Sentiment Analysis- The Emotional aspect of Social Media, Website Reviews and Games
(Part 4)
Primer
As I discussed in Part-3, sentiment analysis is used to read emotions expressed as a text or as a voice data, using many techniques which include NLP, semantic search engines, dynamic classification, translation etc., which in turn takes us towards the opinion mining architecture, which I introduced in Part -1 of this series. The image is provided to jog your memory. we will continue where I left off, at BERT architecture –Word embeddings.
BERT Inputs
Frequency based word embeddings ( Contd..)
TF/IDF
The frequency-based word embeddings discussed before had one major flaw. The more frequent words which have less relevance on context takes preceding to the less frequent words which have increased emphasis on the context. Nonetheless this flaw is completely eradicated in TF/IDF method. The value of TF when normalized (making a larger value to small enough value to handle) has a score of one or less than one, as the denominator will always be more than…
