This is a basic LSTM sentiment analyser, train on an ad-hoc dataset of movie reviews, yelp reviews and manually classified tweets. Data should be in files with each line being a single entry:
Classification should be a number.
0 indicates negative,
1 indicates positive and
2 indicates neutral.
This is an attempt at training a word2vec model. Using three datasets:
- The ad-hoc sentiment dataset in the previous project
Gutenbergdammitdataset from https://github.com/aparrish/gutenberg-dammit
Wikipediadataset from polyglot project https://sites.google.com/site/rmyeid/projects/polyglot