I want to rate the similarity between two tags. For example the words technology, computer and chip should have high similarity, a word like food should be low similarity.
Given the recent advancements in AI and more traditional approaches from NLP, how would you solve that problem nowadays in Node JS?
I tried to get the Google News dataset from 2013 to run with word2vec, but neither this or other of the word2vec libraries seem to be running with modern versions of node (or at least i didnt manage to make any work yet with the dataset). There are also repos like fast-text or Text Similarity Test (using TensorFlow.js).
Right now I'm wondering both which approach would be technically the best to fullfil my problem (also approaches that I didn't consider yet) and how to technically implement it in code.