6
$\begingroup$

How can one process natural language (English) to extract named entities (aka NER)? Or at least, entities which can be mapped to a Mathematica Entity. Note that trying to map entities via simple tokenization, POS or the dependency tree is not the same as NER. Also, I am aware that one can use neural networks to train NER but I would hope there is an easier solution within Mathematica.

Improve this question
$\endgroup$
3
  • 4
    $\begingroup$ Mathematica 12 will support NER. Coming soon. $\endgroup$
    – Alexey Golyshev
    Commented Mar 15, 2019 at 8:06
  • $\begingroup$ I've encountered this problem too. In my case, it was sufficient to look for capitalised words within a sentence, or the first word returning WordData a its head instead of List on POS queries. Your mileage may vary. $\endgroup$
    – Shredderroy
    Commented Mar 15, 2019 at 15:03
  • $\begingroup$ Partial answer: (1) you can use R within Mathematica, and R has several packages for NER; (2) you can use JLink and connect to, say, the Java Stanford NER library. (And with the results from those do Entity mapping...) $\endgroup$
    – Anton Antonov
    Commented Mar 31, 2019 at 1:39

1 Answer 1

Reset to default
5
$\begingroup$

As mentioned in the comments, Mathematica v12 has indeed named entity recognition via the TextContents command. Details in the documentation.

Improve this answer
$\endgroup$

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.