The Design And Implementation Of Sentiment Analysis System Based On Social Media

With the rapid development of Web2.0and mobile internet, a variety of new applications have given rise to abundant of users’reviews data, and most of these reviews data contains users’point of view and emotional tendencies. Sentiment analysis of users’review data has great help for companies that need to know users’feedback on their products and services, as well as for the government that wants to know people’s opionions on hot events. And the goal of sentiment analysis is to discover users’views or emotional tendencies of an entity expressed in the text.This thesis first describes relevant researches on sentiment analysis and research situation. Then we start to explain the principle of texts’ sentiment analysis algorithm. The principle of the sentiment analysis algorithm includes two levels:The first level is the polarity classification algorithm of emotional words. This thesis designs and implements emotional words polarity classification algorithm based on emoticons. This algorithm classifies the emotional words into positive or negative emotional tendency by the orientation of emoticons and the corelation between emotional words and emoticons. The second level is the emotional tendencies analysis algorithm of the text. This thesis classifies the text into positive or negative emotional tendency by classifying the subjective and objective of the text and classifying the emotional polarity of the subjective text with the emotional model.After completing the core algorithm of emotional tendency analyais, the thesis designs and implements a sentiment analysis system of the social media based on this core algorithm, which we called SAS. This system analysises and statistics the emotional tendencies of blogging topics from the large number of micro-bloggings published by users, and shows the results of emtional tendency analysis via web pages. The thesis explains the analysis of system requirements, architecture design, database desgin and web pages, then tests the system carefully.