nlp - How can one resolve synonyms in named-entity recognition? -
in natural language processing, named-entity recognition challenge of, well, recognizing named entities such organizations, places, , importantly names.
there major challenge in though call of synonymy: the count , dracula in fact referring same person, it possible never discussed directly in text.
what best algorithm resolve these synonyms?
if there feature in python-based library, i'm eager educated. i'm using nltk.
you describing problem of coreference resolution , named entity linking. i'm providing separate links not entirely sure 1 meant.
- coreference: stanford corenlp has 1 of best implementations, in java. have used python bindings , wasn't happy- ended running data through stanford pipeline once, , loading processed xml files in python. obviously, doesn't work if have processing in real time.
- named entity linking: check out apache stanbol , links in following stackoverflow post.
Comments
Post a Comment