CODEWITHSUNDEEP
nlptf-idfcosine-similarityvsm
divyum
divyum

Reputation: 1316

Vector Space Model Introduction

What are different types of VSM (vector space model)?

One which I know (as per wiki) is tf-idf (cosine similarity is used in this method, but its not a separate method). Which are other ways?

Also what are different dimensions of a word in a document (except frequency) being talked about in wiki?

Is there any hierarchy for VSMs?

P.S. Please correct me if I am wrong anywhere...

Upvotes: 0

Views: 437

Answers (1)

Nikita Astrakhantsev
Nikita Astrakhantsev

Reputation: 4749

Take a look at the paper From Frequency to Meaning: Vector Space Models of Semantics - survey for Vector space models of semantics. For your particular question about frequency, see sections 4.2 and 4.3: in short, one uses tf-idf, PMI, PPM with further smoothing (mainly by some variations of SVD).

But, since this survey is dated by 2010, there is no mention of deep learning-based VSM, see the paper.

Upvotes: 1

Related Questions