python - NLTK Relation Extraction - custom corpus in relextract.extract_rels -
i learnt there built-in function in nltk extract relations ner-tagged sentences according following:
import re in = re.compile(r'.*\bin\b(?!\b.+ing\b)') fileid in ieer.fileids(): doc in ieer.parsed_docs(fileid): rel in relextract.extract_rels('org', 'loc', doc, corpus='ieer', pattern = in): print(relextract.rtuple(rel)) it seems me promising general purpose, understood relextract.extract_rels accepts 'ieer' or 'conll2002' parameter corpus. in case, usage restricted these 2 corpora, isn't it? how 1 utilize own corpus (presuming, of course, ner-tagged).
this should more of comment don't have enough reputation.
you can pass custom corpus doc argument after it's been pos-tagged , converted list of chunked trees. custom corpus, should use corpus='ace'.
for example, in this answer, use extract_rels tag custom corpus.
Comments
Post a Comment