Git-Mirrors/textfiles-politics
1
0
Fork 0
mirror of https://github.com/nhammer514/textfiles-politics.git synced 2025-01-09 06:27:59 -05:00
Code Issues Packages Projects Releases Wiki Activity
22a4906740
textfiles-politics/pythonCode/personTagger.py

1656 lines
124 KiB
Python
Raw Normal View History

work on Python with Jermaine
2023-04-08 20:16:14 -04:00
import spacy
from collections import Counter
import re as regex
import os
from saxonche import PySaxonProcessor
#### Loads all of the necessary variables and functions.
restores xml root element to pre-src
2023-04-28 02:15:04 -04:00
#nlp = spacy.cli.download("en_core_web_lg")
work on Python with Jermaine
2023-04-08 20:16:14 -04:00
nlp = spacy.load("en_core_web_lg")
#########################################################################################
# ebb: After reading the NLP output, we know spaCy is making some mistakes.
# So, here let's try adding an EntityRuler to customize spaCy's classification. We need
# to configure this BEFORE we send the tokens off to nlp() for processing.
##########################################################################################
# Create the EntityRuler and set it so the ner comes after, so OUR rules take precedence
# Sources:
# W. J. B. Mattingly: https://ner.pythonhumanities.com/02_01_spaCy_Entity_Ruler.html
# spaCy documentation on NER Entity Ruler: https://spacy.io/usage/rule-based-matching#entityruler
config = {"spans_key": None, "annotate_ents": True, "overwrite": True, "validate": True}
ruler = nlp.add_pipe("span_ruler", before="ner", config=config)
# 2023-04-07: ebb: NOTE: before="ner" setting seems to allow the spaCy NER rules to prevail over these patterns where
# there is a conflict.
repaired regex in checkTags
2023-04-09 10:47:55 -04:00
# after="ner" means that the spaCy NER is TOTALLY OVERWRITTEN and invalidated by our patterns.
work on Python with Jermaine
2023-04-08 20:16:14 -04:00
# Notes: Mattingly has this: ruler = nlp.add_pipe("entity_ruler", after="ner", config={"validate": True})
# But this only works when spaCy doesn't recognize a word / phrase as a named entity of any kind.
# If it recognizes a named entity but tags it wrong, we correct it with the span_ruler, not the entity_ruler
patterns = [
repaired regex in checkTags
2023-04-09 10:47:55 -04:00
{"label": "NULL", "pattern": [{"TEXT": {"REGEX": "^-\w+?"}}]},
{"label": "NULL", "pattern": [{"TEXT": {"REGEX": "^.$"}}]},
lots of stuff, python names, xquery to html, xml regex clean-ish
2023-04-24 14:37:38 -04:00
{"label": "NULL", "pattern": [{"TEXT": {"REGEX": "\^+"}}]},
repaired regex in checkTags
2023-04-09 10:47:55 -04:00
{"label": "NULL", "pattern": [{"TEXT": {"REGEX": "^\w\w$"}}]},
{"label": "NULL", "pattern": [{"TEXT": {"REGEX": "^[a-z]+\s+[a-z]+$"}}]},
{"label": "NULL", "pattern": [{"TEXT": {"REGEX": "^.*?__{2,}.*?$"}}]},
{"label": "NORP", "pattern": [{"TEXT": {"REGEX": "CHRISTIAN(ITY|DOM)"}}]},
{"label": "NORP", "pattern": [{"TEXT": {"REGEX": "CHRISTIAN\s+NETWORK"}}]},
# ebb: Don't match on any single characters!
{"label": "NULL", "pattern": [{"TEXT": {"REGEX": "[A-Z]{2,}[A-Z][a-z]+"}}]},
{"label": "NULL", "pattern": [{"TEXT": {"REGEX": "[a-z]{2,}[A-Z][a-z]+"}}]},
{"label": "NULL", "pattern": [{"TEXT": {"REGEX": "^.*?[a-z][A-Z].*?$"}}]},
# ebb: Above line attempts to stop matching things like Oak IslandThe Method
{"label": "NULL", "pattern": [{"TEXT" : {"REGEX": "^[Mm\-]+$"}}]},
lots of stuff, python names, xquery to html, xml regex clean-ish
2023-04-24 14:37:38 -04:00
{"label": "NULL", "pattern": [{"TEXT": {"REGEX": "\w+cia\w+"}}]},
{"label": "NULL", "pattern": [{"TEXT": {"REGEX": "fed\w+"}}]},
repaired regex in checkTags
2023-04-09 10:47:55 -04:00
# SOCIALISMBY RICHARD
# ebb: Above line attempts to stop matching things Mmm-mm or mm , etc.
trying my best to fix the regex and editing python code
2023-04-27 00:54:09 -04:00
{"label": "NULL", "pattern": [{"TEXT": {"REGEX": "TITIOUS CONCEPTS"}}]},
{"label": "NULL", "pattern": [{"TEXT": {"REGEX": "TORTIOUS"}}]},
{"label": "NULL", "pattern": [{"TEXT": {"REGEX": "THROUGHWAYS"}}]},
{"label": "NULL", "pattern": [{"TEXT": {"REGEX": "TARGET"}}]},
{"label": "NULL", "pattern": [{"TEXT": {"REGEX": "NULL"}}]},
{"label": "NULL", "pattern": [{"TEXT": {"REGEX": "Sympathisers"}}]},
{"label": "NULL", "pattern": [{"TEXT": {"REGEX": "Sympathy"}}]},
{"label": "NULL", "pattern": [{"TEXT": {"REGEX": "UNTIL"}}]},
{"label": "NULL", "pattern": [{"TEXT": {"REGEX": "UNCLE"}}]},
{"label": "NULL", "pattern": [{"TEXT": {"REGEX": "VCI"}}]},
{"label": "NULL", "pattern": [{"TEXT": {"REGEX": "VISUALIZATION"}}]},
{"label": "NULL", "pattern": [{"TEXT": {"REGEX": "YOGA"}}]},
{"label": "NULL", "pattern": [{"TEXT": {"REGEX": "ack Canfield"}}]},
{"label": "NULL", "pattern": [{"TEXT": {"REGEX": "aurem concepisti"}}]},
{"label": "NULL", "pattern": [{"TEXT": {"REGEX": "Xref"}}]},
{"label": "NULL", "pattern": [{"TEXT": {"REGEX": "Wspomnienia"}}]},
{"label": "NULL", "pattern": [{"TEXT": {"REGEX": "Untrustworthy"}}]},
{"label": "NULL", "pattern": [{"TEXT": {"REGEX": "Twelve Apostles"}}]},
{"label": "NULL", "pattern": [{"TEXT": {"REGEX": "Tsar"}}]},
{"label": "NULL", "pattern": [{"TEXT": {"REGEX": "Treasurer"}}]},
{"label": "NULL", "pattern": [{"TEXT": {"REGEX": "Thunderer"}}]},
{"label": "NULL", "pattern": [{"TEXT": {"REGEX": "Thou raisest"}}]},
{"label": "NULL", "pattern": [{"TEXT": {"REGEX": "Thou cleansest"}}]},
{"label": "NULL", "pattern": [{"TEXT": {"REGEX": "Thou castest"}}]},
{"label": "NULL", "pattern": [{"TEXT": {"REGEX": "This Leopold the Traitor"}}]},
{"label": "NULL", "pattern": [{"TEXT": {"REGEX": "Swore"}}]},
{"label": "NULL", "pattern": [{"TEXT": {"REGEX": "Swastika"}}]},
{"label": "NULL", "pattern": [{"TEXT": {"REGEX": "Superstrings"}}]},
{"label": "NULL", "pattern": [{"TEXT": {"REGEX": "Subj"}}]},
{"label": "NULL", "pattern": [{"TEXT": {"REGEX": "Subterrene"}}]},
{"label": "NULL", "pattern": [{"TEXT": {"REGEX": "Sts"}}]},
{"label": "NULL", "pattern": [{"TEXT": {"REGEX": "Strength"}}]},
{"label": "NULL", "pattern": [{"TEXT": {"REGEX": "Sol Deus Invictus"}}]},
{"label": "NULL", "pattern": [{"TEXT": {"REGEX": "Skis"}}]},
{"label": "NULL", "pattern": [{"TEXT": {"REGEX": "Sect"}}]},
{"label": "NULL", "pattern": [{"TEXT": {"REGEX": "Sen"}}]},
{"label": "NULL", "pattern": [{"TEXT": {"REGEX": "Senza Aristotele noi"}}]},
{"label": "NULL", "pattern": [{"TEXT": {"REGEX": "Salubrious Living"}}]},
{"label": "NULL", "pattern": [{"TEXT": {"REGEX": "Saint vouches"}}]},
{"label": "NULL", "pattern": [{"TEXT": {"REGEX": "Safehaven"}}]},
{"label": "NULL", "pattern": [{"TEXT": {"REGEX": "SURPRISED"}}]},
{"label": "NULL", "pattern": [{"TEXT": {"REGEX": "SURE"}}]},
{"label": "NULL", "pattern": [{"TEXT": {"REGEX": "SUCCESSIVE"}}]},
{"label": "NULL", "pattern": [{"TEXT": {"REGEX": "SUBSEQUENT"}}]},
{"label": "NULL", "pattern": [{"TEXT": {"REGEX": "STUNTS"}}]},
{"label": "NULL", "pattern": [{"TEXT": {"REGEX": "STREAMLINED"}}]},
{"label": "NULL", "pattern": [{"TEXT": {"REGEX": "SPECULATION"}}]},
{"label": "NULL", "pattern": [{"TEXT": {"REGEX": "SPECULA"}}]},
{"label": "NULL", "pattern": [{"TEXT": {"REGEX": "SOVEREIGN"}}]},
{"label": "NULL", "pattern": [{"TEXT": {"REGEX": "SHOCKED"}}]},
{"label": "NULL", "pattern": [{"TEXT": {"REGEX": "SEPT"}}]},
{"label": "NULL", "pattern": [{"TEXT": {"REGEX": "SAVIOR"}}]},
{"label": "NULL", "pattern": [{"TEXT": {"REGEX": "Roman Emperor"}}]},
{"label": "NULL", "pattern": [{"TEXT": {"REGEX": "Revelation"}}]},
{"label": "NULL", "pattern": [{"TEXT": {"REGEX": "Resurrection Bone"}}]},
{"label": "NULL", "pattern": [{"TEXT": {"REGEX": "Relearning"}}]},
{"label": "NULL", "pattern": [{"TEXT": {"REGEX": "Reason"}}]},
{"label": "NULL", "pattern": [{"TEXT": {"REGEX": "Read again Knocking the Key Weapon"}}]},
{"label": "NULL", "pattern": [{"TEXT": {"REGEX": "Rea"}}]},
{"label": "NULL", "pattern": [{"TEXT": {"REGEX": "RESOLVE"}}]},
{"label": "NULL", "pattern": [{"TEXT": {"REGEX": "RESALE"}}]},
{"label": "NULL", "pattern": [{"TEXT": {"REGEX": "REBELS"}}]},
{"label": "NULL", "pattern": [{"TEXT": {"REGEX": "RATIFICATION"}}]},
{"label": "NULL", "pattern": [{"TEXT": {"REGEX": "Qui"}}]},
{"label": "NULL", "pattern": [{"TEXT": {"REGEX": "Priest"}}]},
{"label": "NULL", "pattern": [{"TEXT": {"REGEX": "Price"}}]},
{"label": "NULL", "pattern": [{"TEXT": {"REGEX": "Popes"}}]},
{"label": "NULL", "pattern": [{"TEXT": {"REGEX": "Place"}}]},
{"label": "NULL", "pattern": [{"TEXT": {"REGEX": "Philosophizing"}}]},
{"label": "NULL", "pattern": [{"TEXT": {"REGEX": "Pharmacopaeia"}}]},
{"label": "NULL", "pattern": [{"TEXT": {"REGEX": "Pharaoh"}}]},
{"label": "NULL", "pattern": [{"TEXT": {"REGEX": "Pauline Privilege"}}]},
{"label": "NULL", "pattern": [{"TEXT": {"REGEX": "Paul the Persecutor"}}]},
{"label": "NULL", "pattern": [{"TEXT": {"REGEX": "Page"}}]},
{"label": "NULL", "pattern": [{"TEXT": {"REGEX": "Pagan Fathers"}}]},
{"label": "NULL", "pattern": [{"TEXT": {"REGEX": "PWs"}}]},
{"label": "NULL", "pattern": [{"TEXT": {"REGEX": "PROPAGANDA"}}]},
{"label": "NULL", "pattern": [{"TEXT": {"REGEX": "PRIVILEGES"}}]},
{"label": "NULL", "pattern": [{"TEXT": {"REGEX": "PRIOR RESTRAINT"}}]},
{"label": "NULL", "pattern": [{"TEXT": {"REGEX": "PRINCE satropic"}}]},
{"label": "NULL", "pattern": [{"TEXT": {"REGEX": "PRESERVATION"}}]},
{"label": "NULL", "pattern": [{"TEXT": {"REGEX": "POPE"}}]},
{"label": "NULL", "pattern": [{"TEXT": {"REGEX": "Obersturmbannfuehrer"}}]},
{"label": "NULL", "pattern": [{"TEXT": {"REGEX": "Obersturmbannfuehrer Brandt"}}]},
{"label": "NULL", "pattern": [{"TEXT": {"REGEX": "Obersturmbannfuehrer Eichmann"}}]},
{"label": "NULL", "pattern": [{"TEXT": {"REGEX": "Obersturmbannfuehrer Liebehenschel"}}]},
{"label": "NULL", "pattern": [{"TEXT": {"REGEX": "Obersturmbannfuehrer Rauff"}}]},
{"label": "NULL", "pattern": [{"TEXT": {"REGEX": "Obersturmbannfuehrer Rodl"}}]},
{"label": "NULL", "pattern": [{"TEXT": {"REGEX": "Obersturmbannfuehrer Sievers"}}]},
{"label": "NULL", "pattern": [{"TEXT": {"REGEX": "Obersturmbannfuehrer Walter Rauff"}}]},
{"label": "NULL", "pattern": [{"TEXT": {"REGEX": "OPIUM CULTIVATION"}}]},
{"label": "NULL", "pattern": [{"TEXT": {"REGEX": "Nutshell"}}]},
{"label": "NULL", "pattern": [{"TEXT": {"REGEX": "Numberless"}}]},
{"label": "NULL", "pattern": [{"TEXT": {"REGEX": "Num"}}]},
{"label": "NULL", "pattern": [{"TEXT": {"REGEX": "Northwest"}}]},
{"label": "NULL", "pattern": [{"TEXT": {"REGEX": "Northshore"}}]},
{"label": "NULL", "pattern": [{"TEXT": {"REGEX": "Northeast"}}]},
{"label": "NULL", "pattern": [{"TEXT": {"REGEX": "Ngo Dinh Diem"}}]},
{"label": "NULL", "pattern": [{"TEXT": {"REGEX": "Ngo Din Diem"}}]},
{"label": "NULL", "pattern": [{"TEXT": {"REGEX": "Nay"}}]},
{"label": "NULL", "pattern": [{"TEXT": {"REGEX": "NAZI CARTEL"}}]},
{"label": "NULL", "pattern": [{"TEXT": {"REGEX": "NATURE"}}]},
{"label": "NULL", "pattern": [{"TEXT": {"REGEX": "Msgr"}}]},
{"label": "NULL", "pattern": [{"TEXT": {"REGEX": "Mother Goose"}}]},
{"label": "NULL", "pattern": [{"TEXT": {"REGEX": "Ministerially"}}]},
{"label": "NULL", "pattern": [{"TEXT": {"REGEX": "Martial Law Threat"}}]},
{"label": "NULL", "pattern": [{"TEXT": {"REGEX": "Marquis Who"}}]},
{"label": "NULL", "pattern": [{"TEXT": {"REGEX": "Mapmakers"}}]},
{"label": "NULL", "pattern": [{"TEXT": {"REGEX": "Mao"}}]},
{"label": "NULL", "pattern": [{"TEXT": {"REGEX": "MIllbourn"}}]},
{"label": "NULL", "pattern": [{"TEXT": {"REGEX": "MEDLEY"}}]},
{"label": "NULL", "pattern": [{"TEXT": {"REGEX": "MAR"}}]},
{"label": "NULL", "pattern": [{"TEXT": {"REGEX": "Lop"}}]},
{"label": "NULL", "pattern": [{"TEXT": {"REGEX": "LYINGLY FOUNDED"}}]},
{"label": "NULL", "pattern": [{"TEXT": {"REGEX": "Law"}}]},
{"label": "NULL", "pattern": [{"TEXT": {"REGEX": "LAW"}}]},
{"label": "NULL", "pattern": [{"TEXT": {"REGEX": "Keys"}}]},
{"label": "NULL", "pattern": [{"TEXT": {"REGEX": "KOD"}}]},
{"label": "NULL", "pattern": [{"TEXT": {"REGEX": "Juifs"}}]},
{"label": "NULL", "pattern": [{"TEXT": {"REGEX": "Judo"}}]},
{"label": "NULL", "pattern": [{"TEXT": {"REGEX": "Joseph the Carpenter"}}]},
{"label": "NULL", "pattern": [{"TEXT": {"REGEX": "Jerome arraigns"}}]},
{"label": "NULL", "pattern": [{"TEXT": {"REGEX": "Jack Built"}}]},
{"label": "NULL", "pattern": [{"TEXT": {"REGEX": "JUDAICA"}}]},
{"label": "NULL", "pattern": [{"TEXT": {"REGEX": "JURIS"}}]},
{"label": "NULL", "pattern": [{"TEXT": {"REGEX": "JOINT"}}]},
{"label": "NULL", "pattern": [{"TEXT": {"REGEX": "JHVH"}}]},
{"label": "NULL", "pattern": [{"TEXT": {"REGEX": "JESUS THE"}}]},
{"label": "NULL", "pattern": [{"TEXT": {"REGEX": "JESUS DIED"}}]},
{"label": "NULL", "pattern": [{"TEXT": {"REGEX": "Izvestiya VUZ"}}]},
{"label": "NULL", "pattern": [{"TEXT": {"REGEX": "Iron Shins"}}]},
{"label": "NULL", "pattern": [{"TEXT": {"REGEX": "Iren"}}]},
{"label": "NULL", "pattern": [{"TEXT": {"REGEX": "Infotechnology"}}]},
{"label": "NULL", "pattern": [{"TEXT": {"REGEX": "Inerrancy"}}]},
{"label": "NULL", "pattern": [{"TEXT": {"REGEX": "Imprimatur"}}]},
{"label": "NULL", "pattern": [{"TEXT": {"REGEX": "Imposture"}}]},
{"label": "NULL", "pattern": [{"TEXT": {"REGEX": "Implore"}}]},
{"label": "NULL", "pattern": [{"TEXT": {"REGEX": "Immortale Dei"}}]},
{"label": "NULL", "pattern": [{"TEXT": {"REGEX": "Illuminatus Lanz"}}]},
{"label": "NULL", "pattern": [{"TEXT": {"REGEX": "Illiterates Illiterates"}}]},
{"label": "NULL", "pattern": [{"TEXT": {"REGEX": "INTELLECTUALS"}}]},
{"label": "NULL", "pattern": [{"TEXT": {"REGEX": "IMMOVABLE"}}]},
{"label": "NULL", "pattern": [{"TEXT": {"REGEX": "IMMIGRANT AZIZ Rehman"}}]},
{"label": "NULL", "pattern": [{"TEXT": {"REGEX": "ILLUNINATI USE MAFIAS"}}]},
{"label": "NULL", "pattern": [{"TEXT": {"REGEX": "Hun"}}]},
{"label": "NULL", "pattern": [{"TEXT": {"REGEX": "Hoo boy"}}]},
{"label": "NULL", "pattern": [{"TEXT": {"REGEX": "Holy Virginity"}}]},
{"label": "NULL", "pattern": [{"TEXT": {"REGEX": "Holy Vehm"}}]},
{"label": "NULL", "pattern": [{"TEXT": {"REGEX": "Hoc Signo"}}]},
{"label": "NULL", "pattern": [{"TEXT": {"REGEX": "His Covenant"}}]},
{"label": "NULL", "pattern": [{"TEXT": {"REGEX": "Henry the Navigator"}}]},
{"label": "NULL", "pattern": [{"TEXT": {"REGEX": "Heathens don"}}]},
{"label": "NULL", "pattern": [{"TEXT": {"REGEX": "auptsturmfuehrer Truehe"}}]},
{"label": "NULL", "pattern": [{"TEXT": {"REGEX": "Hauptscharfuehrer Ruebe"}}]},
{"label": "NULL", "pattern": [{"TEXT": {"REGEX": "Hamstring Yugo"}}]},
{"label": "NULL", "pattern": [{"TEXT": {"REGEX": "HYPERSPACE"}}]},
{"label": "NULL", "pattern": [{"TEXT": {"REGEX": "HUMANE"}}]},
{"label": "NULL", "pattern": [{"TEXT": {"REGEX": "HEED"}}]},
{"label": "NULL", "pattern": [{"TEXT": {"REGEX": "HEARINGS"}}]},
{"label": "NULL", "pattern": [{"TEXT": {"REGEX": "Guillotine"}}]},
{"label": "NULL", "pattern": [{"TEXT": {"REGEX": "Grupenfuehrer Globocnik"}}]},
{"label": "NULL", "pattern": [{"TEXT": {"REGEX": "Gruppenfuehrer Katzmann"}}]},
{"label": "NULL", "pattern": [{"TEXT": {"REGEX": "Groom"}}]},
{"label": "NULL", "pattern": [{"TEXT": {"REGEX": "Gregory the Great"}}]},
{"label": "NULL", "pattern": [{"TEXT": {"REGEX": "Grammarians"}}]},
{"label": "NULL", "pattern": [{"TEXT": {"REGEX": "Grammar"}}]},
{"label": "NULL", "pattern": [{"TEXT": {"REGEX": "Gospel"}}]},
{"label": "NULL", "pattern": [{"TEXT": {"REGEX": "Gort"}}]},
{"label": "NULL", "pattern": [{"TEXT": {"REGEX": "Goebbelesque"}}]},
{"label": "NULL", "pattern": [{"TEXT": {"REGEX": "Godhead"}}]},
{"label": "NULL", "pattern": [{"TEXT": {"REGEX": "Girkalinei Comm"}}]},
{"label": "NULL", "pattern": [{"TEXT": {"REGEX": "Geschichte der Staatstheorien"}}]},
{"label": "NULL", "pattern": [{"TEXT": {"REGEX": "Genus Shamanensis"}}]},
{"label": "NULL", "pattern": [{"TEXT": {"REGEX": "Genius"}}]},
{"label": "NULL", "pattern": [{"TEXT": {"REGEX": "Gene Pool"}}]},
{"label": "NULL", "pattern": [{"TEXT": {"REGEX": "Gemeter"}}]},
{"label": "NULL", "pattern": [{"TEXT": {"REGEX": "Gaude Virgo"}}]},
{"label": "NULL", "pattern": [{"TEXT": {"REGEX": "Gamel Abdul Nasser"}}]},
{"label": "NULL", "pattern": [{"TEXT": {"REGEX": "Gameplayers"}}]},
{"label": "NULL", "pattern": [{"TEXT": {"REGEX": "Galicie Ukrainienne"}}]},
{"label": "NULL", "pattern": [{"TEXT": {"REGEX": "Gal"}}]},
{"label": "NULL", "pattern": [{"TEXT": {"REGEX": "GUNJAMMER"}}]},
{"label": "NULL", "pattern": [{"TEXT": {"REGEX": "GUILTY"}}]},
{"label": "NULL", "pattern": [{"TEXT": {"REGEX": "GREED"}}]},
{"label": "NULL", "pattern": [{"TEXT": {"REGEX": "GRAMS"}}]},
{"label": "NULL", "pattern": [{"TEXT": {"REGEX": "GOVERNORS"}}]},
{"label": "NULL", "pattern": [{"TEXT": {"REGEX": "GENIUS"}}]},
{"label": "NULL", "pattern": [{"TEXT": {"REGEX": "GAER"}}]},
{"label": "NULL", "pattern": [{"TEXT": {"REGEX": "Furness"}}]},
{"label": "NULL", "pattern": [{"TEXT": {"REGEX": "Free Inq"}}]},
{"label": "NULL", "pattern": [{"TEXT": {"REGEX": "Foxx"}}]},
{"label": "NULL", "pattern": [{"TEXT": {"REGEX": "Foo Fighters"}}]},
{"label": "NULL", "pattern": [{"TEXT": {"REGEX": "Fig"}}]},
{"label": "NULL", "pattern": [{"TEXT": {"REGEX": "Figs"}}]},
{"label": "NULL", "pattern": [{"TEXT": {"REGEX": "Feverfew"}}]},
{"label": "NULL", "pattern": [{"TEXT": {"REGEX": "Feathers"}}]},
{"label": "NULL", "pattern": [{"TEXT": {"REGEX": "Fathers"}}]},
{"label": "NULL", "pattern": [{"TEXT": {"REGEX": "FUNC"}}]},
{"label": "NULL", "pattern": [{"TEXT": {"REGEX": "FOUO"}}]},
{"label": "NULL", "pattern": [{"TEXT": {"REGEX": "FILES KIT"}}]},
{"label": "NULL", "pattern": [{"TEXT": {"REGEX": "FEVERFEW"}}]},
{"label": "NULL", "pattern": [{"TEXT": {"REGEX": "FHK"}}]},
{"label": "NULL", "pattern": [{"TEXT": {"REGEX": "Exodus"}}]},
{"label": "NULL", "pattern": [{"TEXT": {"REGEX": "Eur"}}]},
{"label": "NULL", "pattern": [{"TEXT": {"REGEX": "Epistle clxvi"}}]},
{"label": "NULL", "pattern": [{"TEXT": {"REGEX": "Epistle"}}]},
{"label": "NULL", "pattern": [{"TEXT": {"REGEX": "Enfranchised"}}]},
{"label": "NULL", "pattern": [{"TEXT": {"REGEX": "Endothermic"}}]},
{"label": "NULL", "pattern": [{"TEXT": {"REGEX": "Encyc"}}]},
{"label": "NULL", "pattern": [{"TEXT": {"REGEX": "Encloged"}}]},
{"label": "NULL", "pattern": [{"TEXT": {"REGEX": "Emperor"}}]},
{"label": "NULL", "pattern": [{"TEXT": {"REGEX": "Emperor Vespasian"}}]},
{"label": "NULL", "pattern": [{"TEXT": {"REGEX": "Electromagnetically"}}]},
{"label": "NULL", "pattern": [{"TEXT": {"REGEX": "Ecclesiastici Publici"}}]},
{"label": "NULL", "pattern": [{"TEXT": {"REGEX": "Eccles"}}]},
{"label": "NULL", "pattern": [{"TEXT": {"REGEX": "Ebrietatis Enconium"}}]},
{"label": "NULL", "pattern": [{"TEXT": {"REGEX": "EXTRAORDINAIRE"}}]},
{"label": "NULL", "pattern": [{"TEXT": {"REGEX": "EXPATRIATION"}}]},
{"label": "NULL", "pattern": [{"TEXT": {"REGEX": "EPISTLES"}}]},
{"label": "NULL", "pattern": [{"TEXT": {"REGEX": "ENORMOUS"}}]},
{"label": "NULL", "pattern": [{"TEXT": {"REGEX": "EMPLOY"}}]},
{"label": "NULL", "pattern": [{"TEXT": {"REGEX": "Dupes"}}]},
{"label": "NULL", "pattern": [{"TEXT": {"REGEX": "Dueling Spymasters"}}]},
{"label": "NULL", "pattern": [{"TEXT": {"REGEX": "Dual Amiga"}}]},
{"label": "NULL", "pattern": [{"TEXT": {"REGEX": "Driver"}}]},
{"label": "NULL", "pattern": [{"TEXT": {"REGEX": "Dritte Reich"}}]},
{"label": "NULL", "pattern": [{"TEXT": {"REGEX": "Drang nach Osten"}}]},
{"label": "NULL", "pattern": [{"TEXT": {"REGEX": "Dorman"}}]},
{"label": "NULL", "pattern": [{"TEXT": {"REGEX": "Don sez"}}]},
{"label": "NULL", "pattern": [{"TEXT": {"REGEX": "Divinatione Daemonorum"}}]},
{"label": "NULL", "pattern": [{"TEXT": {"REGEX": "Distinguished paedagogists"}}]},
{"label": "NULL", "pattern": [{"TEXT": {"REGEX": "Dies irae"}}]},
{"label": "NULL", "pattern": [{"TEXT": {"REGEX": "Die Geschichte von Joel Brand"}}]},
{"label": "NULL", "pattern": [{"TEXT": {"REGEX": "Dew"}}]},
{"label": "NULL", "pattern": [{"TEXT": {"REGEX": "Deut"}}]},
{"label": "NULL", "pattern": [{"TEXT": {"REGEX": "Desert Rat"}}]},
{"label": "NULL", "pattern": [{"TEXT": {"REGEX": "Deregulatory Creep"}}]},
{"label": "NULL", "pattern": [{"TEXT": {"REGEX": "Der Weg des Politischen Katholizismus"}}]},
{"label": "NULL", "pattern": [{"TEXT": {"REGEX": "Depository"}}]},
{"label": "NULL", "pattern": [{"TEXT": {"REGEX": "Debasement"}}]},
{"label": "NULL", "pattern": [{"TEXT": {"REGEX": "Death Ray"}}]},
{"label": "NULL", "pattern": [{"TEXT": {"REGEX": "Das Sogenannte Boese"}}]},
{"label": "NULL", "pattern": [{"TEXT": {"REGEX": "CORRUPT"}}]},
{"label": "NULL", "pattern": [{"TEXT": {"REGEX": "COMMISSIONED THE UNDERSIGNED"}}]},
{"label": "NULL", "pattern": [{"TEXT": {"REGEX": "CLEARFIELD"}}]},
{"label": "WORK_OF_ART", "pattern": [{"TEXT": {"REGEX": "Pagan Sibylline Oracles"}}]},
{"label": "NULL", "pattern": [{"TEXT": {"REGEX": "CIRCUMSTANTIAL"}}]},
{"label": "NULL", "pattern": [{"TEXT": {"REGEX": "CDT"}}]},
{"label": "NULL", "pattern": [{"TEXT": {"REGEX": "CANNOT"}}]},
{"label": "ORG", "pattern": [{"TEXT": {"REGEX": "BATF"}}]},
{"label": "NULL", "pattern": [{"TEXT": {"REGEX": "BEEN"}}]},
{"label": "NULL", "pattern": [{"TEXT": {"REGEX": "BLACKMAIL"}}]},
{"label": "NULL", "pattern": [{"TEXT": {"REGEX": "BOURGEOIS GENTILHOMME"}}]},
{"label": "GPE", "pattern": [{"TEXT": {"REGEX": "Bagdad"}}]},
{"label": "LAW", "pattern": [{"TEXT": {"REGEX": "Balfour Declaration"}}]},
{"label": "NULL", "pattern": [{"TEXT": {"REGEX": "Battle"}}]},
{"label": "NULL", "pattern": [{"TEXT": {"REGEX": "Biblical"}}]},
{"label": "NULL", "pattern": [{"TEXT": {"REGEX": "Bishop"}}]},
{"label": "NULL", "pattern": [{"TEXT": {"REGEX": "Blade Runner"}}]},
{"label": "NULL", "pattern": [{"TEXT": {"REGEX": "Blackmail"}}]},
{"label": "NULL", "pattern": [{"TEXT": {"REGEX": "Blessing"}}]},
{"label": "NULL", "pattern": [{"TEXT": {"REGEX": "Briefly"}}]},
{"label": "NULL", "pattern": [{"TEXT": {"REGEX": "Brigadefuehrer"}}]},
{"label": "NULL", "pattern": [{"TEXT": {"REGEX": "Brigadefuehrer Globocnik"}}]},
{"label": "NULL", "pattern": [{"TEXT": {"REGEX": "Brigadefuehrer Gluecks"}}]},
{"label": "NULL", "pattern": [{"TEXT": {"REGEX": "Brigadefuehrer Zenner"}}]},
{"label": "NULL", "pattern": [{"TEXT": {"REGEX": "Brigadefuhrer Richard Glucks"}}]},
{"label": "NULL", "pattern": [{"TEXT": {"REGEX": "Brigadefuehrer"}}]},
{"label": "NULL", "pattern": [{"TEXT": {"REGEX": "Brigadefuhrer"}}]},
{"label": "NULL", "pattern": [{"TEXT": {"REGEX": "CREATORS"}}]},
{"label": "NULL", "pattern": [{"TEXT": {"REGEX": "Cap"}}]},
{"label": "NULL", "pattern": [{"TEXT": {"REGEX": "Capitula Angilrammi"}}]},
{"label": "NULL", "pattern": [{"TEXT": {"REGEX": "Case"}}]},
{"label": "NULL", "pattern": [{"TEXT": {"REGEX": "Charles Mentesana Filmed"}}]},
{"label": "NULL", "pattern": [{"TEXT": {"REGEX": "Chillingworth"}}]},
{"label": "NULL", "pattern": [{"TEXT": {"REGEX": "Christian Pope"}}]},
{"label": "NULL", "pattern": [{"TEXT": {"REGEX": "Chronicles xxviii"}}]},
{"label": "NULL", "pattern": [{"TEXT": {"REGEX": "Coddled Omar Abdel Rahman"}}]},
{"label": "NULL", "pattern": [{"TEXT": {"REGEX": "Chronicles xxxiv"}}]},
{"label": "NULL", "pattern": [{"TEXT": {"REGEX": "Clem"}}]},
{"label": "NULL", "pattern": [{"TEXT": {"REGEX": "Commrce"}}]},
{"label": "NULL", "pattern": [{"TEXT": {"REGEX": "Corinthians"}}]},
{"label": "NULL", "pattern": [{"TEXT": {"REGEX": "Cosi"}}]},
{"label": "NULL", "pattern": [{"TEXT": {"REGEX": "Crucifixion"}}]},
{"label": "NULL", "pattern": [{"TEXT": {"REGEX": "Czar"}}]},
{"label": "NULL", "pattern": [{"TEXT": {"REGEX": "DECRETALS"}}]},
{"label": "NULL", "pattern": [{"TEXT": {"REGEX": "DECAY"}}]},
{"label": "NULL", "pattern": [{"TEXT": {"REGEX": "DEPLETION"}}]},
{"label": "NULL", "pattern": [{"TEXT": {"REGEX": "DISASTROUS THAN PEARL"}}]},
{"label": "NULL", "pattern": [{"TEXT": {"REGEX": "DISEASES"}}]},
{"label": "NULL", "pattern": [{"TEXT": {"REGEX": "DOPE"}}]},
{"label": "NULL", "pattern": [{"TEXT": {"REGEX": "Damit Ich"}}]},
{"label": "WORK_OF_ART", "pattern": [{"TEXT": {"REGEX": "Pagan Oracles"}}]},
{"label": "NULL", "pattern": [{"TEXT": {"REGEX": "Deceits"}}]},
{"label": "NULL", "pattern": [{"TEXT": {"REGEX": "Deep Cover"}}]},
{"label": "NULL", "pattern": [{"TEXT": {"REGEX": "Demagnetize"}}]},
{"label": "WORK_OF_ART", "pattern": [{"TEXT": {"REGEX": "Mona Lisa"}}]},
{"label": "NULL", "pattern": [{"TEXT": {"REGEX": "Demigod"}}]},
{"label": "NULL", "pattern": [{"TEXT": {"REGEX": "Demjanjuk"}}]},
{"label": "NORP", "pattern": [{"TEXT": {"REGEX": "Democratic Centralism"}}]},
{"label": "WORK_OF_ART", "pattern": [{"TEXT": {"REGEX": "Mein Kampf"}}]},
{"label": "WORK_OF_ART", "pattern": [{"TEXT": {"REGEX": "Mary Magdalin Gospel"}}]},
{"label": "WORK_OF_ART", "pattern": [{"TEXT": {"REGEX": "Horst Wessel Song"}}]},
{"label": "WORK_OF_ART", "pattern": [{"TEXT": {"REGEX": "Fama Fraternitatis"}}]},
{"label": "NORP", "pattern": [{"TEXT": {"REGEX": "Physiocrats"}}]},
{"label": "NORP", "pattern": [{"TEXT": {"REGEX": "Pagan"}}]},
{"label": "NORP", "pattern": [{"TEXT": {"REGEX": "Meo"}}]},
{"label": "NORP", "pattern": [{"TEXT": {"REGEX": "PAGAN"}}]},
{"label": "NORP", "pattern": [{"TEXT": {"REGEX": "Obersturmfuehrer"}}]},
{"label": "NORP", "pattern": [{"TEXT": {"REGEX": "MARTYRS"}}]},
{"label": "NORP", "pattern": [{"TEXT": {"REGEX": "Jurisconsults"}}]},
{"label": "NORP", "pattern": [{"TEXT": {"REGEX": "Jewess"}}]},
{"label": "NORP", "pattern": [{"TEXT": {"REGEX": "Jeremiads"}}]},
{"label": "NORP", "pattern": [{"TEXT": {"REGEX": "Jehovah's Witnesses"}}]},
{"label": "NORP", "pattern": [{"TEXT": {"REGEX": "Jehovah's Witness"}}]},
{"label": "NORP", "pattern": [{"TEXT": {"REGEX": "Jehovahs Witness"}}]},
{"label": "NORP", "pattern": [{"TEXT": {"REGEX": "JURORS"}}]},
{"label": "NORP", "pattern": [{"TEXT": {"REGEX": "Heathens"}}]},
{"label": "NORP", "pattern": [{"TEXT": {"REGEX": "Gypsy"}}]},
{"label": "NORP", "pattern": [{"TEXT": {"REGEX": "Geronimo Chiricahua Apache"}}]},
{"label": "NORP", "pattern": [{"TEXT": {"REGEX": "Gauleiter"}}]},
{"label": "NORP", "pattern": [{"TEXT": {"REGEX": "Frankists"}}]},
{"label": "NORP", "pattern": [{"TEXT": {"REGEX": "Framers"}}]},
{"label": "NORP", "pattern": [{"TEXT": {"REGEX": "Framer"}}]},
{"label": "NORP", "pattern": [{"TEXT": {"REGEX": "Felloweraft"}}]},
{"label": "NORP", "pattern": [{"TEXT": {"REGEX": "Evangelist"}}]},
{"label": "NORP", "pattern": [{"TEXT": {"REGEX": "Egotists"}}]},
{"label": "NORP", "pattern": [{"TEXT": {"REGEX": "Duce"}}]},
repaired regex in checkTags
2023-04-09 10:47:55 -04:00
{"label": "GPE", "pattern": [{"TEXT": {"REGEX": "Babylon(ia)?"}}]},
trying my best to fix the regex and editing python code
2023-04-27 00:54:09 -04:00
{"label": "NORP", "pattern": [{"TEXT": {"REGEX": "Diluvian Patriarchs"}}]},
repaired regex in checkTags
2023-04-09 10:47:55 -04:00
{"label": "NORP", "pattern": [{"TEXT": {"REGEX": "Christiani\s*ty"}}]},
trying my best to fix the regex and editing python code
2023-04-27 00:54:09 -04:00
{"label": "ORG", "pattern": [{"TEXT": {"REGEX": "CAIB"}}]},
{"label": "NULL", "pattern": [{"TEXT": {"REGEX": "the .+?"}}]},
{"label": "NORP", "pattern": [{"TEXT": {"REGEX": "Infidels"}}]},
{"label": "NULL", "pattern": [{"TEXT": {"REGEX": "Christiani"}}]},
{"label": "NULL", "pattern": [{"TEXT": {"REGEX": "MAHAT"}}]},
{"label": "NULL", "pattern": [{"TEXT": {"REGEX": "Mahat"}}]},
{"label": "NULL", "pattern": [{"TEXT": {"REGEX": "Atta Boy"}}]},
{"label": "NULL", "pattern": [{"TEXT": {"REGEX": "Assailants"}}]},
{"label": "NULL", "pattern": [{"TEXT": {"REGEX": "Assassinated Robert"}}]},
{"label": "NULL", "pattern": [{"TEXT": {"REGEX": "Article"}}]},
{"label": "NULL", "pattern": [{"TEXT": {"REGEX": "Adrift"}}]},
{"label": "NULL", "pattern": [{"TEXT": {"REGEX": "MAD"}}]},
{"label": "NULL", "pattern": [{"TEXT": {"REGEX": "bacchanalian"}}]},
{"label": "NULL", "pattern": [{"TEXT": {"REGEX": "behests"}}]},
{"label": "NULL", "pattern": [{"TEXT": {"REGEX": "Parallel"}}]},
{"label": "NULL", "pattern": [{"TEXT": {"REGEX": ""}}]},
{"label": "NULL", "pattern": [{"TEXT": {"REGEX": "the\s+?user"}}]},
{"label": "NULL", "pattern": [{"TEXT": {"REGEX": "NETWORK"}}]},
{"label": "NULL", "pattern": [{"TEXT": {"REGEX": "Attemps"}}]},
{"label": "NULL", "pattern": [{"TEXT": {"REGEX": "Foundation"}}]},
{"label": "NULL", "pattern": [{"TEXT": {"REGEX": "Iena"}}]},
{"label": "NULL", "pattern": [{"TEXT": {"REGEX": "attemps"}}]},
{"label": "NULL", "pattern": [{"TEXT": {"REGEX": "pgh"}}]},
{"label": "NULL", "pattern": [{"TEXT": {"REGEX": "Christiani"}}]},
{"label": "NULL", "pattern": [{"TEXT": {"REGEX": "MAHAT"}}]},
{"label": "NULL", "pattern": [{"TEXT": {"REGEX": "Mahat"}}]},
{"label": "NULL", "pattern": [{"TEXT": {"REGEX": "MAD"}}]},
{"label": "NULL", "pattern": [{"TEXT": {"REGEX": "Parallel"}}]},
{"label": "NULL", "pattern": [{"TEXT": {"REGEX": "the\s+?user"}}]},
{"label": "NULL", "pattern": [{"TEXT": {"REGEX": "NETWORK"}}]},
{"label": "NULL", "pattern": [{"TEXT": {"REGEX": "Attemps"}}]},
{"label": "NULL", "pattern": [{"TEXT": {"REGEX": "Iena"}}]},
{"label": "NULL", "pattern": [{"TEXT": {"REGEX": "attemps"}}]},
{"label": "NULL", "pattern": [{"TEXT": {"REGEX": "Advanced"}}]},
{"label": "NULL", "pattern": [{"TEXT": {"REGEX": "Believability"}}]},
{"label": "NULL", "pattern": [{"TEXT": {"REGEX": "Onesuch"}}]},
{"label": "NULL", "pattern": [{"TEXT": {"REGEX": "the\s+?People"}}]},
{"label": "NULL", "pattern": [{"TEXT": {"REGEX": "REPRINT"}}]},
{"label": "NULL", "pattern": [{"TEXT": {"REGEX": "The\s+?Next\s+?Banking\s+?Crisis"}}]},
{"label": "NULL", "pattern": [{"TEXT": {"REGEX": "Incidently"}}]},
{"label": "NULL", "pattern": [{"TEXT": {"REGEX": "Martini\s+?Glass"}}]},
{"label": "NULL", "pattern": [{"TEXT": {"REGEX": "Customs"}}]},
{"label": "NULL", "pattern": [{"TEXT": {"REGEX": "the\s+?Sheriff"}}]},
{"label": "NULL", "pattern": [{"TEXT": {"REGEX": "Greets"}}]},
{"label": "NULL", "pattern": [{"TEXT": {"REGEX": "bey"}}]},
{"label": "NULL", "pattern": [{"TEXT": {"REGEX": "conven"}}]},
{"label": "NULL", "pattern": [{"TEXT": {"REGEX": "Pro"}}]},
{"label": "NULL", "pattern": [{"TEXT": {"REGEX": "THUS"}}]},
{"label": "NULL", "pattern": [{"TEXT": {"REGEX": "TUNED CIRCUT"}}]},
{"label": "NULL", "pattern": [{"TEXT": {"REGEX": "the\s+?supreme\s+?end"}}]},
{"label": "NULL", "pattern": [{"TEXT": {"REGEX": "Ebe"}}]},
{"label": "NULL", "pattern": [{"TEXT": {"REGEX": "Big\s+?Government"}}]},
{"label": "NULL", "pattern": [{"TEXT": {"REGEX": "State"}}]},
{"label": "NULL", "pattern": [{"TEXT": {"REGEX": "pseudoscientists"}}]},
{"label": "NULL", "pattern": [{"TEXT": {"REGEX": "undam"}}]},
{"label": "NULL", "pattern": [{"TEXT": {"REGEX": "News"}}]},
{"label": "NULL", "pattern": [{"TEXT": {"REGEX": "Times"}}]},
{"label": "NULL", "pattern": [{"TEXT": {"REGEX": "PERIODICALS"}}]},
{"label": "NULL", "pattern": [{"TEXT": {"REGEX": "SPEECHES"}}]},
{"label": "NULL", "pattern": [{"TEXT": {"REGEX": "INCLUSION"}}]},
{"label": "NULL", "pattern": [{"TEXT": {"REGEX": "MAJESTY"}}]},
{"label": "NULL", "pattern": [{"TEXT": {"REGEX": "interpet"}}]},
{"label": "NULL", "pattern": [{"TEXT": {"REGEX": "manifesta"}}]},
{"label": "NULL", "pattern": [{"TEXT": {"REGEX": "aspetic"}}]},
{"label": "NULL", "pattern": [{"TEXT": {"REGEX": "mish"}}]},
{"label": "NULL", "pattern": [{"TEXT": {"REGEX": "petros"}}]},
{"label": "NULL", "pattern": [{"TEXT": {"REGEX": "shevirat"}}]},
{"label": "NULL", "pattern": [{"TEXT": {"REGEX": "sanguinis"}}]},
{"label": "NULL", "pattern": [{"TEXT": {"REGEX": "resop"}}]},
{"label": "NULL", "pattern": [{"TEXT": {"REGEX": "tryed"}}]},
{"label": "NULL", "pattern": [{"TEXT": {"REGEX": "quandry"}}]},
{"label": "NULL", "pattern": [{"TEXT": {"REGEX": "imbeds"}}]},
{"label": "NULL", "pattern": [{"TEXT": {"REGEX": "germain"}}]},
{"label": "NULL", "pattern": [{"TEXT": {"REGEX": "orleans"}}]},
{"label": "NULL", "pattern": [{"TEXT": {"REGEX": "tgg"}}]},
{"label": "NULL", "pattern": [{"TEXT": {"REGEX": "Families"}}]},
{"label": "NULL", "pattern": [{"TEXT": {"REGEX": "preparingits"}}]},
{"label": "NULL", "pattern": [{"TEXT": {"REGEX": "germain"}}]},
{"label": "NULL", "pattern": [{"TEXT": {"REGEX": "wintry"}}]},
{"label": "NULL", "pattern": [{"TEXT": {"REGEX": "trogan"}}]},
{"label": "NULL", "pattern": [{"TEXT": {"REGEX": "Interested"}}]},
{"label": "NULL", "pattern": [{"TEXT": {"REGEX": "Time"}}]},
{"label": "NULL", "pattern": [{"TEXT": {"REGEX": "ACTUALLY"}}]},
{"label": "NULL", "pattern": [{"TEXT": {"REGEX": "ACCIDENTALLY"}}]},
{"label": "NULL", "pattern": [{"TEXT": {"REGEX": "The\s+?Transmission"}}]},
{"label": "NULL", "pattern": [{"TEXT": {"REGEX": "contra"}}]},
{"label": "NULL", "pattern": [{"TEXT": {"REGEX": "Mental\s+?Health"}}]},
{"label": "LAW", "pattern": [{"TEXT": {"REGEX": "Bill\s+?of\s+?Rights"}}]},
{"label": "LAW", "pattern": [{"TEXT": {"REGEX": "MCCARREN\s+?ACT"}}]},
{"label": "LAW", "pattern": [{"TEXT": {"REGEX": "Emergency\s+?Detention\s+?Act"}}]},
{"label": "LAW", "pattern": [{"TEXT": {"REGEX": "Geneva\s+?Convention"}}]},
{"label": "LAW", "pattern": [{"TEXT": {"REGEX": "Official\s+?Secrets\s+?Act"}}]},
{"label": "LAW", "pattern": [{"TEXT": {"REGEX": "Executive\s+?Order"}}]},
{"label": "LAW", "pattern": [{"TEXT": {"REGEX": "American\s+?Press\s+?freedom"}}]},
{"label": "LAW", "pattern": [{"TEXT": {"REGEX": "State\s+?Constitution"}}]},
{"label": "LAW", "pattern": [{"TEXT": {"REGEX": "Constitution"}}]},
{"label": "LAW", "pattern": [{"TEXT": {"REGEX": "Martial\s+?Law"}}]},
{"label": "LAW", "pattern": [{"TEXT": {"REGEX": "Martial\s+?Rule"}}]},
{"label": "LAW", "pattern": [{"TEXT": {"REGEX": "Unalienable Rights"}}]},
{"label": "LAW", "pattern": [{"TEXT": {"REGEX": "Alaska\s+?Mental\s+?Health\s+?Bill"}}]},
{"label": "LAW", "pattern": [{"TEXT": {"REGEX": "Multilateral\s+?Protection\s+?of\s+?War\s+?Victims/Prisoners\s+?of\s+?War"}}]},
{"label": "LAW", "pattern": [{"TEXT": {"REGEX": "Multilateral\s+?Protection\s+?of\s+?War\s+?Victims/Civilian\s+?Persons"}}]},
{"label": "LAW", "pattern": [{"TEXT": {"REGEX": "Public\s+?Health\s+?Service\s+?Draft\s+?Act"}}]},
{"label": "LAW", "pattern": [{"TEXT": {"REGEX": "Uniform\s+?Mental\s+?Health\s+?Act"}}]},
{"label": "NULL", "pattern": [{"TEXT": {"REGEX": "Executive\s+?Order\s+?#[0-9]+"}}]},
{"label": "NULL", "pattern": [{"TEXT": {"REGEX": "Median"}}]},
{"label": "NULL", "pattern": [{"TEXT": {"REGEX": "Next"}}]},
{"label": "NULL", "pattern": [{"TEXT": {"REGEX": "Daily"}}]},
{"label": "NULL", "pattern": [{"TEXT": {"REGEX": "the\s+?General"}}]},
{"label": "NULL", "pattern": [{"TEXT": {"REGEX": "Diplomatic"}}]},
{"label": "NULL", "pattern": [{"TEXT": {"REGEX": "televi"}}]},
{"label": "NULL", "pattern": [{"TEXT": {"REGEX": "patsy"}}]},
{"label": "NULL", "pattern": [{"TEXT": {"REGEX": "mike"}}]},
{"label": "NULL", "pattern": [{"TEXT": {"REGEX": "MODIFY"}}]},
{"label": "NULL", "pattern": [{"TEXT": {"REGEX": "Nations"}}]},
{"label": "NULL", "pattern": [{"TEXT": {"REGEX": "substanti"}}]},
{"label": "NULL", "pattern": [{"TEXT": {"REGEX": "audi"}}]},
{"label": "NULL", "pattern": [{"TEXT": {"REGEX": "Justice"}}]},
{"label": "NULL", "pattern": [{"TEXT": {"REGEX": "Cancer"}}]},
{"label": "NULL", "pattern": [{"TEXT": {"REGEX": "Natural\s+?causes"}}]},
{"label": "NULL", "pattern": [{"TEXT": {"REGEX": "PLATO"}}]},
{"label": "NULL", "pattern": [{"TEXT": {"REGEX": "darksuckers"}}]},
{"label": "NULL", "pattern": [{"TEXT": {"REGEX": "Darksuckers"}}]},
{"label": "NULL", "pattern": [{"TEXT": {"REGEX": "MAJESTY"}}]},
{"label": "NULL", "pattern": [{"TEXT": {"REGEX": "SNOWBIRD"}}]},
{"label": "NULL", "pattern": [{"TEXT": {"REGEX": "GARNET"}}]},
{"label": "NULL", "pattern": [{"TEXT": {"REGEX": "AGL"}}]},
{"label": "NULL", "pattern": [{"TEXT": {"REGEX": "DELTA"}}]},
{"label": "NULL", "pattern": [{"TEXT": {"REGEX": "the King"}}]},
{"label": "NULL", "pattern": [{"TEXT": {"REGEX": "EXCALIBUR"}}]},
{"label": "NULL", "pattern": [{"TEXT": {"REGEX": "the\s+?secret\s+?control\s+?group"}}]},
{"label": "NULL", "pattern": [{"TEXT": {"REGEX": "JASON"}}]},
{"label": "NULL", "pattern": [{"TEXT": {"REGEX": "Heart"}}]},
{"label": "NULL", "pattern": [{"TEXT": {"REGEX": "sentatives"}}]},
{"label": "NULL", "pattern": [{"TEXT": {"REGEX": "Peter\s+?Gregory\s+?Original"}}]},
{"label": "NULL", "pattern": [{"TEXT": {"REGEX": "Billy\s+?Lovelady\s+?Depository"}}]},
{"label": "NULL", "pattern": [{"TEXT": {"REGEX": "The\s+?World"}}]},
{"label": "NULL", "pattern": [{"TEXT": {"REGEX": "John\s+?Holbrook\s+?Psychiatrist"}}]},
{"label": "NULL", "pattern": [{"TEXT": {"REGEX": "Paul\s+?Mandal\s+?Life"}}]},
{"label": "NULL", "pattern": [{"TEXT": {"REGEX": "Jackson\s+?Life\s+?Magazine"}}]},
{"label": "NULL", "pattern": [{"TEXT": {"REGEX": "Drug"}}]},
{"label": "NULL", "pattern": [{"TEXT": {"REGEX": "thou mayest"}}]},
{"label": "NULL", "pattern": [{"TEXT": {"REGEX": "thou eatest"}}]},
{"label": "NULL", "pattern": [{"TEXT": {"REGEX": "Unity\s+?movement"}}]},
{"label": "NULL", "pattern": [{"TEXT": {"REGEX": "Conn"}}]},
{"label": "NULL", "pattern": [{"TEXT": {"REGEX": "intel"}}]},
{"label": "NULL", "pattern": [{"TEXT": {"REGEX": "xxx xxxxx"}}]},
{"label": "NULL", "pattern": [{"TEXT": {"REGEX": "Xxxxxx Xxxxxxxx"}}]},
{"label": "NULL", "pattern": [{"TEXT": {"REGEX": "Xxxxxxx Xxxxxxx"}}]},
{"label": "NULL", "pattern": [{"TEXT": {"REGEX": "Wie hoch"}}]},
{"label": "NULL", "pattern": [{"TEXT": {"REGEX": "WOULDN"}}]},
{"label": "NULL", "pattern": [{"TEXT": {"REGEX": "Waar"}}]},
{"label": "NULL", "pattern": [{"TEXT": {"REGEX": "WCH"}}]},
{"label": "NULL", "pattern": [{"TEXT": {"REGEX": "WCR"}}]},
{"label": "NULL", "pattern": [{"TEXT": {"REGEX": "yeldeth ben"}}]},
{"label": "NULL", "pattern": [{"TEXT": {"REGEX": "zwite\s+?Aufseher"}}]},
{"label": "NULL", "pattern": [{"TEXT": {"REGEX": "wuz"}}]},
{"label": "NULL", "pattern": [{"TEXT": {"REGEX": "beeb"}}]},
{"label": "NULL", "pattern": [{"TEXT": {"REGEX": "duetschen Juden"}}]},
{"label": "NULL", "pattern": [{"TEXT": {"REGEX": "bison"}}]},
{"label": "NULL", "pattern": [{"TEXT": {"REGEX": "enim pejor"}}]},
{"label": "NULL", "pattern": [{"TEXT": {"REGEX": "enshroud"}}]},
{"label": "NULL", "pattern": [{"TEXT": {"REGEX": "entorce"}}]},
{"label": "NULL", "pattern": [{"TEXT": {"REGEX": "epidemica"}}]},
{"label": "NULL", "pattern": [{"TEXT": {"REGEX": "blasphemously"}}]},
{"label": "NULL", "pattern": [{"TEXT": {"REGEX": "bull dudu"}}]},
{"label": "NULL", "pattern": [{"TEXT": {"REGEX": "bowie asst"}}]},
{"label": "NULL", "pattern": [{"TEXT": {"REGEX": "canonizes"}}]},
{"label": "NULL", "pattern": [{"TEXT": {"REGEX": "clarinews"}}]},
{"label": "NULL", "pattern": [{"TEXT": {"REGEX": "columno"}}]},
{"label": "NULL", "pattern": [{"TEXT": {"REGEX": "complicitly"}}]},
{"label": "NULL", "pattern": [{"TEXT": {"REGEX": "conspiracy buff"}}]},
{"label": "NULL", "pattern": [{"TEXT": {"REGEX": "creo"}}]},
{"label": "NULL", "pattern": [{"TEXT": {"REGEX": "decrepid"}}]},
{"label": "NULL", "pattern": [{"TEXT": {"REGEX": "Motor\s+?Collision"}}]},
{"label": "NULL", "pattern": [{"TEXT": {"REGEX": "pro\s+?tem"}}]},
{"label": "NULL", "pattern": [{"TEXT": {"REGEX": "Electrotherapeutic"}}]},
{"label": "NULL", "pattern": [{"TEXT": {"REGEX": "megs"}}]},
{"label": "NULL", "pattern": [{"TEXT": {"REGEX": "JRD"}}]},
{"label": "NULL", "pattern": [{"TEXT": {"REGEX": "The Wealth"}}]},
{"label": "NULL", "pattern": [{"TEXT": {"REGEX": "state"}}]},
{"label": "NULL", "pattern": [{"TEXT": {"REGEX": "Trade"}}]},
{"label": "NULL", "pattern": [{"TEXT": {"REGEX": "Atomic"}}]},
{"label": "NULL", "pattern": [{"TEXT": {"REGEX": "Expo"}}]},
{"label": "NULL", "pattern": [{"TEXT": {"REGEX": "looney"}}]},
{"label": "NULL", "pattern": [{"TEXT": {"REGEX": "the\s+?Kingdom"}}]},
{"label": "ORG", "pattern": [{"TEXT": {"REGEX": "([Tt]he\s+?)??Office\s+?of\s+?Strategic\s+?Services"}}]},
{"label": "ORG", "pattern": [{"TEXT": {"REGEX": "([Tt]he\s+?)?Life\s+?Magazine"}}]},
{"label": "ORG", "pattern": [{"TEXT": {"REGEX": "([Tt]he\s+?)?Seattle\s+?Post\s+?Intellegencer"}}]},
{"label": "ORG", "pattern": [{"TEXT": {"REGEX": "Industrias\s+?Cardoen"}}]},
{"label": "NULL", "pattern": [{"TEXT": {"REGEX": "di"}}]},
{"label": "NULL", "pattern": [{"TEXT": {"REGEX": "VENTURE"}}]},
{"label": "NULL", "pattern": [{"TEXT": {"REGEX": "ALLEGED\s+?REAGAN"}}]},
{"label": "NULL", "pattern": [{"TEXT": {"REGEX": "Constitu"}}]},
{"label": "NULL", "pattern": [{"TEXT": {"REGEX": "econonic\s+?aid"}}]},
{"label": "NULL", "pattern": [{"TEXT": {"REGEX": "Edgar\s+?Hoover"}}]},
{"label": "NULL", "pattern": [{"TEXT": {"REGEX": "fed"}}]},
{"label": "NULL", "pattern": [{"TEXT": {"REGEX": "Apollo\s+?Astronauts"}}]},
{"label": "NULL", "pattern": [{"TEXT": {"REGEX": "the\s+?Temple"}}]},
{"label": "NULL", "pattern": [{"TEXT": {"REGEX": "CONNECTED\s+?WACKENHUT\s+?CORP"}}]},
{"label": "NULL", "pattern": [{"TEXT": {"REGEX": "GREAT"}}]},
{"label": "NULL", "pattern": [{"TEXT": {"REGEX": "Lung"}}]},
{"label": "NULL", "pattern": [{"TEXT": {"REGEX": "the\s+?Bill"}}]},
{"label": "NULL", "pattern": [{"TEXT": {"REGEX": "Said"}}]},
{"label": "NULL", "pattern": [{"TEXT": {"REGEX": "CAPACITOR"}}]},
{"label": "NULL", "pattern": [{"TEXT": {"REGEX": "SPARK"}}]},
{"label": "NULL", "pattern": [{"TEXT": {"REGEX": "DAMPED"}}]},
{"label": "NULL", "pattern": [{"TEXT": {"REGEX": "ANEW"}}]},
{"label": "NULL", "pattern": [{"TEXT": {"REGEX": "THEREBY\s+?REDUCING\s+?EFFICIENCY"}}]},
{"label": "NULL", "pattern": [{"TEXT": {"REGEX": "HOWEVER"}}]},
{"label": "NULL", "pattern": [{"TEXT": {"REGEX": "TAPPED"}}]},
{"label": "NULL", "pattern": [{"TEXT": {"REGEX": "ELECTRICALLY"}}]},
{"label": "NULL", "pattern": [{"TEXT": {"REGEX": "ELECTRODE"}}]},
{"label": "NULL", "pattern": [{"TEXT": {"REGEX": "TUNED\s+?CIRCUT"}}]},
{"label": "NULL", "pattern": [{"TEXT": {"REGEX": "WAVELENGTH"}}]},
{"label": "NULL", "pattern": [{"TEXT": {"REGEX": "HORIZONTALLY"}}]},
{"label": "NULL", "pattern": [{"TEXT": {"REGEX": "THE\s+?OUDIN\s+?PRIMARY"}}]},
{"label": "NULL", "pattern": [{"TEXT": {"REGEX": "AUTOTRANSFORMER"}}]},
{"label": "NULL", "pattern": [{"TEXT": {"REGEX": "the Allied"}}]},
{"label": "NULL", "pattern": [{"TEXT": {"REGEX": "the\s+?National"}}]},
{"label": "NULL", "pattern": [{"TEXT": {"REGEX": "Cheez\s+?Whiz"}}]},
{"label": "NULL", "pattern": [{"TEXT": {"REGEX": "the\s+?Rich\s+?Discover\s+?Worthy"}}]},
{"label": "NULL", "pattern": [{"TEXT": {"REGEX": "Examiner"}}]},
{"label": "NULL", "pattern": [{"TEXT": {"REGEX": "msen"}}]},
{"label": "NULL", "pattern": [{"TEXT": {"REGEX": "ORG"}}]},
{"label": "NULL", "pattern": [{"TEXT": {"REGEX": "Physics\s+?A\.\s+?Mathematical"}}]},
{"label": "NULL", "pattern": [{"TEXT": {"REGEX": "PALE"}}]},
{"label": "NULL", "pattern": [{"TEXT": {"REGEX": "Operation"}}]},
{"label": "NULL", "pattern": [{"TEXT": {"REGEX": "Order"}}]},
{"label": "NULL", "pattern": [{"TEXT": {"REGEX": "Command"}}]},
{"label": "NULL", "pattern": [{"TEXT": {"REGEX": "jackboots"}}]},
{"label": "NULL", "pattern": [{"TEXT": {"REGEX": "DARK"}}]},
{"label": "NULL", "pattern": [{"TEXT": {"REGEX": "COLD"}}]},
{"label": "NULL", "pattern": [{"TEXT": {"REGEX": "Long\s+?Range"}}]},
{"label": "NULL", "pattern": [{"TEXT": {"REGEX": "\S+@\S+\.\S+"}}]},
{"label": "NULL", "pattern": [{"TEXT": {"REGEX": "Human\s+?Behavior"}}]},
{"label": "NULL", "pattern": [{"TEXT": {"REGEX": "SINCE\s+?LAWYERS\s+?OCCUPY"}}]},
{"label": "WORK_OF_ART", "pattern": [{"TEXT": {"REGEX": "Digha\s+?Nikaya"}}]},
{"label": "PERSON", "pattern": [{"TEXT": {"REGEX": "(Ludwig\s+?[Vv]an\s+?)?Beethoven"}}]},
{"label": "ORG", "pattern": [{"TEXT": {"REGEX": "([Tt]he\s+?)?Falangist"}}]},
{"label": "ORG", "pattern": [{"TEXT": {"REGEX": "([Tt]he\s+?)?Future\s+?of\s+?Freedom\s+?Foundation"}}]},
{"label": "ORG", "pattern": [{"TEXT": {"REGEX": "([Tt]he\s+?)?Congressional\s+?committee"}}]},
{"label": "ORG", "pattern": [{"TEXT": {"REGEX": "([Tt]he\s+?)?The\s+?Federal\s+?Bureau\s+?of\s+?Prisons"}}]},
{"label": "ORG", "pattern": [{"TEXT": {"REGEX": "([Tt]he\s+?)?American\s+?Red\s+?Cross"}}]},
{"label": "ORG", "pattern": [{"TEXT": {"REGEX": "([Tt]he\s+?)?Annals\s+?of\s+?Internal\s+?Medicine"}}]},
{"label": "ORG", "pattern": [{"TEXT": {"REGEX": "([Tt]he\s+?)?Houston\s+?Post"}}]},
{"label": "ORG", "pattern": [{"TEXT": {"REGEX": "MKULTRA"}}]},
{"label": "ORG", "pattern": [{"TEXT": {"REGEX": "KTOK"}}]},
{"label": "ORG", "pattern": [{"TEXT": {"REGEX": "([Tt]he\s+?)?Houston\s+?Chronicle"}}]},
{"label": "ORG", "pattern": [{"TEXT": {"REGEX": "([Tt]he\s+?)?Concentration\s+?Camp\s+?Program"}}]},
{"label": "ORG", "pattern": [{"TEXT": {"REGEX": "([Tt]he\s+?)?Operation\s+?Garden\s+?Plot"}}]},
{"label": "ORG", "pattern": [{"TEXT": {"REGEX": "([Tt]he\s+?)?federal\s+?government"}}]},
{"label": "ORG", "pattern": [{"TEXT": {"REGEX": "([Tt]he\s+?)?British\s+?intelligence"}}]},
{"label": "ORG", "pattern": [{"TEXT": {"REGEX": "([Tt]he\s+?)?JASON\s+?SOCIETY"}}]},
{"label": "ORG", "pattern": [{"TEXT": {"REGEX": "([Tt]he\s+?)?Composite\s+?Service\s+?Organization"}}]},
{"label": "ORG", "pattern": [{"TEXT": {"REGEX": "([Tt]he\s+?)?Psychological\s+?Operations\s+?Organization"}}]},
{"label": "ORG", "pattern": [{"TEXT": {"REGEX": "([Tt]he\s+?)?Council\s+?on\s+?Foreign\s+?Relations"}}]},
{"label": "ORG", "pattern": [{"TEXT": {"REGEX": "([Tt]he\s+?)?Dept\.\s+?of\s+?Defense"}}]},
{"label": "ORG", "pattern": [{"TEXT": {"REGEX": "([Tt]he\s+?)?Mental\s+?Health\s+?Institution"}}]},
{"label": "ORG", "pattern": [{"TEXT": {"REGEX": "([Tt]he\s+?)?Dept\.\s+?of\s+?Transportation"}}]},
{"label": "ORG", "pattern": [{"TEXT": {"REGEX": "([Tt]he\s+?)?Dept\.\s+?of\s+?Justice"}}]},
{"label": "ORG", "pattern": [{"TEXT": {"REGEX": "([Tt]he\s+?)?L\.E\.A\.F\."}}]},
{"label": "ORG", "pattern": [{"TEXT": {"REGEX": "Junio\s+?Valerio\s+?Black\s+?Prince\s+?Borghese"}}]},
{"label": "ORG", "pattern": [{"TEXT": {"REGEX": "([Tt]he\s+?)?Las\s+?Vegas\s+?Bullet"}}]},
{"label": "ORG", "pattern": [{"TEXT": {"REGEX": "([Tt]he\s+?)?C\.I\.A\."}}]},
{"label": "ORG", "pattern": [{"TEXT": {"REGEX": "([Tt]he\s+?)?J\.C\.\s+?Penney"}}]},
{"label": "ORG", "pattern": [{"TEXT": {"REGEX": "([Tt]he\s+?)?[Aa]-albionic\s+?Research"}}]},
{"label": "ORG", "pattern": [{"TEXT": {"REGEX": "([Tt]he\s+?)?Law\s+?Enforcement\s+?Assistance\s+?Force"}}]},
{"label": "ORG", "pattern": [{"TEXT": {"REGEX": "([Tt]he\s+?)?Young\s+?Americans\s+?for\s+?Freedom"}}]},
{"label": "ORG", "pattern": [{"TEXT": {"REGEX": "([Tt]he\s+?)?Military\s+?Police\s+?Unit"}}]},
{"label": "ORG", "pattern": [{"TEXT": {"REGEX": "([Tt]he\s+?)?Annals"}}]},
{"label": "ORG", "pattern": [{"TEXT": {"REGEX": "([Tt]he\s+?)?Inslaw"}}]},
{"label": "ORG", "pattern": [{"TEXT": {"REGEX": "WACL"}}]},
{"label": "ORG", "pattern": [{"TEXT": {"REGEX": "JDL"}}]},
{"label": "ORG", "pattern": [{"TEXT": {"REGEX": "SJG"}}]},
{"label": "ORG", "pattern": [{"TEXT": {"REGEX": "LEAA"}}]},
{"label": "ORG", "pattern": [{"TEXT": {"REGEX": "Newe Anthropologie"}}]},
{"label": "ORG", "pattern": [{"TEXT": {"REGEX": "NSTAC"}}]},
{"label": "ORG", "pattern": [{"TEXT": {"REGEX": "NEWSBYTES"}}]},
{"label": "ORG", "pattern": [{"TEXT": {"REGEX": "NDPC"}}]},
{"label": "ORG", "pattern": [{"TEXT": {"REGEX": "National\s+?Democratic\s+?Policy\s+?Committee"}}]},
{"label": "ORG", "pattern": [{"TEXT": {"REGEX": "National\s+?Security\s+?Telecommunications\s+?Advisory\s+?Committee"}}]},
{"label": "ORG", "pattern": [{"TEXT": {"REGEX": "Tom Davis Research"}}]},
{"label": "ORG", "pattern": [{"TEXT": {"REGEX": "([Tt]he\s+?)?Civil\s+?Affairs\s+?Operations"}}]},
{"label": "ORG", "pattern": [{"TEXT": {"REGEX": "([Tt]he\s+?)?Civil\s+?Affairs\s+?Organization"}}]},
{"label": "ORG", "pattern": [{"TEXT": {"REGEX": "([Tt]he\s+?)?Big\s+?Brother"}}]},
{"label": "ORG", "pattern": [{"TEXT": {"REGEX": "([Tt]he\s+?)?Big\s+?brother"}}]},
{"label": "ORG", "pattern": [{"TEXT": {"REGEX": "([Tt]he\s+?)?State\s+?Youthful\s+?Offenders\s+?Division"}}]},
{"label": "ORG", "pattern": [{"TEXT": {"REGEX": "([Tt]he\s+?)?The\s+?California\s+?State\s+?Bar's\s+?Standing\s+?Committee\s+?on\s+?Professional\s+?Responsibility\s+?and\s+?Conduct"}}]},
{"label": "ORG", "pattern": [{"TEXT": {"REGEX": "([Tt]he\s+?)?ILLUMINATI"}}]},
{"label": "ORG", "pattern": [{"TEXT": {"REGEX": "([Tt]he\s+?)?Crystic\s+?Institute"}}]},
{"label": "ORG", "pattern": [{"TEXT": {"REGEX": "([Tt]he\s+?)?Crystic\s+?institute"}}]},
{"label": "ORG", "pattern": [{"TEXT": {"REGEX": "([Tt]he\s+?)?New\s+?World\s+?Order"}}]},
{"label": "ORG", "pattern": [{"TEXT": {"REGEX": "([Tt]he\s+?)?Roundtables"}}]},
{"label": "ORG", "pattern": [{"TEXT": {"REGEX": "([Tt]he\s+?)?Club\s+?of\s+?Rome"}}]},
{"label": "ORG", "pattern": [{"TEXT": {"REGEX": "([Tt]he\s+?)?Tri-lateral\s+?Commission"}}]},
{"label": "ORG", "pattern": [{"TEXT": {"REGEX": "([Tt]he\s+?)?Committee\s+?of\s+?300"}}]},
{"label": "ORG", "pattern": [{"TEXT": {"REGEX": "([Tt]he\s+?)?Alaska\s+?Bar\s+?Association"}}]},
{"label": "ORG", "pattern": [{"TEXT": {"REGEX": "([Tt]he\s+?)?Rose\s+?Law\s+?Firm"}}]},
{"label": "ORG", "pattern": [{"TEXT": {"REGEX": "([Tt]he\s+?)?University\s+?of\s+?Wisconsin"}}]},
{"label": "ORG", "pattern": [{"TEXT": {"REGEX": "([Tt]he\s+?)?University\s+?of\s+?Southern\s+?California"}}]},
{"label": "ORG", "pattern": [{"TEXT": {"REGEX": "([Tt]he\s+?)?ROTC"}}]},
{"label": "ORG", "pattern": [{"TEXT": {"REGEX": "([Tt]he\s+?)?Plunge"}}]},
{"label": "EVENT", "pattern": [{"TEXT": {"REGEX": "Fifth\s+?International\s+?Martin\s+?Luther\s+?King\s+?Tribunal"}}]},
{"label": "ORG", "pattern": [{"TEXT": {"REGEX": "([Tt]he\s+?)?Bayshore\s+?Inn"}}]},
{"label": "ORG", "pattern": [{"TEXT": {"REGEX": "([Tt]he\s+?)?Los\s+?Angeles\s+?Sheriff's\s+?Dept\."}}]},
{"label": "ORG", "pattern": [{"TEXT": {"REGEX": "([Tt]he\s+?)?U\.N\.\s+?Security\s+?Council"}}]},
{"label": "ORG", "pattern": [{"TEXT": {"REGEX": "([Tt]he\s+?)?U\.N\."}}]},
{"label": "ORG", "pattern": [{"TEXT": {"REGEX": "([Tt]he\s+?)?Daily\s+?Telegraph"}}]},
{"label": "ORG", "pattern": [{"TEXT": {"REGEX": "([Tt]he\s+?)?Alternative\s+?3"}}]},
{"label": "ORG", "pattern": [{"TEXT": {"REGEX": "([Tt]he\s+?)?Library\s+?of\s+?Congress"}}]},
{"label": "ORG", "pattern": [{"TEXT": {"REGEX": "([Tt]he\s+?)?International\s+?Congress\s+?on\s+?Mental\s+?health"}}]},
{"label": "ORG", "pattern": [{"TEXT": {"REGEX": "([Tt]he\s+?)?National\s+?Academy\s+?of\s+?Sciences"}}]},
{"label": "ORG", "pattern": [{"TEXT": {"REGEX": "([Tt]he\s+?)?U\.\s+?S\s+?.\s+?Secret\s+?Service"}}]},
{"label": "ORG", "pattern": [{"TEXT": {"REGEX": "([Tt]he\s+?)?Office\s+?of\s+?Thrift\s+?Supervision"}}]},
{"label": "ORG", "pattern": [{"TEXT": {"REGEX": "([Tt]he\s+?)?Comptroller\s+?of\s+?the\s+?Currency"}}]},
{"label": "ORG", "pattern": [{"TEXT": {"REGEX": "([Tt]he\s+?)?Urban\s+?Plunge"}}]},
{"label": "ORG", "pattern": [{"TEXT": {"REGEX": "([Tt]he\s+?)?Institute\s+?of\s+?Atmospheric\s+?Physics"}}]},
{"label": "ORG", "pattern": [{"TEXT": {"REGEX": "([Tt]he\s+?)?State\s+?Department"}}]},
{"label": "ORG", "pattern": [{"TEXT": {"REGEX": "([Tt]he\s+?)?National\s+?Libertarian\s+?Party"}}]},
{"label": "ORG", "pattern": [{"TEXT": {"REGEX": "([Tt]he\s+?)?Libertarian\s+?Party"}}]},
{"label": "ORG", "pattern": [{"TEXT": {"REGEX": "([Tt]he\s+?)?Bank\s+?of\s+?England"}}]},
{"label": "ORG", "pattern": [{"TEXT": {"REGEX": "([Tt]he\s+?)?Dept\.\s+?of\s+?State"}}]},
{"label": "ORG", "pattern": [{"TEXT": {"REGEX": "([Tt]he\s+?)?United\s+?Grand\s+?Lodge"}}]},
{"label": "ORG", "pattern": [{"TEXT": {"REGEX": "([Tt]he\s+?)?Board\s+?of\s+?General\s+?Purposes"}}]},
{"label": "ORG", "pattern": [{"TEXT": {"REGEX": "([Tt]he\s+?)?National\s+?Recon\s+?Organization"}}]},
{"label": "ORG", "pattern": [{"TEXT": {"REGEX": "([Tt]he\s+?)?G\.S\.A\."}}]},
{"label": "ORG", "pattern": [{"TEXT": {"REGEX": "([Tt]he\s+?)?United\s+?States\s+?Military\s+?Establishment"}}]},
{"label": "ORG", "pattern": [{"TEXT": {"REGEX": "([Tt]he\s+?)?State\s+?Dept\."}}]},
{"label": "ORG", "pattern": [{"TEXT": {"REGEX": "([Tt]he\s+?)?Sapata\s+?Oil"}}]},
{"label": "ORG", "pattern": [{"TEXT": {"REGEX": "([Tt]he\s+?)?California\s+?National\s+?Guard"}}]},
{"label": "ORG", "pattern": [{"TEXT": {"REGEX": "([Tt]he\s+?)?H\.U\.D\."}}]},
{"label": "ORG", "pattern": [{"TEXT": {"REGEX": "([Tt]he\s+?)?Royal\s+?Bank\s+?of\s+?Canada"}}]},
{"label": "ORG", "pattern": [{"TEXT": {"REGEX": "([Tt]he\s+?)?Bank\s+?of\s+?Montreal"}}]},
{"label": "ORG", "pattern": [{"TEXT": {"REGEX": "([Tt]he\s+?)?Rockefeller\s+?Bank"}}]},
{"label": "ORG", "pattern": [{"TEXT": {"REGEX": "([Tt]he\s+?)?American\s+?Express"}}]},
{"label": "ORG", "pattern": [{"TEXT": {"REGEX": "([Tt]he\s+?)?Loeb\s+?Company"}}]},
{"label": "ORG", "pattern": [{"TEXT": {"REGEX": "([Tt]he\s+?)?H\.E\.W\."}}]},
{"label": "ORG", "pattern": [{"TEXT": {"REGEX": "([Tt]he\s+?)?U\.S\.\s+?Senate"}}]},
{"label": "ORG", "pattern": [{"TEXT": {"REGEX": "([Tt]he\s+?)?Institute\s+?of\s+?Political\s+?Studies"}}]},
{"label": "ORG", "pattern": [{"TEXT": {"REGEX": "([Tt]he\s+?)?law\s+?enforcement"}}]},
{"label": "ORG", "pattern": [{"TEXT": {"REGEX": "([Tt]he\s+?)?Housing\s+?&\s+?Urban\s+?Development"}}]},
{"label": "ORG", "pattern": [{"TEXT": {"REGEX": "([Tt]he\s+?)?Dept\.\s+?of\s+?Education"}}]},
{"label": "NORP", "pattern": [{"TEXT": {"REGEX": "Dropa"}}]},
{"label": "NORP", "pattern": [{"TEXT": {"REGEX": "Fria Ord"}}]},
{"label": "NORP", "pattern": [{"TEXT": {"REGEX": "Haitian"}}]},
{"label": "NORP", "pattern": [{"TEXT": {"REGEX": "Astronauts"}}]},
{"label": "NORP", "pattern": [{"TEXT": {"REGEX": "ISHVARA"}}]},
{"label": "NORP", "pattern": [{"TEXT": {"REGEX": "Wehrwirtschaftsfuehrer"}}]},
{"label": "NORP", "pattern": [{"TEXT": {"REGEX": "Viet\s+?Cong"}}]},
{"label": "NORP", "pattern": [{"TEXT": {"REGEX": "Egyptian"}}]},
{"label": "NORP", "pattern": [{"TEXT": {"REGEX": "Lybian"}}]},
{"label": "NORP", "pattern": [{"TEXT": {"REGEX": "Cuban"}}]},
{"label": "NORP", "pattern": [{"TEXT": {"REGEX": "Japanese"}}]},
{"label": "NORP", "pattern": [{"TEXT": {"REGEX": "Nicaraguan"}}]},
{"label": "NORP", "pattern": [{"TEXT": {"REGEX": "African"}}]},
{"label": "NORP", "pattern": [{"TEXT": {"REGEX": "Indian"}}]},
{"label": "NORP", "pattern": [{"TEXT": {"REGEX": "Icelandic"}}]},
{"label": "NORP", "pattern": [{"TEXT": {"REGEX": "Russian"}}]},
{"label": "NORP", "pattern": [{"TEXT": {"REGEX": "CANADIAN"}}]},
{"label": "NORP", "pattern": [{"TEXT": {"REGEX": "Clandestinism"}}]},
{"label": "NORP", "pattern": [{"TEXT": {"REGEX": "SCIENTOLOGISTS"}}]},
{"label": "NORP", "pattern": [{"TEXT": {"REGEX": "Trilateralists"}}]},
{"label": "NORP", "pattern": [{"TEXT": {"REGEX": "Homo\s+?Sapiens"}}]},
{"label": "NORP", "pattern": [{"TEXT": {"REGEX": "Spaniard"}}]},
{"label": "NORP", "pattern": [{"TEXT": {"REGEX": "Freemason"}}]},
{"label": "NORP", "pattern": [{"TEXT": {"REGEX": "Freemson"}}]},
{"label": "NORP", "pattern": [{"TEXT": {"REGEX": "Freemasonry"}}]},
{"label": "NORP", "pattern": [{"TEXT": {"REGEX": "Masonic"}}]},
{"label": "LOC", "pattern": [{"TEXT": {"REGEX": "Vietnam\s+?Moratorium"}}]},
{"label": "LOC", "pattern": [{"TEXT": {"REGEX": "Homestead\s+?Air\s+?Force\s+?Base"}}]},
{"label": "LOC", "pattern": [{"TEXT": {"REGEX": "Moon"}}]},
{"label": "LOC", "pattern": [{"TEXT": {"REGEX": "Swartz\s+?Creek"}}]},
{"label": "LOC", "pattern": [{"TEXT": {"REGEX": "([Tt]he\s+?)?Marshall\s+?Fields"}}]},
{"label": "LOC", "pattern": [{"TEXT": {"REGEX": "([Tt]he\s+?)?Gulf\s+?of\s+?Mexico"}}]},
{"label": "LOC", "pattern": [{"TEXT": {"REGEX": "DREAMLAND"}}]},
{"label": "LOC", "pattern": [{"TEXT": {"REGEX": "Pharselis"}}]},
{"label": "LOC", "pattern": [{"TEXT": {"REGEX": "Northlake"}}]},
{"label": "LOC", "pattern": [{"TEXT": {"REGEX": "Oak Cliff"}}]},
{"label": "LOC", "pattern": [{"TEXT": {"REGEX": "Paddington\s+?station"}}]},
{"label": "LOC", "pattern": [{"TEXT": {"REGEX": "David\s+?Munson\s+?Air\s+?Base"}}]},
{"label": "LOC", "pattern": [{"TEXT": {"REGEX": "Southeast\s+?Asia"}}]},
{"label": "LOC", "pattern": [{"TEXT": {"REGEX": "Eielson\s+?Air\s+?Force\s+?Base"}}]},
{"label": "LOC", "pattern": [{"TEXT": {"REGEX": "U\.S\.\s+?Army\s+?Reserves"}}]},
{"label": "LOC", "pattern": [{"TEXT": {"REGEX": "Bay\s+?of\s+?Pigs"}}]},
{"label": "LOC", "pattern": [{"TEXT": {"REGEX": "US\s+?Air\s+?Force\s+?L\.\s+?Fletcher\s+?Prouty"}}]},
{"label": "LOC", "pattern": [{"TEXT": {"REGEX": "Bohemian\s+?Grove"}}]},
{"label": "LOC", "pattern": [{"TEXT": {"REGEX": "Broadway"}}]},
{"label": "LOC", "pattern": [{"TEXT": {"REGEX": "Mockingbird Lane"}}]},
{"label": "LOC", "pattern": [{"TEXT": {"REGEX": "Mockingbird"}}]},
{"label": "LOC", "pattern": [{"TEXT": {"REGEX": "LAOS"}}]},
{"label": "LOC", "pattern": [{"TEXT": {"REGEX": "Hyannisport"}}]},
{"label": "LOC", "pattern": [{"TEXT": {"REGEX": "Lop Nor"}}]},
{"label": "LOC", "pattern": [{"TEXT": {"REGEX": "Gross Wannsee"}}]},
{"label": "LOC", "pattern": [{"TEXT": {"REGEX": "Groom Lake"}}]},
{"label": "NORP", "pattern": [{"TEXT": {"REGEX": "North\s+?American"}}]},
{"label": "NORP", "pattern": [{"TEXT": {"REGEX": "Monk"}}]},
{"label": "NULL", "pattern": [{"TEXT": {"REGEX": "Northern"}}]},
{"label": "GPE", "pattern": [{"TEXT": {"REGEX": "Allanwood"}}]},
{"label": "GPE", "pattern": [{"TEXT": {"REGEX": "Roswell"}}]},
{"label": "GPE", "pattern": [{"TEXT": {"REGEX": "Westminster"}}]},
{"label": "GPE", "pattern": [{"TEXT": {"REGEX": "Portland"}}]},
{"label": "GPE", "pattern": [{"TEXT": {"REGEX": "Richmond"}}]},
{"label": "GPE", "pattern": [{"TEXT": {"REGEX": "Sacramento"}}]},
{"label": "GPE", "pattern": [{"TEXT": {"REGEX": "St\s+?Louis"}}]},
{"label": "GPE", "pattern": [{"TEXT": {"REGEX": "New\s+?Haven"}}]},
{"label": "GPE", "pattern": [{"TEXT": {"REGEX": "Milwaukee"}}]},
{"label": "GPE", "pattern": [{"TEXT": {"REGEX": "Little\s+?Rock"}}]},
{"label": "GPE", "pattern": [{"TEXT": {"REGEX": "Los\s+?Angeles"}}]},
{"label": "GPE", "pattern": [{"TEXT": {"REGEX": "Briton"}}]},
{"label": "GPE", "pattern": [{"TEXT": {"REGEX": "El\s+?Paso"}}]},
{"label": "GPE", "pattern": [{"TEXT": {"REGEX": "Columbia"}}]},
{"label": "GPE", "pattern": [{"TEXT": {"REGEX": "FORT\s+?HUNT"}}]},
{"label": "GPE", "pattern": [{"TEXT": {"REGEX": "Butte"}}]},
{"label": "GPE", "pattern": [{"TEXT": {"REGEX": "State\s+?College"}}]},
{"label": "GPE", "pattern": [{"TEXT": {"REGEX": "Williamsport"}}]},
{"label": "GPE", "pattern": [{"TEXT": {"REGEX": "Landsdale"}}]},
{"label": "GPE", "pattern": [{"TEXT": {"REGEX": "Newtown\s+?Square"}}]},
{"label": "GPE", "pattern": [{"TEXT": {"REGEX": "Allentown"}}]},
{"label": "GPE", "pattern": [{"TEXT": {"REGEX": "New\s+?Castle"}}]},
{"label": "GPE", "pattern": [{"TEXT": {"REGEX": "Beckley"}}]},
{"label": "GPE", "pattern": [{"TEXT": {"REGEX": "Alton"}}]},
{"label": "GPE", "pattern": [{"TEXT": {"REGEX": "Stubenville"}}]},
{"label": "GPE", "pattern": [{"TEXT": {"REGEX": "Buffalo"}}]},
{"label": "GPE", "pattern": [{"TEXT": {"REGEX": "Belleville"}}]},
{"label": "GPE", "pattern": [{"TEXT": {"REGEX": "Bloomington"}}]},
{"label": "GPE", "pattern": [{"TEXT": {"REGEX": "Carbondale"}}]},
{"label": "GPE", "pattern": [{"TEXT": {"REGEX": "Champaign"}}]},
{"label": "GPE", "pattern": [{"TEXT": {"REGEX": "Constitantinoble"}}]},
{"label": "GPE", "pattern": [{"TEXT": {"REGEX": "Rock\s+?Island"}}]},
{"label": "GPE", "pattern": [{"TEXT": {"REGEX": "Birmingham"}}]},
{"label": "GPE", "pattern": [{"TEXT": {"REGEX": "IRAQ"}}]},
{"label": "GPE", "pattern": [{"TEXT": {"REGEX": "CANADA"}}]},
{"label": "GPE", "pattern": [{"TEXT": {"REGEX": "ENGLAND"}}]},
{"label": "GPE", "pattern": [{"TEXT": {"REGEX": "Danvers"}}]},
{"label": "GPE", "pattern": [{"TEXT": {"REGEX": "GERMANY"}}]},
{"label": "GPE", "pattern": [{"TEXT": {"REGEX": "Montreal"}}]},
{"label": "GPE", "pattern": [{"TEXT": {"REGEX": "Canada"}}]},
{"label": "GPE", "pattern": [{"TEXT": {"REGEX": "Baltimore"}}]},
{"label": "GPE", "pattern": [{"TEXT": {"REGEX": "Anchorage"}}]},
{"label": "GPE", "pattern": [{"TEXT": {"REGEX": "Albuquerque"}}]},
{"label": "GPE", "pattern": [{"TEXT": {"REGEX": "Albany"}}]},
{"label": "GPE", "pattern": [{"TEXT": {"REGEX": "CORAL\s+?GABLES"}}]},
{"label": "GPE", "pattern": [{"TEXT": {"REGEX": "Washington\s+?DC"}}]},
{"label": "GPE", "pattern": [{"TEXT": {"REGEX": "Washington,\s+?DC"}}]},
{"label": "GPE", "pattern": [{"TEXT": {"REGEX": "Washington,\s+?D\.C\."}}]},
{"label": "GPE", "pattern": [{"TEXT": {"REGEX": "Avon\s+?Park"}}]},
{"label": "GPE", "pattern": [{"TEXT": {"REGEX": "Mill\s+?Point"}}]},
{"label": "GPE", "pattern": [{"TEXT": {"REGEX": "El\s+?Reno"}}]},
{"label": "GPE", "pattern": [{"TEXT": {"REGEX": "Zagreb"}}]},
{"label": "GPE", "pattern": [{"TEXT": {"REGEX": "ZAGREB"}}]},
{"label": "LOC", "pattern": [{"TEXT": {"REGEX": "Whiskey\s+?Flat"}}]},
{"label": "LOC", "pattern": [{"TEXT": {"REGEX": "site\s+?S-4"}}]},
{"label": "LOC", "pattern": [{"TEXT": {"REGEX": "LUNA"}}]},
{"label": "LOC", "pattern": [{"TEXT": {"REGEX": "AREA\s+?51"}}]},
{"label": "LOC", "pattern": [{"TEXT": {"REGEX": "El\s+?Salvador\s+?air\s+?base"}}]},
{"label": "LOC", "pattern": [{"TEXT": {"REGEX": "Greenville\s+?County"}}]},
{"label": "LOC", "pattern": [{"TEXT": {"REGEX": "Toyland"}}]},
{"label": "LOC", "pattern": [{"TEXT": {"REGEX": "Sun"}}]},
{"label": "LOC", "pattern": [{"TEXT": {"REGEX": "North\s+?Pole"}}]},
{"label": "LOC", "pattern": [{"TEXT": {"REGEX": "Qua\s+?Vieaf"}}]},
{"label": "LOC", "pattern": [{"TEXT": {"REGEX": "North\s+?Viet\s+?Nam"}}]},
{"label": "LOC", "pattern": [{"TEXT": {"REGEX": "Owl's\s+?Nest"}}]},
{"label": "LOC", "pattern": [{"TEXT": {"REGEX": "Dachau"}}]},
{"label": "LOC", "pattern": [{"TEXT": {"REGEX": "Hill\s+?Billies"}}]},
{"label": "LOC", "pattern": [{"TEXT": {"REGEX": "Cave\s+?Man's"}}]},
{"label": "LOC", "pattern": [{"TEXT": {"REGEX": "Cherry\s+?Brook\s+?Valley"}}]},
{"label": "LOC", "pattern": [{"TEXT": {"REGEX": "Postbridge"}}]},
{"label": "LOC", "pattern": [{"TEXT": {"REGEX": "McDill\s+?Air\s+?Force\s+?Base"}}]},
{"label": "LOC", "pattern": [{"TEXT": {"REGEX": "Egland\s+?Air\s+?Force\s+?Base"}}]},
{"label": "GPE", "pattern": [{"TEXT": {"REGEX": "Nazareth"}}]},
{"label": "GPE", "pattern": [{"TEXT": {"REGEX": "Latin\s+?America"}}]},
{"label": "GPE", "pattern": [{"TEXT": {"REGEX": "GREAT\s+?BRITAIN"}}]},
{"label": "GPE", "pattern": [{"TEXT": {"REGEX": "U\.S\.S\.R"}}]},
{"label": "GPE", "pattern": [{"TEXT": {"REGEX": "New\s+?York\s+?City"}}]},
{"label": "GPE", "pattern": [{"TEXT": {"REGEX": "Soviet\s+?Union"}}]},
{"label": "GPE", "pattern": [{"TEXT": {"REGEX": "Viet\s+?Nam"}}]},
{"label": "GPE", "pattern": [{"TEXT": {"REGEX": "U\.S\.A\."}}]},
{"label": "GPE", "pattern": [{"TEXT": {"REGEX": "U\.S\."}}]},
{"label": "GPE", "pattern": [{"TEXT": {"REGEX": "U\.S"}}]},
{"label": "GPE", "pattern": [{"TEXT": {"REGEX": "Saudi\s+?Arabia"}}]},
{"label": "GPE", "pattern": [{"TEXT": {"REGEX": "Washington,\s+?DC"}}]},
{"label": "GPE", "pattern": [{"TEXT": {"REGEX": "the\s+?Vatican"}}]},
{"label": "GPE", "pattern": [{"TEXT": {"REGEX": "([Tt]he\s+?)?British\s+?Empire"}}]},
{"label": "GPE", "pattern": [{"TEXT": {"REGEX": "Pennsylvania"}}]},
{"label": "ORG", "pattern": [{"TEXT": {"REGEX": "([Tt]he\s+?)?Warren\s+?Commission"}}]},
{"label": "ORG", "pattern": [{"TEXT": {"REGEX": "([Tt]he\s+?)?Tesla\s+?Electric\s+?Company"}}]},
{"label": "ORG", "pattern": [{"TEXT": {"REGEX": "([Tt]he\s+?)?Hahn\s+?group"}}]},
{"label": "ORG", "pattern": [{"TEXT": {"REGEX": "([Tt]he\s+?)?U\.S\.\s+?Department\s+?of\s+?Health\s+?and\s+?Human\s+?Services"}}]},
{"label": "ORG", "pattern": [{"TEXT": {"REGEX": "([Tt]he\s+?)?American\s+?Council\s+?of\s+?Life\s+?Insurance"}}]},
{"label": "ORG", "pattern": [{"TEXT": {"REGEX": "([Tt]he\s+?)?Health\s+?Insurance\s+?Association\s+?of\s+?America\."}}]},
{"label": "ORG", "pattern": [{"TEXT": {"REGEX": "([Tt]he\s+?)?U\.S\.\s+?District\s+?Court"}}]},
{"label": "ORG", "pattern": [{"TEXT": {"REGEX": "([Tt]he\s+?)?United\s+?States\s+?Pacific\s+?Fleet"}}]},
{"label": "ORG", "pattern": [{"TEXT": {"REGEX": "([Tt]he\s+?)?Pacific\s+?Fleet"}}]},
{"label": "ORG", "pattern": [{"TEXT": {"REGEX": "([Tt]he\s+?)?National\s+?Guard"}}]},
{"label": "ORG", "pattern": [{"TEXT": {"REGEX": "([Tt]he\s+?)?Christic\s+?Institute"}}]},
{"label": "ORG", "pattern": [{"TEXT": {"REGEX": "([Tt]he\s+?)?Nugan-Hand"}}]},
{"label": "ORG", "pattern": [{"TEXT": {"REGEX": "([Tt]he\s+?)?Military-Industrial\s+?complex"}}]},
{"label": "ORG", "pattern": [{"TEXT": {"REGEX": "([Tt]he\s+?)?Condon\s+?Committee"}}]},
{"label": "ORG", "pattern": [{"TEXT": {"REGEX": "COMMUNICATIONS\s+?CANADA"}}]},
{"label": "ORG", "pattern": [{"TEXT": {"REGEX": "([Tt]he\s+?)?ABC"}}]},
{"label": "ORG", "pattern": [{"TEXT": {"REGEX": "([Tt]he\s+?)?CBS"}}]},
{"label": "ORG", "pattern": [{"TEXT": {"REGEX": "([Tt]he\s+?)?Unity\s+?movement"}}]},
{"label": "ORG", "pattern": [{"TEXT": {"REGEX": "([Tt]he\s+?)?Covert\s+?Action"}}]},
{"label": "ORG", "pattern": [{"TEXT": {"REGEX": "([Tt]he\s+?)?General\s+?Electric"}}]},
{"label": "ORG", "pattern": [{"TEXT": {"REGEX": "([Tt]he\s+?)?US\s+?Army"}}]},
{"label": "ORG", "pattern": [{"TEXT": {"REGEX": "([Tt]he\s+?)?US\s+?Navy"}}]},
{"label": "ORG", "pattern": [{"TEXT": {"REGEX": "([Tt]he\s+?)?London\s+?Underground"}}]},
{"label": "ORG", "pattern": [{"TEXT": {"REGEX": "([Tt]he\s+?)?US\s+?Justice\s+?Department"}}]},
{"label": "ORG", "pattern": [{"TEXT": {"REGEX": "([Tt]he\s+?)?Sur\s+?Coester"}}]},
{"label": "ORG", "pattern": [{"TEXT": {"REGEX": "([Tt]he\s+?)?Colorado\s+?Springs\s+?Gazette"}}]},
{"label": "ORG", "pattern": [{"TEXT": {"REGEX": "([Tt]he\s+?)?SoftQuad\s+?Inc\."}}]},
{"label": "ORG", "pattern": [{"TEXT": {"REGEX": "([Tt]he\s+?)?United\s+?Nations\s+?World\s+?Health\s+?Organization"}}]},
{"label": "ORG", "pattern": [{"TEXT": {"REGEX": "([Tt]he\s+?)?World\s+?Health\s+?Organization"}}]},
{"label": "ORG", "pattern": [{"TEXT": {"REGEX": "([Tt]he\s+?)?L\.E\.A\.A\."}}]},
{"label": "ORG", "pattern": [{"TEXT": {"REGEX": "([Tt]he\s+?)?Justice\s+?Dept\."}}]},
{"label": "ORG", "pattern": [{"TEXT": {"REGEX": "(THE\s+?)?HOUSE\s+?JUDICIARY\s+?COMMITTEE"}}]},
{"label": "ORG", "pattern": [{"TEXT": {"REGEX": "([Tt]he\s+?)?Contemporary\s+?Research,\s+?Inc\."}}]},
{"label": "ORG", "pattern": [{"TEXT": {"REGEX": "([Tt]he\s+?)?British\s+?Government"}}]},
{"label": "ORG", "pattern": [{"TEXT": {"REGEX": "([Tt]he\s+?)?House\s+?Select\s+?Committee\s+?on\s+?Assassinations"}}]},
{"label": "ORG", "pattern": [{"TEXT": {"REGEX": "([Tt]he\s+?)?Office\s+?of\s+?Criminal\s+?Justice\s+?Planning"}}]},
{"label": "ORG", "pattern": [{"TEXT": {"REGEX": "([Tt]he\s+?)?Dept\.\s+?of\s+?Defense\s+?of\s+?the\s+?United\s+?States"}}]},
{"label": "ORG", "pattern": [{"TEXT": {"REGEX": "([Tt]he\s+?)?Dept\.\s+?of\s+?Commerce"}}]},
{"label": "ORG", "pattern": [{"TEXT": {"REGEX": "([Tt]he\s+?)?MAJIC"}}]},
{"label": "ORG", "pattern": [{"TEXT": {"REGEX": "Mother\s+?Jones"}}]},
{"label": "ORG", "pattern": [{"TEXT": {"REGEX": "([Tt]he\s+?)?W\.A\.\s+?Harriman\s+?&\s+?Co\."}}]},
{"label": "ORG", "pattern": [{"TEXT": {"REGEX": "([Tt]he\s+?)?Securities\s+?C\.\s+?Commission"}}]},
{"label": "ORG", "pattern": [{"TEXT": {"REGEX": "([Tt]he\s+?)?Rockefeller\s+?Syndicate"}}]},
{"label": "ORG", "pattern": [{"TEXT": {"REGEX": "([Tt]he\s+?)?Dallas\s+?Citizens\s+?Council"}}]},
{"label": "ORG", "pattern": [{"TEXT": {"REGEX": "([Tt]he\s+?)?Round\s+?Table\s+?groups"}}]},
{"label": "ORG", "pattern": [{"TEXT": {"REGEX": "([Tt]he\s+?)?Devon\s+?Unidentified\s+?Flying\s+?Objects\s+?centre"}}]},
{"label": "ORG", "pattern": [{"TEXT": {"REGEX": "([Tt]he\s+?)?San\s+?Francisco\s+?Sidewalk\s+?Astronomers\s+?Association"}}]},
{"label": "ORG", "pattern": [{"TEXT": {"REGEX": "([Tt]he\s+?)?Dartmoor\s+?Livestock\s+?Protection\s+?Society"}}]},
{"label": "ORG", "pattern": [{"TEXT": {"REGEX": "([Tt]he\s+?)?Animal\s+?Defence\s+?Society"}}]},
{"label": "ORG", "pattern": [{"TEXT": {"REGEX": "([Tt]he\s+?)?U\.S\.\s+?Agriculture\s+?Department"}}]},
{"label": "ORG", "pattern": [{"TEXT": {"REGEX": "([Tt]he\s+?)?Perpetual\s+?Hidden\s+?Government"}}]},
{"label": "ORG", "pattern": [{"TEXT": {"REGEX": "([Tt]he\s+?)?Office\s+?of\s+?the\s+?Deputy\s+?Chief\s+?of\s+?Staff\s+?of\s+?Personnel"}}]},
{"label": "ORG", "pattern": [{"TEXT": {"REGEX": "([Tt]he\s+?)?Harriman\s+?Fifteen\s+?Corp\."}}]},
{"label": "ORG", "pattern": [{"TEXT": {"REGEX": "([Tt]he\s+?)?Israeli\s+?secret\s+?service"}}]},
{"label": "ORG", "pattern": [{"TEXT": {"REGEX": "([Tt]he\s+?)?Bell\s+?Helicopter"}}]},
{"label": "ORG", "pattern": [{"TEXT": {"REGEX": "([Tt]he\s+?)?Eurocheque"}}]},
{"label": "ORG", "pattern": [{"TEXT": {"REGEX": "([Tt]he\s+?)?MasterCard"}}]},
{"label": "ORG", "pattern": [{"TEXT": {"REGEX": "([Tt]he\s+?)?CitiCorp"}}]},
{"label": "ORG", "pattern": [{"TEXT": {"REGEX": "([Tt]he\s+?)?Marconi\s+?company"}}]},
{"label": "ORG", "pattern": [{"TEXT": {"REGEX": "([Tt]he\s+?)?Marconi"}}]},
{"label": "ORG", "pattern": [{"TEXT": {"REGEX": "([Tt]he\s+?)?NRO"}}]},
{"label": "ORG", "pattern": [{"TEXT": {"REGEX": "([Tt]he\s+?)?Illinois\s+?Bell\s+?Telephone\s+?Company"}}]},
{"label": "ORG", "pattern": [{"TEXT": {"REGEX": "([Tt]he\s+?)?UFO\s+?Magazine"}}]},
{"label": "ORG", "pattern": [{"TEXT": {"REGEX": "([Tt]he\s+?)?University\s+?of\s+?Toronto"}}]},
{"label": "ORG", "pattern": [{"TEXT": {"REGEX": "([Tt]he\s+?)?Taymar,\s+?Inc\."}}]},
{"label": "ORG", "pattern": [{"TEXT": {"REGEX": "([Tt]he\s+?)?Kerr\s+?McGree\s+?Nuclear\s+?Corporation"}}]},
{"label": "ORG", "pattern": [{"TEXT": {"REGEX": "([Tt]he\s+?)?Edward\s+?R\.\s+?Murrow\s+?Center"}}]},
{"label": "ORG", "pattern": [{"TEXT": {"REGEX": "([Tt]he\s+?)?American\s+?Ship\s+?and\s+?Commerce\s+?Corp\."}}]},
{"label": "ORG", "pattern": [{"TEXT": {"REGEX": "([Tt]he\s+?)?American\s+?government"}}]},
{"label": "ORG", "pattern": [{"TEXT": {"REGEX": "([Tt]he\s+?)?Silverado\s+?Savings\s+?&\s+?Loan"}}]},
{"label": "ORG", "pattern": [{"TEXT": {"REGEX": "([Tt]he\s+?)?American\s+?Association\s+?for\s+?the\s+?Advancement\s+?of\s+?Science"}}]},
{"label": "ORG", "pattern": [{"TEXT": {"REGEX": "([Tt]he\s+?)?G\.H\.\s+?Walker\s+?&\s+?Co\."}}]},
{"label": "ORG", "pattern": [{"TEXT": {"REGEX": "([Tt]he\s+?)?Office\s+?of\s+?Preparedness"}}]},
{"label": "ORG", "pattern": [{"TEXT": {"REGEX": "([Tt]he\s+?)?National\s+?Aeronautics\s+?and\s+?Space\s+?Agency"}}]},
{"label": "ORG", "pattern": [{"TEXT": {"REGEX": "([Tt]he\s+?)?Martin\s+?Marietta\s+?company"}}]},
{"label": "ORG", "pattern": [{"TEXT": {"REGEX": "([Tt]he\s+?)?John\s+?Birch\s+?Society"}}]},
{"label": "ORG", "pattern": [{"TEXT": {"REGEX": "([Tt]he\s+?)?U\.S\.\s+?government"}}]},
{"label": "ORG", "pattern": [{"TEXT": {"REGEX": "([Tt]he\s+?)?Cleveland\s+?Field\s+?Office"}}]},
{"label": "ORG", "pattern": [{"TEXT": {"REGEX": "The\s+?John\s+?F\.\s+?Kennedy\s+?Center\s+?for\s+?the\s+?Performing\s+?Arts\s+?Education\s+?Program"}}]},
{"label": "ORG", "pattern": [{"TEXT": {"REGEX": "([Tt]he\s+?)?United\s+?States\s+?government"}}]},
{"label": "ORG", "pattern": [{"TEXT": {"REGEX": "([Tt]he\s+?)?National\s+?Institute\s+?of\s+?Agrarian\s+?Reform"}}]},
{"label": "ORG", "pattern": [{"TEXT": {"REGEX": "([Tt]he\s+?)?National\s+?Institute\s+?of\s+?Mental\s+?Health"}}]},
{"label": "ORG", "pattern": [{"TEXT": {"REGEX": "([Tt]he\s+?)?INRA"}}]},
{"label": "ORG", "pattern": [{"TEXT": {"REGEX": "([Tt]he\s+?)?Massachusetts\s+?Institute\s+?of\s+?Technology"}}]},
{"label": "ORG", "pattern": [{"TEXT": {"REGEX": "([Tt]he\s+?)?International\s+?UFO\s+?Bureau"}}]},
{"label": "ORG", "pattern": [{"TEXT": {"REGEX": "([Tt]he\s+?)?National\s+?Investigations\s+?Committee\s+?on\s+?Aerial\s+?Phenomena"}}]},
{"label": "ORG", "pattern": [{"TEXT": {"REGEX": "([Tt]he\s+?)?Dept\.\s+?of\s+?the\s+?Army"}}]},
{"label": "ORG", "pattern": [{"TEXT": {"REGEX": "([Tt]he\s+?)?U\.S\.\s+?Air\s+?Force"}}]},
{"label": "ORG", "pattern": [{"TEXT": {"REGEX": "([Tt]he\s+?)?University\s+?of\s+?Arizona"}}]},
{"label": "ORG", "pattern": [{"TEXT": {"REGEX": "([Tt]he\s+?)?Criminal\s+?Justice\s+?System"}}]},
{"label": "ORG", "pattern": [{"TEXT": {"REGEX": "([Tt]he\s+?)?Sheriffs\s+?Dept\."}}]},
{"label": "ORG", "pattern": [{"TEXT": {"REGEX": "Standard\s+?&\s+?Poors"}}]},
{"label": "ORG", "pattern": [{"TEXT": {"REGEX": "([Tt]he\s+?)?Federal\s+?Reserve"}}]},
{"label": "ORG", "pattern": [{"TEXT": {"REGEX": "([Tt]he\s+?)?L\.E\.A\.A\.\s+?Newsletter"}}]},
{"label": "ORG", "pattern": [{"TEXT": {"REGEX": "([Tt]he\s+?)?Texas\s+?School\s+?Book\s+?Depository"}}]},
{"label": "ORG", "pattern": [{"TEXT": {"REGEX": "([Tt]he\s+?)?Senate\s+?Select\s+?Committee\s+?on\s+?Intelligence"}}]},
{"label": "ORG", "pattern": [{"TEXT": {"REGEX": "([Tt]he\s+?)?Senate\s+?Intelligence\s+?Committee"}}]},
{"label": "ORG", "pattern": [{"TEXT": {"REGEX": "([Tt]he\s+?)?House\s+?Committee\s+?on\s+?Assassinations"}}]},
{"label": "ORG", "pattern": [{"TEXT": {"REGEX": "([Tt]he\s+?)?House\s+?Committee"}}]},
{"label": "ORG", "pattern": [{"TEXT": {"REGEX": "([Tt]he\s+?)?House\s+?Select\s+?Committee"}}]},
{"label": "ORG", "pattern": [{"TEXT": {"REGEX": "([Tt]he\s+?)?House\s+?of\s+?Saud"}}]},
{"label": "ORG", "pattern": [{"TEXT": {"REGEX": "Wackenhut"}}]},
{"label": "ORG", "pattern": [{"TEXT": {"REGEX": "KPFA"}}]},
{"label": "ORG", "pattern": [{"TEXT": {"REGEX": "WACKENHUT\s+?CORP"}}]},
{"label": "ORG", "pattern": [{"TEXT": {"REGEX": "([Tt]he\s+?)?International\s+?Agency\s+?for\s+?Research\s+?on\s+?Cancer"}}]},
{"label": "ORG", "pattern": [{"TEXT": {"REGEX": "([Tt]he\s+?)?The\s+?London\s+?Times"}}]},
{"label": "ORG", "pattern": [{"TEXT": {"REGEX": "WACKENHUT"}}]},
{"label": "ORG", "pattern": [{"TEXT": {"REGEX": "([Tt]he\s+?)?Federal\s+?Emergency\s+?Management\s+?Agency"}}]},
{"label": "ORG", "pattern": [{"TEXT": {"REGEX": "([Tt]he\s+?)?Special\s+?Forces\s+?Unit"}}]},
{"label": "ORG", "pattern": [{"TEXT": {"REGEX": "([Tt]he\s+?)?Defense\s+?Central\s+?Index"}}]},
{"label": "ORG", "pattern": [{"TEXT": {"REGEX": "([Tt]he\s+?)?Joint\s+?Chiefs\s+?of\s+?Staff"}}]},
{"label": "ORG", "pattern": [{"TEXT": {"REGEX": "([Tt]he\s+?)?Federal\s+?Information\s+?Center"}}]},
{"label": "ORG", "pattern": [{"TEXT": {"REGEX": "([Tt]he\s+?)?American\s+?Counter-intelligence\s+?Corps"}}]},
{"label": "ORG", "pattern": [{"TEXT": {"REGEX": "CIC"}}]},
{"label": "ORG", "pattern": [{"TEXT": {"REGEX": "([Tt]he\s+?)?British\s+?Establishment\."}}]},
lots of stuff, python names, xquery to html, xml regex clean-ish
2023-04-24 14:37:38 -04:00
{"label": "PERSON", "pattern": [{"TEXT": {"REGEX": "[A-Z][a-z]+\s+([A-Z]\.\s+[A-Z][a-z]+)?(,\s+Jr\.)"}}]},
{"label": "PERSON", "pattern": [{"TEXT": {"REGEX": "[A-Z][a-z]+\s+([A-Z]\.\s+[A-Z][a-z]+)?(\s+Jr\.)"}}]},
{"label": "PERSON", "pattern": [{"TEXT": {"REGEX": "[A-Z][a-z]+\s+(([A-Z]\.\s+)?[A-Z][a-z]+([A-Z][a-z]+)?)?\s+Jr\."}}]},
{"label": "PERSON", "pattern": [{"TEXT": {"REGEX": "[[A-Z][a-z]+\s+[A-Z]\.\s+([A-Z][a-z])?[A-Z][a-z]+"}}]},
trying my best to fix the regex and editing python code
2023-04-27 00:54:09 -04:00
{"label": "PERSON", "pattern": [{"TEXT": {"REGEX": "Nostradameus"}}]},
{"label": "PERSON", "pattern": [{"TEXT": {"REGEX": "Harrison\s+?Edward\s+?Livingstone"}}]},
{"label": "PERSON", "pattern": [{"TEXT": {"REGEX": "J\.\s+?Herbert\s+?Sawyer"}}]},
{"label": "PERSON", "pattern": [{"TEXT": {"REGEX": "L\.\s+?Harmon\s+?Zeigler"}}]},
{"label": "PERSON", "pattern": [{"TEXT": {"REGEX": "C\.\s+?Everett\s+?Koop"}}]},
{"label": "PERSON", "pattern": [{"TEXT": {"REGEX": "Edward\s+?L\.\s+?van\s+?Roden"}}]},
{"label": "PERSON", "pattern": [{"TEXT": {"REGEX": "Kirschbaum"}}]},
{"label": "PERSON", "pattern": [{"TEXT": {"REGEX": "Raphael\s+?Shumacker"}}]},
{"label": "PERSON", "pattern": [{"TEXT": {"REGEX": "Martin\s+?Marietta"}}]},
{"label": "PERSON", "pattern": [{"TEXT": {"REGEX": "DAVID\s+?MELLOR"}}]},
{"label": "PERSON", "pattern": [{"TEXT": {"REGEX": "J\.\s+?W\.\s+?Willmott"}}]},
{"label": "PERSON", "pattern": [{"TEXT": {"REGEX": "O'Neill"}}]},
{"label": "PERSON", "pattern": [{"TEXT": {"REGEX": "O'Neil"}}]},
{"label": "PERSON", "pattern": [{"TEXT": {"REGEX": "O'Neal"}}]},
{"label": "PERSON", "pattern": [{"TEXT": {"REGEX": "Orsenigo"}}]},
{"label": "PERSON", "pattern": [{"TEXT": {"REGEX": "Neil\s+?Bush"}}]},
{"label": "PERSON", "pattern": [{"TEXT": {"REGEX": "SWAMI\s+?VIVEKANANDA"}}]},
{"label": "PERSON", "pattern": [{"TEXT": {"REGEX": "Morris\s+?Ellowitz"}}]},
{"label": "PERSON", "pattern": [{"TEXT": {"REGEX": "Harry\s+?Thon"}}]},
{"label": "PERSON", "pattern": [{"TEXT": {"REGEX": "David\s+?Ferrie"}}]},
{"label": "PERSON", "pattern": [{"TEXT": {"REGEX": "Darrell\s+?W\.\s+?Garner"}}]},
{"label": "PERSON", "pattern": [{"TEXT": {"REGEX": "David\s+?Goldstein"}}]},
{"label": "PERSON", "pattern": [{"TEXT": {"REGEX": "Mona\s+?B\.\s+?Saenz"}}]},
{"label": "PERSON", "pattern": [{"TEXT": {"REGEX": "Abraham\s+?Zapruder"}}]},
{"label": "PERSON", "pattern": [{"TEXT": {"REGEX": "Clayton\s+?Fowler"}}]},
{"label": "PERSON", "pattern": [{"TEXT": {"REGEX": "WILLIAM\s+?L\.\s+?LAW"}}]},
{"label": "PERSON", "pattern": [{"TEXT": {"REGEX": "Reynolds"}}]},
{"label": "PERSON", "pattern": [{"TEXT": {"REGEX": "Marriner S. Eccles"}}]},
{"label": "PERSON", "pattern": [{"TEXT": {"REGEX": "H\.S\.M\.\s+?Coxeter"}}]},
{"label": "PERSON", "pattern": [{"TEXT": {"REGEX": "A\.D\.\s+?Bowie"}}]},
{"label": "PERSON", "pattern": [{"TEXT": {"REGEX": "Phillip\s+?Geraci"}}]},
{"label": "PERSON", "pattern": [{"TEXT": {"REGEX": "E\.R\.\s+?Walthers"}}]},
{"label": "PERSON", "pattern": [{"TEXT": {"REGEX": "George\s+?McGann"}}]},
{"label": "PERSON", "pattern": [{"TEXT": {"REGEX": "Perry\s+?Russo"}}]},
{"label": "PERSON", "pattern": [{"TEXT": {"REGEX": "JOHN\s+?PHILIP\s+?NICHOLS"}}]},
{"label": "PERSON", "pattern": [{"TEXT": {"REGEX": "SALVADOR\s+?ALLENDE"}}]},
{"label": "PERSON", "pattern": [{"TEXT": {"REGEX": "D\.\s+?CASOLARO"}}]},
{"label": "PERSON", "pattern": [{"TEXT": {"REGEX": "AFFIDAVIT"}}]},
{"label": "PERSON", "pattern": [{"TEXT": {"REGEX": "Koreshan"}}]},
{"label": "PERSON", "pattern": [{"TEXT": {"REGEX": "Koresh"}}]},
{"label": "PERSON", "pattern": [{"TEXT": {"REGEX": "Carl\s+?Oglesby"}}]},
{"label": "PERSON", "pattern": [{"TEXT": {"REGEX": "Earline\s+?Roberts"}}]},
{"label": "PERSON", "pattern": [{"TEXT": {"REGEX": "Lee\s+?Bowers,\s+?Jr\."}}]},
{"label": "PERSON", "pattern": [{"TEXT": {"REGEX": "Locust"}}]},
restores xml root element to pre-src
2023-04-28 02:15:04 -04:00
{"label": "PERSON", "pattern": [{"TEXT": {"REGEX": "Starmaster"}}]},
trying my best to fix the regex and editing python code
2023-04-27 00:54:09 -04:00
{"label": "PERSON", "pattern": [{"TEXT": {"REGEX": "Nuri\s+?Al-Said"}}]},
{"label": "PERSON", "pattern": [{"TEXT": {"REGEX": "Pamela\s+?Courson"}}]},
{"label": "PERSON", "pattern": [{"TEXT": {"REGEX": "Jimmy\s+?Hendrix"}}]},
{"label": "PERSON", "pattern": [{"TEXT": {"REGEX": "Faisal\s+?II"}}]},
{"label": "PERSON", "pattern": [{"TEXT": {"REGEX": "Milton\s+?Cooper"}}]},
{"label": "PERSON", "pattern": [{"TEXT": {"REGEX": "Mostafa\s+?A\s+?Abdelkader"}}]},
{"label": "PERSON", "pattern": [{"TEXT": {"REGEX": "Abdul\s+?Llah"}}]},
{"label": "PERSON", "pattern": [{"TEXT": {"REGEX": "Marc\s+?J\.\s+?Seifer"}}]},
{"label": "PERSON", "pattern": [{"TEXT": {"REGEX": "R\.\s+?J\.\s+?Biggar"}}]},
{"label": "PERSON", "pattern": [{"TEXT": {"REGEX": "Stanford\s+?White"}}]},
{"label": "PERSON", "pattern": [{"TEXT": {"REGEX": "John\s+?Paul\s+?I"}}]},
{"label": "PERSON", "pattern": [{"TEXT": {"REGEX": "John\s+?Paul\s+?II"}}]},
{"label": "PERSON", "pattern": [{"TEXT": {"REGEX": "Ensign\s+?Ball"}}]},
{"label": "PERSON", "pattern": [{"TEXT": {"REGEX": "O\.H\.\s+?Cril"}}]},
{"label": "PERSON", "pattern": [{"TEXT": {"REGEX": "Crill"}}]},
{"label": "PERSON", "pattern": [{"TEXT": {"REGEX": "KRLL"}}]},
{"label": "PERSON", "pattern": [{"TEXT": {"REGEX": "KRLLL"}}]},
{"label": "PERSON", "pattern": [{"TEXT": {"REGEX": "EBE"}}]},
{"label": "PERSON", "pattern": [{"TEXT": {"REGEX": "Rockefeller\s+?III"}}]},
{"label": "PERSON", "pattern": [{"TEXT": {"REGEX": "John\s+?D\.\s+?Rockefeller\s+?III"}}]},
{"label": "PERSON", "pattern": [{"TEXT": {"REGEX": "Dennis\s+?DeConcini"}}]},
{"label": "PERSON", "pattern": [{"TEXT": {"REGEX": "Phil\s+?Gramm"}}]},
{"label": "PERSON", "pattern": [{"TEXT": {"REGEX": "BILL\s+?HAMILTON"}}]},
{"label": "PERSON", "pattern": [{"TEXT": {"REGEX": "Gramm"}}]},
{"label": "PERSON", "pattern": [{"TEXT": {"REGEX": "DeConcini"}}]},
{"label": "PERSON", "pattern": [{"TEXT": {"REGEX": "Michael\s+?Emerling"}}]},
{"label": "PERSON", "pattern": [{"TEXT": {"REGEX": "Rockefeller"}}]},
{"label": "PERSON", "pattern": [{"TEXT": {"REGEX": "Johnston"}}]},
{"label": "PERSON", "pattern": [{"TEXT": {"REGEX": "Dodd"}}]},
{"label": "PERSON", "pattern": [{"TEXT": {"REGEX": "Laurence\s+Rockefeller"}}]},
{"label": "PERSON", "pattern": [{"TEXT": {"REGEX": "John\s+D\.\s+Rockefeller"}}]},
{"label": "PERSON", "pattern": [{"TEXT": {"REGEX": "John\s+D\.\s+Rockefeller\s+IV"}}]},
{"label": "PERSON", "pattern": [{"TEXT": {"REGEX": "John\s+D\.\s+Rockefeller,\s+Jr\."}}]},
{"label": "PERSON", "pattern": [{"TEXT": {"REGEX": "William\s+?H\.\s+?Draper\s+?III"}}]},
{"label": "PERSON", "pattern": [{"TEXT": {"REGEX": "Eduardo"}}]},
{"label": "PERSON", "pattern": [{"TEXT": {"REGEX": "EARL\s+?W\.\s+?BRIAN"}}]},
{"label": "PERSON", "pattern": [{"TEXT": {"REGEX": "RICONOSCIUTO"}}]},
{"label": "PERSON", "pattern": [{"TEXT": {"REGEX": "Ramakrishna"}}]},
{"label": "PERSON", "pattern": [{"TEXT": {"REGEX": "Don\s+?Ecker"}}]},
{"label": "PERSON", "pattern": [{"TEXT": {"REGEX": "John\s+?Galt"}}]},
{"label": "PERSON", "pattern": [{"TEXT": {"REGEX": "C\.D\.\s+?Jackson"}}]},
{"label": "PERSON", "pattern": [{"TEXT": {"REGEX": "G\.H\.\s+?Walker\s+?III"}}]},
{"label": "PERSON", "pattern": [{"TEXT": {"REGEX": "Daniel\s+?Gooch"}}]},
{"label": "PERSON", "pattern": [{"TEXT": {"REGEX": "Ramses\s+?II"}}]},
{"label": "PERSON", "pattern": [{"TEXT": {"REGEX": "Alexander\s+?II"}}]},
{"label": "PERSON", "pattern": [{"TEXT": {"REGEX": "Alexander\s+?V"}}]},
{"label": "PERSON", "pattern": [{"TEXT": {"REGEX": "Alexander\s+?IV"}}]},
{"label": "PERSON", "pattern": [{"TEXT": {"REGEX": "Alexander\s+?VI"}}]},
{"label": "PERSON", "pattern": [{"TEXT": {"REGEX": "Breck\s+?Wall"}}]},
{"label": "PERSON", "pattern": [{"TEXT": {"REGEX": "CARROLL\s+?QUIGLEY"}}]},
{"label": "PERSON", "pattern": [{"TEXT": {"REGEX": "Swami\s+?Vivekananda"}}]},
{"label": "PERSON", "pattern": [{"TEXT": {"REGEX": "DOOLITTLE"}}]},
{"label": "PERSON", "pattern": [{"TEXT": {"REGEX": "James\s+?I"}}]},
{"label": "PERSON", "pattern": [{"TEXT": {"REGEX": "Oswald\s+?II"}}]},
{"label": "PERSON", "pattern": [{"TEXT": {"REGEX": "Oswald\s+?I"}}]},
{"label": "PERSON", "pattern": [{"TEXT": {"REGEX": "Umberto\s+?I"}}]},
{"label": "PERSON", "pattern": [{"TEXT": {"REGEX": "Nikola\s+?Tesla"}}]},
{"label": "PERSON", "pattern": [{"TEXT": {"REGEX": "Tesla"}}]},
{"label": "PERSON", "pattern": [{"TEXT": {"REGEX": "C\.\s+?Hamilton\s+?Ellis"}}]},
{"label": "PERSON", "pattern": [{"TEXT": {"REGEX": "Alfred\s+?Ely\s+?Beach"}}]},
{"label": "PERSON", "pattern": [{"TEXT": {"REGEX": "Ed\s+?Vyrdolak"}}]},
{"label": "PERSON", "pattern": [{"TEXT": {"REGEX": "Beach"}}]},
{"label": "PERSON", "pattern": [{"TEXT": {"REGEX": "Lord\s+?Milner"}}]},
{"label": "PERSON", "pattern": [{"TEXT": {"REGEX": "Agapetus\s+?I"}}]},
{"label": "PERSON", "pattern": [{"TEXT": {"REGEX": "Agapetus\s+?II"}}]},
{"label": "PERSON", "pattern": [{"TEXT": {"REGEX": "C\.\s+?Victor\s+?Raiser\s+?II"}}]},
{"label": "PERSON", "pattern": [{"TEXT": {"REGEX": "C\.V\.\s+?Raiser\s+?II"}}]},
{"label": "PERSON", "pattern": [{"TEXT": {"REGEX": "Boniface\s+?VIII"}}]},
{"label": "PERSON", "pattern": [{"TEXT": {"REGEX": "Boniface\s+?VII"}}]},
{"label": "PERSON", "pattern": [{"TEXT": {"REGEX": "Boniface\s+?III"}}]},
{"label": "PERSON", "pattern": [{"TEXT": {"REGEX": "Boniface\s+?II"}}]},
{"label": "PERSON", "pattern": [{"TEXT": {"REGEX": "Boniface\s+?VI"}}]},
{"label": "PERSON", "pattern": [{"TEXT": {"REGEX": "Gonda"}}]},
{"label": "PERSON", "pattern": [{"TEXT": {"REGEX": "Ollie\s+?North"}}]},
{"label": "PERSON", "pattern": [{"TEXT": {"REGEX": "E\.\s+?Howard\s+?Hunt"}}]},
{"label": "PERSON", "pattern": [{"TEXT": {"REGEX": "MLK"}}]},
{"label": "PERSON", "pattern": [{"TEXT": {"REGEX": "RFK"}}]},
{"label": "PERSON", "pattern": [{"TEXT": {"REGEX": "O\.\s+?W\.\s+?Judd"}}]},
{"label": "PERSON", "pattern": [{"TEXT": {"REGEX": "C\.J\.\s+?Duffner"}}]},
{"label": "PERSON", "pattern": [{"TEXT": {"REGEX": "J\.P\.\s+?Morgan"}}]},
{"label": "PERSON", "pattern": [{"TEXT": {"REGEX": "Richard\s+?II"}}]},
{"label": "PERSON", "pattern": [{"TEXT": {"REGEX": "James\s+?McDonald"}}]},
{"label": "PERSON", "pattern": [{"TEXT": {"REGEX": "McDonald"}}]},
{"label": "PERSON", "pattern": [{"TEXT": {"REGEX": "Saunders"}}]},
{"label": "PERSON", "pattern": [{"TEXT": {"REGEX": "Richard\s+?III"}}]},
{"label": "PERSON", "pattern": [{"TEXT": {"REGEX": "George\s+?Terwilliger\s+?III"}}]},
{"label": "PERSON", "pattern": [{"TEXT": {"REGEX": "C\.E\.\s+?Koop"}}]},
{"label": "PERSON", "pattern": [{"TEXT": {"REGEX": "Callistus\s+?I"}}]},
{"label": "PERSON", "pattern": [{"TEXT": {"REGEX": "Francis\s+?von\s+?Hapsburg"}}]},
{"label": "PERSON", "pattern": [{"TEXT": {"REGEX": "Malcom\s+?X"}}]},
{"label": "PERSON", "pattern": [{"TEXT": {"REGEX": "Paramahansa\s+?Yogananda"}}]},
{"label": "PERSON", "pattern": [{"TEXT": {"REGEX": "Victor\s+?IV"}}]},
{"label": "PERSON", "pattern": [{"TEXT": {"REGEX": "Carles\s+?C\.\s+?Messick\s+?III"}}]},
{"label": "PERSON", "pattern": [{"TEXT": {"REGEX": "John\s+?D\.\s+?Rockefeller\s+?IV"}}]},
{"label": "PERSON", "pattern": [{"TEXT": {"REGEX": "MARCONI"}}]},
{"label": "PERSON", "pattern": [{"TEXT": {"REGEX": "Marconi"}}]},
{"label": "PERSON", "pattern": [{"TEXT": {"REGEX": "Bill\s+?English"}}]},
{"label": "PERSON", "pattern": [{"TEXT": {"REGEX": "Bill\s+?Cooper"}}]},
{"label": "PERSON", "pattern": [{"TEXT": {"REGEX": "Charles\s+?W\.\s+?Bailey\s+?II"}}]},
{"label": "PERSON", "pattern": [{"TEXT": {"REGEX": "Charles\s+?I"}}]},
{"label": "PERSON", "pattern": [{"TEXT": {"REGEX": "Charles\s+?McKee"}}]},
{"label": "PERSON", "pattern": [{"TEXT": {"REGEX": "Charles\s+?X"}}]},
{"label": "PERSON", "pattern": [{"TEXT": {"REGEX": "Charles\s+?V"}}]},
{"label": "PERSON", "pattern": [{"TEXT": {"REGEX": "Jerome\s+?I"}}]},
{"label": "PERSON", "pattern": [{"TEXT": {"REGEX": "Mark\s+?I"}}]},
{"label": "PERSON", "pattern": [{"TEXT": {"REGEX": "John\s+?XXIII"}}]},
{"label": "PERSON", "pattern": [{"TEXT": {"REGEX": "Ferdinand\s+?I"}}]},
{"label": "PERSON", "pattern": [{"TEXT": {"REGEX": "Hillary\s+?Rodham\s+?Clinton"}}]},
{"label": "PERSON", "pattern": [{"TEXT": {"REGEX": "Vincent\s+?Foster,\s+?jr\."}}]},
{"label": "PERSON", "pattern": [{"TEXT": {"REGEX": "Nicholas\s+?I"}}]},
{"label": "PERSON", "pattern": [{"TEXT": {"REGEX": "Nicholas\s+?II"}}]},
{"label": "PERSON", "pattern": [{"TEXT": {"REGEX": "Nicholas\s+?V"}}]},
{"label": "PERSON", "pattern": [{"TEXT": {"REGEX": "Vespasian"}}]},
{"label": "PERSON", "pattern": [{"TEXT": {"REGEX": "John\s+?Zajac"}}]},
{"label": "PERSON", "pattern": [{"TEXT": {"REGEX": "Sylvester\s+?I"}}]},
{"label": "PERSON", "pattern": [{"TEXT": {"REGEX": "Sylvester\s+?II"}}]},
{"label": "PERSON", "pattern": [{"TEXT": {"REGEX": "Engenius\s+?IV"}}]},
{"label": "PERSON", "pattern": [{"TEXT": {"REGEX": "Philip\s+?IV"}}]},
{"label": "PERSON", "pattern": [{"TEXT": {"REGEX": "Stephen\s+?II"}}]},
{"label": "PERSON", "pattern": [{"TEXT": {"REGEX": "Stephen\s+?VI"}}]},
{"label": "PERSON", "pattern": [{"TEXT": {"REGEX": "Pious\s+?X"}}]},
{"label": "PERSON", "pattern": [{"TEXT": {"REGEX": "Honorius\s+?II"}}]},
{"label": "PERSON", "pattern": [{"TEXT": {"REGEX": "Sergius\s+?III"}}]},
{"label": "PERSON", "pattern": [{"TEXT": {"REGEX": "Adrian\s+?I"}}]},
{"label": "PERSON", "pattern": [{"TEXT": {"REGEX": "Adrian\s+?VI"}}]},
{"label": "PERSON", "pattern": [{"TEXT": {"REGEX": "William\s+?F\.\s+?Hamilton\s+?III"}}]},
{"label": "PERSON", "pattern": [{"TEXT": {"REGEX": "James\s+?Baker\s+?III"}}]},
{"label": "PERSON", "pattern": [{"TEXT": {"REGEX": "William\s+?Stamps\s+?Farish\s+?III"}}]},
{"label": "PERSON", "pattern": [{"TEXT": {"REGEX": "Elizabeth\s+?II"}}]},
{"label": "PERSON", "pattern": [{"TEXT": {"REGEX": "Anacletus\s+?II"}}]},
{"label": "PERSON", "pattern": [{"TEXT": {"REGEX": "Edward\s+?II"}}]},
{"label": "PERSON", "pattern": [{"TEXT": {"REGEX": "Edward\s+?III"}}]},
{"label": "PERSON", "pattern": [{"TEXT": {"REGEX": "Edward\s+?V"}}]},
{"label": "PERSON", "pattern": [{"TEXT": {"REGEX": "Matthew\s+?XVI"}}]},
{"label": "PERSON", "pattern": [{"TEXT": {"REGEX": "Vyrdolak"}}]},
{"label": "PERSON", "pattern": [{"TEXT": {"REGEX": "Thothmes\s+?III"}}]},
{"label": "PERSON", "pattern": [{"TEXT": {"REGEX": "Harry\s+?Hurt\s+?III"}}]},
{"label": "PERSON", "pattern": [{"TEXT": {"REGEX": "Napoleon\s+?III"}}]},
{"label": "PERSON", "pattern": [{"TEXT": {"REGEX": "Clement\s+?VIII"}}]},
{"label": "PERSON", "pattern": [{"TEXT": {"REGEX": "Clement\s+?I"}}]},
{"label": "PERSON", "pattern": [{"TEXT": {"REGEX": "Clement\s+?VI"}}]},
{"label": "PERSON", "pattern": [{"TEXT": {"REGEX": "Clement\s+?IV"}}]},
{"label": "PERSON", "pattern": [{"TEXT": {"REGEX": "Pius\s+?X"}}]},
{"label": "PERSON", "pattern": [{"TEXT": {"REGEX": "Pius\s+?II"}}]},
{"label": "PERSON", "pattern": [{"TEXT": {"REGEX": "Pius\s+?IX"}}]},
{"label": "PERSON", "pattern": [{"TEXT": {"REGEX": "Pius\s+?XI"}}]},
{"label": "PERSON", "pattern": [{"TEXT": {"REGEX": "Pius\s+?XII"}}]},
{"label": "PERSON", "pattern": [{"TEXT": {"REGEX": "Leo\s+?IX"}}]},
{"label": "PERSON", "pattern": [{"TEXT": {"REGEX": "Leo\s+?X"}}]},
{"label": "PERSON", "pattern": [{"TEXT": {"REGEX": "Leo\s+?III"}}]},
{"label": "PERSON", "pattern": [{"TEXT": {"REGEX": "Leo\s+?XIII"}}]},
{"label": "PERSON", "pattern": [{"TEXT": {"REGEX": "Amenhotep\s+?IV"}}]},
{"label": "PERSON", "pattern": [{"TEXT": {"REGEX": "Charles\s+?W\.\s+?Bailey\s+?II"}}]},
{"label": "PERSON", "pattern": [{"TEXT": {"REGEX": "Constantine"}}]},
{"label": "PERSON", "pattern": [{"TEXT": {"REGEX": "Tut-ankh-amen"}}]},
{"label": "PERSON", "pattern": [{"TEXT": {"REGEX": "Sixtus\s+?V"}}]},
{"label": "PERSON", "pattern": [{"TEXT": {"REGEX": "Sixtus\s+?IV"}}]},
{"label": "PERSON", "pattern": [{"TEXT": {"REGEX": "Benedict\s+?XIV"}}]},
{"label": "PERSON", "pattern": [{"TEXT": {"REGEX": "Benedict\s+?VI"}}]},
{"label": "PERSON", "pattern": [{"TEXT": {"REGEX": "Benedict\s+?IX"}}]},
{"label": "PERSON", "pattern": [{"TEXT": {"REGEX": "Benedict\s+?XV"}}]},
{"label": "PERSON", "pattern": [{"TEXT": {"REGEX": "George\s+?I"}}]},
{"label": "PERSON", "pattern": [{"TEXT": {"REGEX": "George\s+?III"}}]},
{"label": "PERSON", "pattern": [{"TEXT": {"REGEX": "Gregory\s+?VII"}}]},
{"label": "PERSON", "pattern": [{"TEXT": {"REGEX": "Gregory\s+?I"}}]},
{"label": "PERSON", "pattern": [{"TEXT": {"REGEX": "Gregory\s+?XVI"}}]},
{"label": "PERSON", "pattern": [{"TEXT": {"REGEX": "Gregory\s+?IX"}}]},
{"label": "PERSON", "pattern": [{"TEXT": {"REGEX": "Constantine\s+?II"}}]},
{"label": "PERSON", "pattern": [{"TEXT": {"REGEX": "Albert\s+?I"}}]},
{"label": "PERSON", "pattern": [{"TEXT": {"REGEX": "Albert\s+?V\.\s+?Bryan\s+?Jr"}}]},
{"label": "PERSON", "pattern": [{"TEXT": {"REGEX": "Alfonso\s+?XII"}}]},
{"label": "PERSON", "pattern": [{"TEXT": {"REGEX": "Alfonso\s+?XIII"}}]},
{"label": "PERSON", "pattern": [{"TEXT": {"REGEX": "Gustavus\s+?III"}}]},
{"label": "PERSON", "pattern": [{"TEXT": {"REGEX": "Gustav\s+?III"}}]},
{"label": "PERSON", "pattern": [{"TEXT": {"REGEX": "Felix\s+?III"}}]},
{"label": "PERSON", "pattern": [{"TEXT": {"REGEX": "Valentinian\s+?III"}}]},
{"label": "PERSON", "pattern": [{"TEXT": {"REGEX": "Innocent\s+?III"}}]},
{"label": "PERSON", "pattern": [{"TEXT": {"REGEX": "Innocent\s+?IV"}}]},
{"label": "PERSON", "pattern": [{"TEXT": {"REGEX": "Frederick,\s+?II"}}]},
{"label": "PERSON", "pattern": [{"TEXT": {"REGEX": "Frederick\s+?II"}}]},
{"label": "PERSON", "pattern": [{"TEXT": {"REGEX": "Theodosius\s+?I"}}]},
{"label": "PERSON", "pattern": [{"TEXT": {"REGEX": "Henry\s+?VIII"}}]},
{"label": "PERSON", "pattern": [{"TEXT": {"REGEX": "Henry\s+?VI"}}]},
{"label": "PERSON", "pattern": [{"TEXT": {"REGEX": "Henry\s+?IV"}}]},
{"label": "PERSON", "pattern": [{"TEXT": {"REGEX": "Louis\s+?IX"}}]},
{"label": "PERSON", "pattern": [{"TEXT": {"REGEX": "Louis\s+?XVI"}}]},
{"label": "PERSON", "pattern": [{"TEXT": {"REGEX": "Joseph\s+?II"}}]},
{"label": "PERSON", "pattern": [{"TEXT": {"REGEX": "Catherine\s+?II"}}]},
{"label": "PERSON", "pattern": [{"TEXT": {"REGEX": "James\s+?D\s+?Bryant\s+?II"}}]},
{"label": "PERSON", "pattern": [{"TEXT": {"REGEX": "Paul\s+?III"}}]},
{"label": "PERSON", "pattern": [{"TEXT": {"REGEX": "Paul\s+?II"}}]},
{"label": "PERSON", "pattern": [{"TEXT": {"REGEX": "Julius\s+?II"}}]},
{"label": "PERSON", "pattern": [{"TEXT": {"REGEX": "Paul\s+?VI"}}]},
{"label": "PERSON", "pattern": [{"TEXT": {"REGEX": "William\s+?II"}}]},
{"label": "PERSON", "pattern": [{"TEXT": {"REGEX": "William\s+?III"}}]},
{"label": "PERSON", "pattern": [{"TEXT": {"REGEX": "Baldwin\s+?II"}}]},
{"label": "PERSON", "pattern": [{"TEXT": {"REGEX": "Urban\s+?VIII"}}]},
{"label": "PERSON", "pattern": [{"TEXT": {"REGEX": "Charles\s+?Goodhue,\s+?III"}}]},
{"label": "PERSON", "pattern": [{"TEXT": {"REGEX": "Edwin\s+?Meese,\s+?III"}}]},
{"label": "PERSON", "pattern": [{"TEXT": {"REGEX": "Thomas\s+?Downing"}}]},
{"label": "PERSON", "pattern": [{"TEXT": {"REGEX": "Thomas\s+?E\.\s+?Davis"}}]},
{"label": "PERSON", "pattern": [{"TEXT": {"REGEX": "J\.A\.\s+?Milteer"}}]},
{"label": "PERSON", "pattern": [{"TEXT": {"REGEX": "Charles\s+?Cabell"}}]},
{'label': 'PERSON', 'pattern': [{'TEXT': {'REGEX': 'James\s+?"Bo"\s+?Gritz'}}]},
{'label': 'PERSON', 'pattern': [{'TEXT': {'REGEX': 'Mike\s+?"Chucky"\s+?Peters'}}]},
{"label": "PERSON", "pattern": [{"TEXT": {"REGEX": "James\s+?Earl\s+?Ray"}}]},
{"label": "PERSON", "pattern": [{"TEXT": {"REGEX": "Earl\s+?Wheeler"}}]},
{"label": "PERSON", "pattern": [{"TEXT": {"REGEX": "William\s+?S\.\s+?Thompson"}}]},
{"label": "PERSON", "pattern": [{"TEXT": {"REGEX": "S\.\s+?Rilling"}}]},
{"label": "PERSON", "pattern": [{"TEXT": {"REGEX": "Ralph\s+?Paul"}}]},
{"label": "PERSON", "pattern": [{"TEXT": {"REGEX": "Scott\s+?Weekly"}}]},
{"label": "PERSON", "pattern": [{"TEXT": {"REGEX": "Ford,\s+?L\.H"}}]},
{"label": "PERSON", "pattern": [{"TEXT": {"REGEX": "A\.\s+?Vilenkin"}}]},
{"label": "PERSON", "pattern": [{"TEXT": {"REGEX": "Renate\s+?Viebahn"}}]},
{"label": "PERSON", "pattern": [{"TEXT": {"REGEX": "Lustick,\s+?Ian\s+?S\."}}]},
{"label": "PERSON", "pattern": [{"TEXT": {"REGEX": "Friedman,\s+?Robert\s+?I\."}}]},
{"label": "PERSON", "pattern": [{"TEXT": {"REGEX": "Howard\s+?Sprague"}}]},
{"label": "PERSON", "pattern": [{"TEXT": {"REGEX": "Mark\s+?Clark"}}]},
{"label": "PERSON", "pattern": [{"TEXT": {"REGEX": "Owen"}}]},
{"label": "PERSON", "pattern": [{"TEXT": {"REGEX": "Theodore\s+?Shackley"}}]},
{"label": "PERSON", "pattern": [{"TEXT": {"REGEX": "DeLesseps\s+?Morrison"}}]},
{"label": "PERSON", "pattern": [{"TEXT": {"REGEX": "Betty\s+?McDonald"}}]},
{"label": "PERSON", "pattern": [{"TEXT": {"REGEX": "Clines"}}]},
{"label": "PERSON", "pattern": [{"TEXT": {"REGEX": "L\.\s+?F\.\s+?Prouty"}}]},
{"label": "PERSON", "pattern": [{"TEXT": {"REGEX": "L\.\s+?Fletcher\s+?Prouty"}}]},
{"label": "PERSON", "pattern": [{"TEXT": {"REGEX": "Knight,\s+?Amy\s+?W\."}}]},
{"label": "PERSON", "pattern": [{"TEXT": {"REGEX": "Martindale"}}]},
{"label": "PERSON", "pattern": [{"TEXT": {"REGEX": "R\.\s+?Knox"}}]},
{"label": "PERSON", "pattern": [{"TEXT": {"REGEX": "Leon\s+?Oswald"}}]},
{"label": "PERSON", "pattern": [{"TEXT": {"REGEX": "H\.\s+?Ross\s+?Perot"}}]},
{"label": "PERSON", "pattern": [{"TEXT": {"REGEX": "Stanly\s+?R\.\s+?Larsen"}}]},
{"label": "PERSON", "pattern": [{"TEXT": {"REGEX": "L\.E\.\s+?Allen\s+?Jr\."}}]},
{"label": "PERSON", "pattern": [{"TEXT": {"REGEX": "A\.\s+?Lewis"}}]},
{"label": "PERSON", "pattern": [{"TEXT": {"REGEX": "D\.\s+?Fleming"}}]},
{"label": "PERSON", "pattern": [{"TEXT": {"REGEX": "J\.\s+?Edgar\s+?Hoover"}}]},
{"label": "PERSON", "pattern": [{"TEXT": {"REGEX": "V\.\s+?Pirie"}}]},
{"label": "PERSON", "pattern": [{"TEXT": {"REGEX": "Nancy\s+?B\."}}]},
{"label": "PERSON", "pattern": [{"TEXT": {"REGEX": "Ronald\s+?Reagan"}}]},
{"label": "PERSON", "pattern": [{"TEXT": {"REGEX": "P\.\s+?Gibbs"}}]},
{"label": "PERSON", "pattern": [{"TEXT": {"REGEX": "W\.\s+?Citrine"}}]},
{"label": "PERSON", "pattern": [{"TEXT": {"REGEX": "E\.\s+?Moore"}}]},
{"label": "PERSON", "pattern": [{"TEXT": {"REGEX": "Lyndon\s+?Johnson"}}]},
repairing and restoring XSLT, spot fixes to personTagger.py
2023-04-28 01:26:51 -04:00
{"label": "PERSON", "pattern": [{"TEXT": {"REGEX": "John(\s*F\.?)\s*Kennedy"}}]},
trying my best to fix the regex and editing python code
2023-04-27 00:54:09 -04:00
{"label": "PERSON", "pattern": [{"TEXT": {"REGEX": "Zia\s+?ul-Haque"}}]},
repairing and restoring XSLT, spot fixes to personTagger.py
2023-04-28 01:26:51 -04:00
{"label": "PERSON", "pattern": [{"TEXT": {"REGEX": "Ronald\s*Payne"}}]},
{"label": "PERSON", "pattern": [{"TEXT": {"REGEX": "R\.\s*Muldoon"}}]},
{"label": "PERSON", "pattern": [{"TEXT": {"REGEX": "F\.\s*Orr"}}]},
trying my best to fix the regex and editing python code
2023-04-27 00:54:09 -04:00
{"label": "PERSON", "pattern": [{"TEXT": {"REGEX": "Frank\s+?H\.\s+?Schwable"}}]},
{"label": "PERSON", "pattern": [{"TEXT": {"REGEX": "R\.\s+?F\.\s+?Doyle"}}]},
{"label": "PERSON", "pattern": [{"TEXT": {"REGEX": "David\s+?Munson"}}]},
{"label": "PERSON", "pattern": [{"TEXT": {"REGEX": "Frances\s+?Clark"}}]},
{"label": "PERSON", "pattern": [{"TEXT": {"REGEX": "G\.\s+?Gordon\s+?Broadbent"}}]},
{"label": "PERSON", "pattern": [{"TEXT": {"REGEX": "J\.\s+?Segal"}}]},
{"label": "PERSON", "pattern": [{"TEXT": {"REGEX": "F\.\s+?Brangwyn"}}]},
{"label": "PERSON", "pattern": [{"TEXT": {"REGEX": "Luc\s+?Montagnier"}}]},
{"label": "PERSON", "pattern": [{"TEXT": {"REGEX": "A\.\s+?Luchaire"}}]},
{"label": "PERSON", "pattern": [{"TEXT": {"REGEX": "L\.\s+?Segal"}}]},
{"label": "PERSON", "pattern": [{"TEXT": {"REGEX": "R\.\s+?William\s+?Davis"}}]},
{"label": "PERSON", "pattern": [{"TEXT": {"REGEX": "William\s+?C\.\s+?Sullivan"}}]},
{"label": "PERSON", "pattern": [{"TEXT": {"REGEX": "William\s+?W\.\s+?Quinn"}}]},
{"label": "PERSON", "pattern": [{"TEXT": {"REGEX": "William\s+?R\.\s+?Pabst"}}]},
{"label": "PERSON", "pattern": [{"TEXT": {"REGEX": "Oliver\s+?Nichelson"}}]},
{"label": "PERSON", "pattern": [{"TEXT": {"REGEX": "James\s+?Worrell,\s+?Jr\."}}]},
{"label": "PERSON", "pattern": [{"TEXT": {"REGEX": "Hank\s+?Suydam"}}]},
{"label": "PERSON", "pattern": [{"TEXT": {"REGEX": "William\s+?Pabst"}}]},
{"label": "PERSON", "pattern": [{"TEXT": {"REGEX": "G\.\s+?Maeferren"}}]},
{"label": "PERSON", "pattern": [{"TEXT": {"REGEX": "Dian-Lanz"}}]},
{"label": "PERSON", "pattern": [{"TEXT": {"REGEX": "Buria"}}]},
{"label": "PERSON", "pattern": [{"TEXT": {"REGEX": "North"}}]},
{"label": "PERSON", "pattern": [{"TEXT": {"REGEX": "Ortho\s+?III"}}]},
{"label": "PERSON", "pattern": [{"TEXT": {"REGEX": "Swami\s+?Nikhilananda"}}]},
{"label": "PERSON", "pattern": [{"TEXT": {"REGEX": "Santo\s+?Trafficante"}}]},
{"label": "PERSON", "pattern": [{"TEXT": {"REGEX": "Felix\s+?Rodreguez"}}]},
{"label": "PERSON", "pattern": [{"TEXT": {"REGEX": "Stanly\s+?F\.\s+?Yoles"}}]},
{"label": "PERSON", "pattern": [{"TEXT": {"REGEX": "G\.\s+?B\.\s+?Chisholm"}}]},
{"label": "PERSON", "pattern": [{"TEXT": {"REGEX": "Baron\s+?Kurt\s+?von\s+?Schroeder"}}]},
{"label": "PERSON", "pattern": [{"TEXT": {"REGEX": "George\s+?Bush"}}]},
{"label": "PERSON", "pattern": [{"TEXT": {"REGEX": "Roland\s+?Harriman"}}]},
{"label": "PERSON", "pattern": [{"TEXT": {"REGEX": "John\s+?Foster\s+?Dulles"}}]},
{"label": "PERSON", "pattern": [{"TEXT": {"REGEX": "Sherman\s+?Skolnick"}}]},
{"label": "PERSON", "pattern": [{"TEXT": {"REGEX": "Francis\s+?G\.\s+?Powers"}}]},
{"label": "PERSON", "pattern": [{"TEXT": {"REGEX": "Kenneth\s+?O'Donnell"}}]},
{"label": "PERSON", "pattern": [{"TEXT": {"REGEX": "J\.M\.\s+?English"}}]},
{"label": "PERSON", "pattern": [{"TEXT": {"REGEX": "William\s+?Sullivan"}}]},
{'label': 'PERSON', 'pattern': [{'TEXT': {'REGEX': 'C\.L\. "Lummie" Lewis'}}]},
{"label": "PERSON", "pattern": [{"TEXT": {"REGEX": "Garland\s+?Slack"}}]},
{"label": "PERSON", "pattern": [{"TEXT": {"REGEX": "Billy\s+?Lovelady"}}]},
{"label": "PERSON", "pattern": [{"TEXT": {"REGEX": "John\s+?Holbrook"}}]},
{"label": "PERSON", "pattern": [{"TEXT": {"REGEX": "Marguerite\s+?Oswald"}}]},
{"label": "PERSON", "pattern": [{"TEXT": {"REGEX": "Andrew\s+?Cetti"}}]},
{"label": "PERSON", "pattern": [{"TEXT": {"REGEX": "Will\s+?H\.\s+?Griffin"}}]},
{"label": "PERSON", "pattern": [{"TEXT": {"REGEX": "Peter\s+?Gregory"}}]},
{"label": "PERSON", "pattern": [{"TEXT": {"REGEX": "Harry\s+?Grindell-Matthews"}}]},
{"label": "PERSON", "pattern": [{"TEXT": {"REGEX": "Grindell-Matthews"}}]},
{"label": "PERSON", "pattern": [{"TEXT": {"REGEX": "Rosenholtz"}}]},
{"label": "PERSON", "pattern": [{"TEXT": {"REGEX": "Karen\s+?Silkwood"}}]},
{"label": "PERSON", "pattern": [{"TEXT": {"REGEX": "N\.\s+?McQuire"}}]},
{"label": "PERSON", "pattern": [{"TEXT": {"REGEX": "William\s+?Highland"}}]},
{"label": "PERSON", "pattern": [{"TEXT": {"REGEX": "Harriman"}}]},
{"label": "PERSON", "pattern": [{"TEXT": {"REGEX": "William\s+?Saxby"}}]},
{"label": "PERSON", "pattern": [{"TEXT": {"REGEX": "Ellis\s+?O\.\s+?Briggs"}}]},
{"label": "PERSON", "pattern": [{"TEXT": {"REGEX": "Daniel\s+?Ludwig"}}]},
{"label": "PERSON", "pattern": [{"TEXT": {"REGEX": "Robert\s+?C\.\s+?Klowers"}}]},
{"label": "PERSON", "pattern": [{"TEXT": {"REGEX": "Tatum\s+?B\.\s+?Laird"}}]},
{"label": "PERSON", "pattern": [{"TEXT": {"REGEX": "William\s+?F\.\s+?Buckley,\s+?Jr\."}}]},
{"label": "PERSON", "pattern": [{"TEXT": {"REGEX": "William\s+?Randolph\s+?Hearst,\s+?Jr\."}}]},
{"label": "PERSON", "pattern": [{"TEXT": {"REGEX": "Justin\s+?Dart"}}]},
{"label": "PERSON", "pattern": [{"TEXT": {"REGEX": "Larry\s+?McDonald"}}]},
{"label": "PERSON", "pattern": [{"TEXT": {"REGEX": "Mac\s+?MacPherson"}}]},
{"label": "PERSON", "pattern": [{"TEXT": {"REGEX": "MacPherson"}}]},
{"label": "PERSON", "pattern": [{"TEXT": {"REGEX": "Callen"}}]},
{"label": "PERSON", "pattern": [{"TEXT": {"REGEX": "Lievense"}}]},
{"label": "PERSON", "pattern": [{"TEXT": {"REGEX": "CASTRO"}}]},
{"label": "PERSON", "pattern": [{"TEXT": {"REGEX": "Erik\s+?Jonsson"}}]},
{"label": "PERSON", "pattern": [{"TEXT": {"REGEX": "Charles\s+?E\.\s+?Allen"}}]},
{"label": "PERSON", "pattern": [{"TEXT": {"REGEX": "Laurence\s+?H\.\s+?Shoup"}}]},
{"label": "PERSON", "pattern": [{"TEXT": {"REGEX": "David\s+?Rieff"}}]},
{"label": "PERSON", "pattern": [{"TEXT": {"REGEX": "David\s+?Rockefeller"}}]},
{"label": "PERSON", "pattern": [{"TEXT": {"REGEX": "Diamond"}}]},
{"label": "PERSON", "pattern": [{"TEXT": {"REGEX": "Donald\s+?DeFreeze"}}]},
{"label": "PERSON", "pattern": [{"TEXT": {"REGEX": "McGovern"}}]},
{"label": "PERSON", "pattern": [{"TEXT": {"REGEX": "McCarthy"}}]},
{"label": "PERSON", "pattern": [{"TEXT": {"REGEX": "John\s+?DiNardo"}}]},
{"label": "PERSON", "pattern": [{"TEXT": {"REGEX": "Gyeorgos\s+?C\.\s+?Hatonn"}}]},
{"label": "PERSON", "pattern": [{"TEXT": {"REGEX": "John\s+?Connally"}}]},
{"label": "PERSON", "pattern": [{"TEXT": {"REGEX": "Lyndon\s+?LaRouche"}}]},
{"label": "PERSON", "pattern": [{"TEXT": {"REGEX": "LaRouche"}}]},
{"label": "PERSON", "pattern": [{"TEXT": {"REGEX": "W\.H\.\s+?Bowart"}}]},
{"label": "PERSON", "pattern": [{"TEXT": {"REGEX": "Alan\s+?H\.\s+?Belmont"}}]},
{"label": "PERSON", "pattern": [{"TEXT": {"REGEX": "Alan\s+?Belmont"}}]},
{"label": "PERSON", "pattern": [{"TEXT": {"REGEX": "Joseph\s+?C\.\s+?Ayres"}}]},
{"label": "PERSON", "pattern": [{"TEXT": {"REGEX": "Lehrman"}}]},
{"label": "PERSON", "pattern": [{"TEXT": {"REGEX": "John\s+?McCone"}}]},
{"label": "PERSON", "pattern": [{"TEXT": {"REGEX": "Al\s+?Haig"}}]},
{"label": "PERSON", "pattern": [{"TEXT": {"REGEX": "J\.\s+?Latimer\s+?Clark"}}]},
{"label": "PERSON", "pattern": [{"TEXT": {"REGEX": "T\.\s+?W\.\s+?Rammell"}}]},
{"label": "PERSON", "pattern": [{"TEXT": {"REGEX": "Elizabeth\s+?II"}}]},
{"label": "PERSON", "pattern": [{"TEXT": {"REGEX": "Martha\s+?Honey"}}]},
{"label": "PERSON", "pattern": [{"TEXT": {"REGEX": "Vince\s+?Bielski"}}]},
{"label": "PERSON", "pattern": [{"TEXT": {"REGEX": "Anton\s+?Chaitkin"}}]},
{"label": "PERSON", "pattern": [{"TEXT": {"REGEX": "Dean\s+?Burch"}}]},
{"label": "PERSON", "pattern": [{"TEXT": {"REGEX": "Burch"}}]},
{"label": "PERSON", "pattern": [{"TEXT": {"REGEX": "Gaius\s+?Caesar"}}]},
{"label": "PERSON", "pattern": [{"TEXT": {"REGEX": "Gaius\s+?Chaerea"}}]},
{"label": "PERSON", "pattern": [{"TEXT": {"REGEX": "McCone"}}]},
{"label": "PERSON", "pattern": [{"TEXT": {"REGEX": "Richard\s+?Nixon"}}]},
{"label": "PERSON", "pattern": [{"TEXT": {"REGEX": "David\s+?Wise"}}]},
{"label": "PERSON", "pattern": [{"TEXT": {"REGEX": "Zapruder"}}]},
{"label": "PERSON", "pattern": [{"TEXT": {"REGEX": "Mortimer\s+?J\.\s+?Adler"}}]},
{"label": "PERSON", "pattern": [{"TEXT": {"REGEX": "Angela\s+?Davis"}}]},
{"label": "PERSON", "pattern": [{"TEXT": {"REGEX": "Segals"}}]},
{"label": "PERSON", "pattern": [{"TEXT": {"REGEX": "COLIN\s+?WALLACE"}}]},
{"label": "PERSON", "pattern": [{"TEXT": {"REGEX": "Thomas\s+?B\.\s+?Ross"}}]},
{"label": "PERSON", "pattern": [{"TEXT": {"REGEX": "JACOB\s+?G\.\s+?HORNBERGER"}}]},
{"label": "PERSON", "pattern": [{"TEXT": {"REGEX": "Bowart"}}]},
{"label": "PERSON", "pattern": [{"TEXT": {"REGEX": "Frank\s+?H\.\s+?Schwable"}}]},
{"label": "PERSON", "pattern": [{"TEXT": {"REGEX": "Billy\s+?Goodman"}}]},
{"label": "PERSON", "pattern": [{"TEXT": {"REGEX": "Leonard\s+?Pullin"}}]},
{"label": "PERSON", "pattern": [{"TEXT": {"REGEX": "R\.\s+?J\.\s+?Biggar"}}]},
{"label": "PERSON", "pattern": [{"TEXT": {"REGEX": "David\s+?R\.\s+?Hunter"}}]},
{"label": "PERSON", "pattern": [{"TEXT": {"REGEX": "Hugh\s+?Everett,\s+?III"}}]},
{"label": "PERSON", "pattern": [{"TEXT": {"REGEX": "Sigmund\s+?Diamond"}}]},
{"label": "PERSON", "pattern": [{"TEXT": {"REGEX": "J\.\s+?Pierpont\s+?Morgan"}}]},
{"label": "PERSON", "pattern": [{"TEXT": {"REGEX": "Montagnier"}}]},
{"label": "PERSON", "pattern": [{"TEXT": {"REGEX": "W\.\s+?Schmunger"}}]},
{"label": "PERSON", "pattern": [{"TEXT": {"REGEX": "Mary\s+?Pat\s+?Flaherty"}}]},
{"label": "PERSON", "pattern": [{"TEXT": {"REGEX": "Vince\s+?Bielski"}}]},
{"label": "PERSON", "pattern": [{"TEXT": {"REGEX": "Lee\s+?Harvey\s+?Oswald"}}]},
{"label": "PERSON", "pattern": [{"TEXT": {"REGEX": "Schmunger"}}]},
{"label": "PERSON", "pattern": [{"TEXT": {"REGEX": "Stranglove"}}]},
{"label": "PERSON", "pattern": [{"TEXT": {"REGEX": "James\s+?Calcutt"}}]},
{"label": "PERSON", "pattern": [{"TEXT": {"REGEX": "Florentine\s+?Giovanni\s+?Boccaccio"}}]},
{"label": "PERSON", "pattern": [{"TEXT": {"REGEX": "Prescott\s+?Bush"}}]},
{"label": "PERSON", "pattern": [{"TEXT": {"REGEX": "Maxwell"}}]},
{"label": "PERSON", "pattern": [{"TEXT": {"REGEX": "Daniel\s+?Schorr"}}]},
{"label": "PERSON", "pattern": [{"TEXT": {"REGEX": "Allard\s+?K\.\s+?Lowenstein"}}]},
{"label": "PERSON", "pattern": [{"TEXT": {"REGEX": "Daniel\s+?Patrick\s+?Moynihan"}}]},
{"label": "ORG", "pattern": [{"TEXT": {"REGEX": "([Tt]he\s+?)?National\s+?Institute\s+?of\s+?Health"}}]},
{"label": "ORG", "pattern": [{"TEXT": {"REGEX": "([Tt]he\s+?)?Department\s+?of\s+?Justice"}}]},
{"label": "ORG", "pattern": [{"TEXT": {"REGEX": "([Tt]he\s+?)?Yakuza"}}]},
{"label": "ORG", "pattern": [{"TEXT": {"REGEX": "YAKUZA"}}]},
{"label": "ORG", "pattern": [{"TEXT": {"REGEX": "([Tt]he\s+?)?Citizens'\s+?Committee\s+?to\s+?Clean\s+?Up\s+?the\s+?Courts"}}]},
{"label": "ORG", "pattern": [{"TEXT": {"REGEX": "([Tt]he\s+?)?Bank\s+?of\s+?Criminals\s+?and\s+?Conspirators\s+?International"}}]},
{"label": "ORG", "pattern": [{"TEXT": {"REGEX": "([Tt]he\s+?)?U\.S\.\s+?Department\s+?of\s+?Justice"}}]},
{"label": "ORG", "pattern": [{"TEXT": {"REGEX": "([Tt]he\s+?)?U\.S\.\s+?Justice\s+?Department"}}]},
{"label": "ORG", "pattern": [{"TEXT": {"REGEX": "([Tt]he\s+?)?FBI\s+?Field\s+?Office"}}]},
{"label": "ORG", "pattern": [{"TEXT": {"REGEX": "([Tt]he\s+?)?Bureau\s+?of\s+?Investigation"}}]},
{"label": "ORG", "pattern": [{"TEXT": {"REGEX": "([Tt]he\s+?)?Dallas\s+?Citizens\s+?Council"}}]},
{"label": "ORG", "pattern": [{"TEXT": {"REGEX": "([Tt]he\s+?)?University\s+?of\s+?Maryland,"}}]},
{"label": "ORG", "pattern": [{"TEXT": {"REGEX": "([Tt]he\s+?)?Carnegie\s+?Corporation"}}]},
{"label": "ORG", "pattern": [{"TEXT": {"REGEX": "([Tt]he\s+?)?Katzenbach\s+?Committee"}}]},
{"label": "ORG", "pattern": [{"TEXT": {"REGEX": "([Tt]he\s+?)?Democratic\s+?Congress"}}]},
{"label": "ORG", "pattern": [{"TEXT": {"REGEX": "([Tt]he\s+?)?National\s+?Endowment\s+?for\s+?Democracy"}}]},
{"label": "ORG", "pattern": [{"TEXT": {"REGEX": "([Tt]he\s+?)?Congress\s+?for\s+?Cultural\s+?Freedom"}}]},
{"label": "ORG", "pattern": [{"TEXT": {"REGEX": "([Tt]he\s+?)?Cambridge\s+?University\s+?Press"}}]},
{"label": "ORG", "pattern": [{"TEXT": {"REGEX": "([Tt]he\s+?)?Association\s+?of\s+?National\s+?Security\s+?Alumni"}}]},
{"label": "ORG", "pattern": [{"TEXT": {"REGEX": "([Tt]he\s+?)?Avon\s+?Books"}}]},
{"label": "ORG", "pattern": [{"TEXT": {"REGEX": "([Tt]he\s+?)?American\s+?Psychiatric\s+?Association"}}]},
{"label": "ORG", "pattern": [{"TEXT": {"REGEX": "([Tt]he\s+?)?Ministry\s+?of\s+?Defence"}}]},
{"label": "ORG", "pattern": [{"TEXT": {"REGEX": "([Tt]he\s+?)?Ramparts\s+?Press"}}]},
{"label": "ORG", "pattern": [{"TEXT": {"REGEX": "([Tt]he\s+?)?Grove\s+?Press"}}]},
{"label": "ORG", "pattern": [{"TEXT": {"REGEX": "([Tt]he\s+?)?National\s+?Association\s+?of\s+?Scholars"}}]},
{"label": "ORG", "pattern": [{"TEXT": {"REGEX": "([Tt]he\s+?)?Council\s+?on\s+?Foreign\s+?Relations"}}]},
{"label": "ORG", "pattern": [{"TEXT": {"REGEX": "([Tt]he\s+?)?Symbionese\s+?Liberation\s+?Army"}}]},
{"label": "ORG", "pattern": [{"TEXT": {"REGEX": "([Tt]he\s+?)?Washington\s+?Times"}}]},
{"label": "ORG", "pattern": [{"TEXT": {"REGEX": "([Tt]he\s+?)?Internal\s+?Revenue\s+?Service"}}]},
{"label": "ORG", "pattern": [{"TEXT": {"REGEX": "([Tt]he\s+?)?Terminate\s+?With\s+?Extreme\s+?Prejudice"}}]},
{"label": "ORG", "pattern": [{"TEXT": {"REGEX": "([Tt]he\s+?)?Center\s+?for\s+?Strategic\s+?and\s+?International\s+?Studies"}}]},
{"label": "ORG", "pattern": [{"TEXT": {"REGEX": "([Tt]he\s+?)?School\s+?of\s+?Foreign\s+?Service"}}]},
{"label": "ORG", "pattern": [{"TEXT": {"REGEX": "([Tt]he\s+?)?The\s+?Anglo-American\s+?Establishment"}}]},
{"label": "ORG", "pattern": [{"TEXT": {"REGEX": "([Tt]he\s+?)?Centre\s+?of\s+?Eternity"}}]},
{"label": "ORG", "pattern": [{"TEXT": {"REGEX": "([Tt]he\s+?)?Life\s+?magazine"}}]},
{"label": "ORG", "pattern": [{"TEXT": {"REGEX": "([Tt]he\s+?)?Full\s+?Disclosure\s+?Newspaper"}}]},
{"label": "ORG", "pattern": [{"TEXT": {"REGEX": "([Tt]he\s+?)?Trilateral\s+?Commision"}}]},
{"label": "ORG", "pattern": [{"TEXT": {"REGEX": "([Tt]he\s+?)?Ruling\s+?Elite"}}]},
{"label": "ORG", "pattern": [{"TEXT": {"REGEX": "([Tt]he\s+?)?US\s+?Congress"}}]},
{"label": "ORG", "pattern": [{"TEXT": {"REGEX": "([Tt]he\s+?)?Church\s+?Committee"}}]},
{"label": "ORG", "pattern": [{"TEXT": {"REGEX": "([Tt]he\s+?)?National\s+?Academy\s+?of\s+?Sciences"}}]},
{"label": "ORG", "pattern": [{"TEXT": {"REGEX": "([Tt]he\s+?)?Ramparts"}}]},
{"label": "NULL", "pattern": [{"TEXT": {"REGEX": "Bab"}}]},
{"label": "NULL", "pattern": [{"TEXT": {"REGEX": "Century"}}]},
{"label": "NULL", "pattern": [{"TEXT": {"REGEX": "The\s+?Problem"}}]},
{"label": "LOC", "pattern": [{"TEXT": {"REGEX": "Air\s+?Force\s+?One"}}]},
{"label": "LOC", "pattern": [{"TEXT": {"REGEX": "Kingdom\s+?of\s+?God"}}]},
{"label": "LOC", "pattern": [{"TEXT": {"REGEX": "Grassy\s+?Knoll"}}]},
{"label": "LOC", "pattern": [{"TEXT": {"REGEX": "Frankfurt\s+?airport"}}]},
{"label": "LOC", "pattern": [{"TEXT": {"REGEX": "Dealey\s+?Plaza"}}]},
{"label": "LOC", "pattern": [{"TEXT": {"REGEX": "Central\s+?Africa"}}]},
{"label": "LOC", "pattern": [{"TEXT": {"REGEX": "Fort\s+?Detrick"}}]},
{"label": "LOC", "pattern": [{"TEXT": {"REGEX": "Area-51"}}]},
{"label": "LOC", "pattern": [{"TEXT": {"REGEX": "Clifton\s+?suspension\s+?bridge"}}]},
{"label": "NULL", "pattern": [{"TEXT": {"REGEX": "mandkind"}}]},
{"label": "NULL", "pattern": [{"TEXT": {"REGEX": "Motor"}}]},
{"label": "MISC", "pattern": [{"TEXT": {"REGEX": "AIDS\s+?virus"}}]},
{"label": "MISC", "pattern": [{"TEXT": {"REGEX": "AIDS"}}]},
{"label": "MISC", "pattern": [{"TEXT": {"REGEX": "AIDS-1"}}]},
{"label": "MISC", "pattern": [{"TEXT": {"REGEX": "AIDS\s+?VIRUS"}}]},
{"label": "MISC", "pattern": [{"TEXT": {"REGEX": "Human\s+?Adult\s+?Leukemia\s+?virus"}}]},
{"label": "MISC", "pattern": [{"TEXT": {"REGEX": "LAV"}}]},
{"label": "MISC", "pattern": [{"TEXT": {"REGEX": "Malaria"}}]},
{"label": "MISC", "pattern": [{"TEXT": {"REGEX": "ARC"}}]},
{"label": "MISC", "pattern": [{"TEXT": {"REGEX": "HTLV-1"}}]},
{"label": "MISC", "pattern": [{"TEXT": {"REGEX": "HTLV-III"}}]},
{"label": "MISC", "pattern": [{"TEXT": {"REGEX": "HTLV"}}]},
{"label": "MISC", "pattern": [{"TEXT": {"REGEX": "HIV"}}]},
{"label": "MISC", "pattern": [{"TEXT": {"REGEX": "HIV\s+?virus"}}]},
{"label": "MISC", "pattern": [{"TEXT": {"REGEX": "Sheep\s+?Visna\s+?Virus"}}]},
{"label": "MISC", "pattern": [{"TEXT": {"REGEX": "Pneumocystis\s+?carinii\s+?pneumonia"}}]},
{"label": "MISC", "pattern": [{"TEXT": {"REGEX": "Kaposi's\s+?sarcoma"}}]},
{"label": "MISC", "pattern": [{"TEXT": {"REGEX": "Smallpox"}}]},
{"label": "MISC", "pattern": [{"TEXT": {"REGEX": "Polio"}}]},
{"label": "MISC", "pattern": [{"TEXT": {"REGEX": "Gunshot"}}]},
{"label": "MISC", "pattern": [{"TEXT": {"REGEX": "LSD"}}]},
{"label": "MISC", "pattern": [{"TEXT": {"REGEX": "Poison"}}]},
{"label": "MISC", "pattern": [{"TEXT": {"REGEX": "MHz"}}]},
{"label": "MISC", "pattern": [{"TEXT": {"REGEX": "TNT"}}]},
{"label": "MISC", "pattern": [{"TEXT": {"REGEX": "Acquired\s+?Immuno-Deficiency\s+?Syndrome"}}]},
{"label": "MISC", "pattern": [{"TEXT": {"REGEX": "Equal\s+?Opportunity\s+?Employer"}}]},
{"label": "MISC", "pattern": [{"TEXT": {"REGEX": "Visna"}}]},
{"label": "MISC", "pattern": [{"TEXT": {"REGEX": "visna"}}]},
{"label": "MISC", "pattern": [{"TEXT": {"REGEX": "visna\s+?virus"}}]},
{"label": "EVENT", "pattern": [{"TEXT": {"REGEX": "Korean\s+?War"}}]},
{"label": "EVENT", "pattern": [{"TEXT": {"REGEX": "National\s+?Jury\s+?Rights\s+?Day"}}]},
{"label": "EVENT", "pattern": [{"TEXT": {"REGEX": "Miranda\s+?Case"}}]},
{"label": "EVENT", "pattern": [{"TEXT": {"REGEX": "World\s+?War\s+?II"}}]},
{"label": "EVENT", "pattern": [{"TEXT": {"REGEX": "Second\s+?World\s+?War"}}]},
{"label": "EVENT", "pattern": [{"TEXT": {"REGEX": "KRISTALLNACHT"}}]},
{"label": "EVENT", "pattern": [{"TEXT": {"REGEX": "First\s+?World\s+?War"}}]},
{"label": "EVENT", "pattern": [{"TEXT": {"REGEX": "World\s+?War\s+?I"}}]},
{"label": "EVENT", "pattern": [{"TEXT": {"REGEX": "Parliament\s+?of\s+?Religions"}}]},
{"label": "EVENT", "pattern": [{"TEXT": {"REGEX": "JFK\s+?assassination"}}]},
{"label": "EVENT", "pattern": [{"TEXT": {"REGEX": "Apollo\s+?14\s+?mission"}}]},
{"label": "EVENT", "pattern": [{"TEXT": {"REGEX": "Apollo\s+?11"}}]},
{"label": "EVENT", "pattern": [{"TEXT": {"REGEX": "([Tt]he\s+?)?Andrew\s+?Cetti\s+?Affair"}}]},
{"label": "EVENT", "pattern": [{"TEXT": {"REGEX": "Railway\s+?Mania"}}]},
{"label": "EVENT", "pattern": [{"TEXT": {"REGEX": "EVENT"}}]},
{"label": "EVENT", "pattern": [{"TEXT": {"REGEX": "CHRISTIAN ERA"}}]},
{"label": "EVENT", "pattern": [{"TEXT": {"REGEX": "ante-Niccne"}}]},
{"label": "EVENT", "pattern": [{"TEXT": {"REGEX": "Vietnam\s+?War"}}]},
{"label": "EVENT", "pattern": [{"TEXT": {"REGEX": "Project\s+?Red-light"}}]},
{"label": "EVENT", "pattern": [{"TEXT": {"REGEX": "Moon\s+?mission"}}]},
{"label": "EVENT", "pattern": [{"TEXT": {"REGEX": "UFO"}}]},
{"label": "EVENT", "pattern": [{"TEXT": {"REGEX": "Apollo\s+?11"}}]},
{"label": "EVENT", "pattern": [{"TEXT": {"REGEX": "Johnson's\s+?war\s+?on\s+?poverty"}}]},
{"label": "MISC", "pattern": [{"TEXT": {"REGEX": "AIDS"}}]},
{"label": "NORP", "pattern": [{"TEXT": {"REGEX": "Chinesse"}}]},
{"label": "NORP", "pattern": [{"TEXT": {"REGEX": "Vietnamese"}}]},
{"label": "EVENT", "pattern": [{"TEXT": {"REGEX": "Identified\s+?Alien\s+?Crafts"}}]},
{"label": "EVENT", "pattern": [{"TEXT": {"REGEX": "IAC"}}]},
{"label": "NORP", "pattern": [{"TEXT": {"REGEX": "multiculturalists"}}]},
{"label": "NORP", "pattern": [{"TEXT": {"REGEX": "Vietnamese"}}]},
{"label": "NORP", "pattern": [{"TEXT": {"REGEX": "Anglo-Saxon"}}]},
{"label": "NORP", "pattern": [{"TEXT": {"REGEX": "Patriots"}}]},
{"label": "NORP", "pattern": [{"TEXT": {"REGEX": "Patriot"}}]},
work on Python with Jermaine
2023-04-08 20:16:14 -04:00
]
ruler.add_patterns(patterns)
workingDir = os.getcwd()
repairing and restoring XSLT, spot fixes to personTagger.py
2023-04-28 01:26:51 -04:00
CollPath = os.path.join(workingDir, '../pre-src-xml')
redirecting filepaths in personTagger.py
2023-04-28 01:40:42 -04:00
outputPath = os.path.join(workingDir, '../src-xml')
work on Python with Jermaine
2023-04-08 20:16:14 -04:00
# Everything in original conspiracy directory.
insideDir = os.listdir(CollPath)
print(insideDir)
# Copies files in case they do not exist
def copyTextFiles(file):
content = []
# Reads the contents of file, and saves each line of file into the content array.
with open(CollPath + "/" + file, 'r', encoding='utf8') as inFile:
for line in inFile:
content.append(line)
print(" ~~~~~~~~~~~~~~~~~~~~~~~~~~~ copying " + file + " ~~~~~~~~~~~~~~~~~~~~~~~~~~~ ")
inFile.close()
# With the contents copied, a loop will go through the array and write it all in a new file in output folder.
with open(outputPath + "/" + file, 'w', encoding='utf8') as f:
for line in content:
f.write(str(line))
# Function runs through the tokens of given file. Entities are stored in array, then returned. Called by regexFile().
def entitycollector(tokens):
# creates a new file that includes all of the found entities.
lots of stuff, python names, xquery to html, xml regex clean-ish
2023-04-24 14:37:38 -04:00
with open('conspPERSON.txt', 'w') as f:
work on Python with Jermaine
2023-04-08 20:16:14 -04:00
entities = {}
# goes through each entity in the token list.
for ent in sorted(tokens.ents):
lots of stuff, python names, xquery to html, xml regex clean-ish
2023-04-24 14:37:38 -04:00
entityInfo = [ent.text, ent.label_]
work on Python with Jermaine
2023-04-08 20:16:14 -04:00
stringify = str(entityInfo)
f.write(stringify)
f.write('\n')
entities[ent.text] = ent.label_
# return all entities with its label and text.
return entities
# Function runs regex through given file.
def regexFile(file):
fileDir = os.path.join(outputPath, file)
with PySaxonProcessor(license=False) as proc:
# grabs the original xml file and stores it in a variable for later. this some xquery bs
xml = open(fileDir, encoding='utf-8').read()
xp = proc.new_xpath_processor()
node = proc.parse_xml(xml_text=xml)
xp.set_context(xdm_item=node)
# xquery goes through original text, and stores it all in a single string.
xpath = xp.evaluate('//p ! normalize-space() => string-join()')
string = str(xpath)
# regex goes through the text and deletes anything that is not a letter or space.
cleanedText = regex.sub(r'[^A-z ]+', ' ', string)
lots of stuff, python names, xquery to html, xml regex clean-ish
2023-04-24 14:37:38 -04:00
cleanedText = regex.sub(r'\n+', ' ', cleanedText)
work on Python with Jermaine
2023-04-08 20:16:14 -04:00
# gets the tokens of the clean text.
tokens = nlp(cleanedText)
wrappedText = xml
# grabs all the entities in file and stores it in a list/array.
dictEntities = entitycollector(tokens)
# if anything exists in the list, the following code will run.
if dictEntities:
# it will check through each entity in the list and see its entity type. it is looking for "PERSON" tokens
# in this instance, which includes of nouns and names.
for entity in dictEntities.keys():
lots of stuff, python names, xquery to html, xml regex clean-ish
2023-04-24 14:37:38 -04:00
if dictEntities[entity] == "PERSON" or dictEntities[entity] == "LOC" or dictEntities[entity] == "ORG" or dictEntities[entity] == "GPE" or dictEntities[entity] == "NORP" or dictEntities[entity] == "EVENT":
work on Python with Jermaine
2023-04-08 20:16:14 -04:00
# key_template variable is the elements we wrap around found instances.
lots of stuff, python names, xquery to html, xml regex clean-ish
2023-04-24 14:37:38 -04:00
key_template = "<ent type='" + dictEntities[entity] + "'>" + entity + "</ent>"
work on Python with Jermaine
2023-04-08 20:16:14 -04:00
# loops through wrappedText until all entities are wrapped.
wrappedText = wrappedText.replace(entity, key_template)
# Saves newly wrapped elements and then writes it into new file.
with open(fileDir, 'w', encoding='utf8') as f:
f.write(wrappedText)
print("WRAPPING " + entity)
repaired regex in checkTags
2023-04-09 10:47:55 -04:00
checkTags(file)
# ebb: Added above line to send the tagged file to the checkTags() function for cleaning.
work on Python with Jermaine
2023-04-08 20:16:14 -04:00
repaired regex in checkTags
2023-04-09 10:47:55 -04:00
# This part of the code is a WIP.
# ebb: I just activated it, and it works! (Nice job.) I altered it just a bit. May need more regexes to match.
work on Python with Jermaine
2023-04-08 20:16:14 -04:00
## It tries to find weird or invalid elements/tags and fix them.
def checkTags(file):
content = []
fileDir = os.path.join(outputPath, file)
with open(fileDir, 'r', encoding='utf8') as inFile:
for line in inFile:
content.append(line)
# With the contents copied, a loop will go through the array and write it all in a new file in output folder.
with open(fileDir, 'w', encoding='utf8') as f:
for line in content:
# match = regex.search(r"(<ent type='.+?'>[^<>]*?)<ent[^>]+?>([^<>]+?)</ent>([^<>]*?</ent>)", line)
# if match:
repaired regex in checkTags
2023-04-09 10:47:55 -04:00
# print("broken line found, fixing...")
# ebb: NOTE: IF this function only processes a line when there's a regex match, we'd have a serious problem:
# we'd not output the rest of the file--only the cleaned matches. So the output files would be mostly empty!
# Better to just string-clean every line using regex.sub(). Where there's no regex match, no substitution will happen.
origLine = line
work on Python with Jermaine
2023-04-08 20:16:14 -04:00
# newLine = regex.sub(r"(<ent type='.+?'>[^<>]*?)<ent[^>]+?>([^<>]+?)</ent>([^<>]*?</ent>)", r"\1\2\3",line)
repaired regex in checkTags
2023-04-09 10:47:55 -04:00
# <spe<ent type='ORG'>cia</ent>l>
newLine = regex.sub(r"(</?spe)<ent type='ORG'>(cia)</ent>(l>)", r"\1\2\3", origLine)
# newLine = regex.sub(r"(<)<ent type='ORG'>(di)</ent>(v>)", r"\1\2\3", newLine)
newLine = regex.sub(r"(<ent type='[A-z]+?'>[^<]*?)<ent type='[A-z]+?'>([^<]+?)</ent>([^<]*?</ent>)", r"\1\2\3", newLine)
newLine = regex.sub(r"(<ent type='[A-z]+?'>[^<]*?)<ent type='[A-z]+?'>([^<]+?)</ent>([^<]*?</ent>)", r"\1\2\3", newLine)
newLine = regex.sub(r"(<ent type='[A-z]+?'>[^<]*?)<ent type='[A-z]+?'>([^<]+?)</ent>([^<]*?</ent>)", r"\1\2\3", newLine)
lots of stuff, python names, xquery to html, xml regex clean-ish
2023-04-24 14:37:38 -04:00
newLine = regex.sub(r"(<ent type='[A-z]+?'>[^<]*?)<ent type='[A-z]+?'>([^<]+?)</ent>([^<]*?</ent>)", r"\1\2\3", newLine)
newLine = regex.sub(r"(<ent type='[A-z]+?'>[^<]*?)<ent type='[A-z]+?'>([^<]+?)</ent>([^<]*?</ent>)", r"\1\2\3", newLine)
newLine = regex.sub(r"(<ent type='[A-z]+?'>[^<]*?)<ent type='[A-z]+?'>([^<]+?)</ent>([^<]*?</ent>)", r"\1\2\3", newLine)
repaired regex in checkTags
2023-04-09 10:47:55 -04:00
# ebb: I'm repeating the above just in case of the weird event of triple or quadruple nested <ent> tags in <ent> tags.
# We saw it happen on the LOTR project and running it through multiple passes of the above line ultimately got rid of them all
# preserving only the outermost tags.
newLine = regex.sub(r"(<ent type=')<ent type='ORG'>(ORG)</ent>('>)", r"\1\2\3", newLine)
trying my best to fix the regex and editing python code
2023-04-27 00:54:09 -04:00
newLine = regex.sub(r"(<ent type='[A-Z]+'>)<ent type='[A-Z]+'>(\w+)</ent><ent type='[A-Z]+'>(\w+)</ent>(</ent>)", "\1\2 \3\4", newLine)
newLine = regex.sub(r"(<ent type='[A-Z]+?'>)(\w+)\s+?(<ent type='[A-Z+?]'>)(\w+)(</ent>)(\w+)(<ent type='[A-Z]+?'>)(\w+)(</ent>)(</ent>)", r"\1\2 \4 \6 \8\9", newLine)
repairing and restoring XSLT, spot fixes to personTagger.py
2023-04-28 01:26:51 -04:00
# ebb: Problem line below: eliminates <ent type="PERSON">John Kennedy</ent>'s
# newLine = regex.sub(r"<ent type='\w+'>(\w+)</ent>('\w)", r"\1\2", newLine)
repaired regex in checkTags
2023-04-09 10:47:55 -04:00
#
# <spe<ent type='ORG'>cia</ent>l>
# <<ent type='ORG'>di</ent>v>
if origLine != newLine:
print("broken line found, fixing...")
print(origLine + "\n INTO.")
print(newLine)
f.write(str(newLine))
print("File checking finished.")
work on Python with Jermaine
2023-04-08 20:16:14 -04:00
for file in insideDir:
copyTextFiles(file)
regexFile(file)
#checkTags(file)
repaired regex in checkTags
2023-04-09 10:47:55 -04:00
# ebb: You don't really want to activate checkTags here,
# because it would run over the untagged input files.
Reference in New Issue Copy Permalink