TextMoodClassification

This is a text sentiment classification project.
It is a short text classification problem and there are two classes: positive sentiment and negative sentiment.
In the .tsv files, 'id' is unique for each text, 'sentiment' is sentiment classes.(positive is 1, negative is 0) , 'review' is the text.
labeledTrainData.tsv：Training data, which has 24500 texts and each text has its label.
unlabeledTrainData.tsv：Training data, which has 49000 texts, but each text has no label. This dataset could be used to unsupervised learning.
testData.tsv：Test data, which has 22000 texts.(Output the probability of each test text using trained model, and submit result as submission.txt)