krinkv / information-ranking

The Unlicense
0 stars 0 forks source link

Fix vectorizing and calculating cosine similarity. #29

Closed KristiyanCvetanov closed 9 months ago

KristiyanCvetanov commented 9 months ago

Current solution for creating vectors is too inefficient. Make them maps (index in english dictionary-> tf-idf value for term) so we save memory and perform faster (would it be really faster?) cosine similarity.