Design And Implement Of Scoial Relationship Mining System Based On Emotion Analysis

With the rapid development of web 2.0 and we media,more and more users are becoming protagonists of the internet.The communications&social applications such as FaceBook,Twitter and Weibo are going deeply into people's daily life.In China,Weibo as a platform of information exchange and sharing was deeply loved by people because of its ability of information spreading and membership organization.In Weibo,every user builds relationship to exchange information and share information by using follow.With the accumulation of long time,every user constitutes a complex social network system by using follow.The user relationship of Weibo is different from Facebook and QQ,the attention mechanism of Weibo doesn't clearly express the true social relationship of the users.For example,the big V mechanism of Weibo can make big V to gain more follower than others,but the users maybe don't support the big V users,with which they have followed.So,we need to mining the true social relationships of users of Weibo and find out the users social attributes and show the result by using a directly way.In this thesis,we introduce a system of discovering the real social relations by using emotion analysis-a social relationship mining system based on emotion analysis.In this thesis,we use the real Weibo data as research object,and the core of the system is a supervised algorithm based on Support vector machine and a unsupervised algorithm based semantics.The mian functions of the system are emotional tendency analysis and mining the social relationship of Weibo users based on emotion analysis.In the unsupervised algorithm based on semantics,first,we give different weight of verb,adjective,adverb and the punctuations that have emotional tendency.Then,we grade every sentence combined with emotional word dictionary and the feature of the Weibo short text.At last,we judge the emotional tendency of every single Weibo based on the total points.In the supervised algorithm based on Support vector machine,first,we use TF-IDF to compute the weight of every words and make a weight dictionary.Then,because the weight of many general service words and stop word is very big,we must rid of some terms that weight is greater than a certain threshold to avoid influence the result of emotion analysis.But when we rid of the terms that weight is greater,we maybe rid of some emotion terms,negative word and adverb of degree.To avoid rids of these terms,we must traverse the dictionary before removing these terms.Again,we give every term of emotional word dictionary a sole id,and add punctuations such as exclamatory mark that can reinforce emotion.At last,we also judge the emotional tendency of every single Weibo based on the total points.In this thesis,first of all,we introduce the domestic research status of correlation technique of the system.Then,we confirm the feasibility and the function of the system by analyzing the system from requirements analysis,data analysis and feasibility analysis.Again,we design the general design and function module design to the system.Moreover,we make detailed design and implementation of every module based on overall design.At last,we test the system,and assess the functions of the system,and summarize the advantages and deficiencies of the system,and point out the direction of the future work.