How to do semantic keyword search with nlp

Question

I want to do SEMANTIC keyword search on list of topics with NLP(Natural Language Processing ). It would be very appreciable if you post any reference links or ideas.

Please consider marking the answer as accepted (grey tick mark on the left of the answer) and upvoting it if you find it useful. — sophros, Feb 15 '21 at 10:27

score 0 · Answer 1 · answered Jul 18 '16 at 13:25

Your questions is somewhat vague but I will try nonetheless...

If I understand you correctly then what you want to do (depending on the effort you want to spend) is the following:

Expand the keyword to a synonym list that you will use to search for in the topics (you can use WordNet for this).
Use collocations (n-gram model) to extend the keyword to the likely bi-, tri-grams and search for these in the texts.
Depending on the availability of the data you may also want to create a classifier (e.g. using good old SVM or CRF) that maps list of keywords into topics (where topic is a class).
Assuming that you have a number of documents per each topic, you may also want to create a list of most frequent words per topic (eliminating stop-words).

Most of the functionality is available via NLTK, Pandas, etc. for Python and OpenNLP, libsvm, LingPipe in Java.

How to do semantic keyword search with nlp

1 Answers1