KBP
Table of contents
Description
Extracts relation triples meeting the TAC-KBP competition specifications.
For example when run on the input sentence:
Joe Smith was born in Oregon.
The annotator will find the following ("subject", "relation", "object") triple:
("Joe Smith", "per:stateorprovince_of_birth", "Oregon" }
| Property name | Annotator class name | Generated Annotation |
|---|---|---|
| kbp | KBPAnnotator | KBPTriplesAnnotation |
Example Usage
Command Line
java edu.stanford.nlp.pipeline.StanfordCoreNLP -annotators tokenize,pos,lemma,ner,parse,coref,kbp -coref.md.type RULE -file example.txt
List Of Relations
| Relation name |
|---|
| gpe_subsidiaries |
| org_alternate_names |
| org_city_of_headquarters |
| org_country_of_headquarters |
| org_date_dissolved |
| org_date_founded |
| org_dissolved |
| org_founded |
| org_founded_by |
| org_member_of |
| org_members |
| org_number_of_employees/members |
| org_parents |
| org_political/religious_affiliation |
| org_shareholders |
| org_stateorprovince_of_headquarters |
| org_subsidiaries |
| org_top_members/employees |
| org_website |
| per_age |
| per_alternate_names |
| per_cause_of_death |
| per_charges |
| per_children |
| per_cities_of_residence |
| per_city_of_birth |
| per_city_of_death |
| per_countries_of_residence |
| per_country_of_birth |
| per_country_of_death |
| per_date_of_birth |
| per_date_of_death |
| per_employee_of |
| per_member_of |
| per_origin |
| per_other_family |
| per_parents |
| per_religion |
| per_schools_attended |
| per_siblings |
| per_spouse |
| per_stateorprovince_of_birth |
| per_stateorprovince_of_death |
| per_stateorprovinces_of_residence |
| per_title |
Options
| Option name | Type | Default | Description |
|---|---|---|---|
| kbp.model | file, classpath, or URL | edu/stanford/nlp/models/kbp/english/tac-re-lr.ser.gz | Relation extraction model to be used, set to “none” to use no statistical model |
| kbp.semgrex | file, classpath, or URL | edu/stanford/nlp/models/kbp/english/semgrex | Directory containing semgrex rules (rules over dependency patterns) to be used by relation extractor, set to “none” to use no semgrex rules |
| kbp.tokensregex | file, classpath, or URL | edu/stanford/nlp/models/kbp/english/tokensregex | Directory containing tokensregex rules (rules over token patterns) to be used, set to “none” to use no tokensregex rules |
- There are currently models for
Chinese, English, Spanish. The properties files for those languages have the proper settings for using theKBPAnnotatorin that language
More information
There are descriptions of the sentence level statistical model, semgrex rules, and tokensregex rules in the write up for our 2016 TAC-KBP submission, though this paper also includes details about our overall KBP system which is not included in Stanford CoreNLP.