Bag of Embedding Words for Sentiment Analysis of Tweets

Abstract

This paper presents an alternative of solution based in artificial intelligence to simplify the human effort that implies the analysis of the impact for businesses of their publications in social networks services. This analysis is very important because the audience manifest its opinion mostly in texts that must be processed one by one to know their content and use it in benefit of the business, this implies the use of resources to read each comment and extract characteristics that make possible to determine whether the comments are, positive reactions or negative. Our solution can obtain most effective reports than the ones generated by manual procedures, it means that demands less resources and leads to the save of time and money during the extraction of the answers to a Twitter’s publication. We use BOEW and Word2vec to generate the characteristic vector for each of the answers. Finally, to make the sentiment analysis we use statistic classification models to polarize comments.