Opinion Mining System for Twitter Sentiment Analysis

After the paradigm shift produced by Web 2.0, the volume of opinion on the Internet has increased exponentially. The expansion of social media, whose textual content is somewhat subjective and comes loaded with opinions and assessments, can be very useful for recommending a product or brand. This information is an interesting challenge from the perspective of natural language processing, but is also an aspect of deep interest and great value not only as a marketing strategy for companies and political campaigns, but also as an indicator measuring consumer satisfaction with a product or service. In this paper, we present an opinion mining system that uses text mining techniques and natural language processing to automatically obtain useful knowledge about opinions, preferences and user trends. We studied improvements in the quality of opinion classification by using a voting system to choose the best classification of each tweet, base on of the absolute majority of the votes of the algorithms considered. In addition we developed a visualization tool that automatically combines these algorithms to assist end-user decision making. The opinion mining tool makes it possible to analyze and visualize data published on Twitter, to understand the sentiment analysis of users in relation to a product or service, by identifying the positive or negative sentiment expressed in Twitter messages.