CODEWITHSUNDEEP
nlpword2vecsentiment-analysisdoc2vec
mrc_03
mrc_03

Reputation: 21

Doc2Vec vs Avg Word Vectors : Which is better for Sentiment Analysis?

I was performing Sentiment Analysis on the IMdb dataset on Kaggle. I used the BOW approach with bigrams and that gave me a decent accuracy of ~89%. But I dont know how to approach the same using word embeddings: Should i go for averaged word vectors or doc2vec?

Someone please help. Thanks in advance.

Upvotes: 1

Views: 453

Answers (1)

Adnan S
Adnan S

Reputation: 1882

Here's a recent blog post comparing word2vec averaging vs doc2vec performance. The post favors doc2vec. It also depends on what classification model you are using (logistic regression, SVM, LSTM, etc.)

Upvotes: 1

Related Questions