The final project of introduction of ML. All the data including dataset(labeled) and dictionary are acquired from Google.
To be specific: Dataset2 and Dataset3 are the same, and Dataset3 and label3 are splitted from Dataset2. In the program, you should use Dataset3, label3 or Dataset1, label1.
When classifying the twitter data with dictionary1, the accuray is around 0.6, which depends on the threshold. In this coding, classes are Positive, Negative and Neutral. With the sentiment dictionary, you can get a final score for every sentence; and comparing the final scores with the threshold, you can have the final classification rsult.
The dictionry I use is: SemEval-2015 English Twitter Sentiment Lexicon