hajaulee / sentiment

LSTM CODE: Sentiment Analysis

Install library

• pip : pip install -r requirements.txt
• pip3: pip3 install -r requirements.txt

Download dictionary

• Glove 50 dimension, 40000 words of dictionary file

DataSet

We have a dataset (X, Y) from amazon imdb where:

X contains 1000 sentences (strings) consist of 500 negative samples and 500 positive samples Y contains a integer label between 0 and 1 corresponding to each sentence

Embeddings

Glove 50 dimension, 40000 words of dictionary file is used for word embeddings. It should be downloaded from http://akii.tk/QVNCUL (file size = ~168MB))

• word_to_index: dictionary mapping from words to their indices in the vocabulary (400,001 words, with the valid indices ranging from 0 to 400,000)
• index_to_word: dictionary mapping from indices to their corresponding words in the vocabulary
• word_to_vec_map: dictionary mapping words to their GloVe vector representation.