Weighted Word Embeddings for Twitter Sentiment Classification

What is it about?

Sentiment classification plays an important role in Sentiment Analysis. It is challenging to develop an automatic method for classification problems without annotated training data. In this paper, we present a WWE (weighted word embeddings) method, which uses a continuous word representations algorithm (Word2Vec) to train a vector model. According to the cosine similarity between the vector of a word and the vectors of seed words, a polarity score of this word can be calculated. We then use the weighted polarity scores of words to compute a polarity score of the whole tweet. Unlike the previous learning-based approaches, our method does not require annotated data gathered for the purpose of training models. We collected the Super Bowl related tweets to demonstrate the WWE classification method. Experiments are performed with promising outcomes.

Featured Image