Weibo sentiment analysis (i)

      say Weibo has been a long time, but the micro-blog information mining is just beginning, this one of the reasons of course there is information mining technology is not mature, but I think the main problem is still in the embryonic stage of the technology of Chinese processing. The Chinese language itself is a large amount of information, ambiguous vocabulary, coupled with the micro-blogging language semantics, micro-BO Media ontology mixed with a large number of tags, resulting in the slow development of micro-BO technology. On the current network, the user actively express their views through the network or the attitude towards other people or events, subjective strong, the language of the micro-Bo carrier is only 140 words, so that the information in the microblog show fragmentation, immediacy and mobile features, and no longer have complete contextual information. Through the free, convenient and instant expression of their emotions, Weibo has become a fashion on the internet, and it has become an important place to generate and talk about hot events, in which hot events refer to events, topics or information that are widely concerned, debated, and discussed in a given time, so the discovery of hot events in Weibo platforms, Research on monitoring and management has become very important.     Weibo as a new type of media, has its own unique text structure. Topic-based micro-Bo refers to a topic that is the label of the analysis of opinions, the discussion of the micro-blog form, so in the use of views, the expression of the use of the language means and the object of evaluation of the looming also has a distinctive feature.       I think the question of how much emotion 140 words can express is worth discussing. Perhaps in most cases, the discussion of an event will be more than 140 simple, let alone express a profound opinion. The user's comment on the event is more of a joke than a real comment, which leads to two problems: first, the user's comments on the issue can not be fully expressed in the premise of the user's attitude to the problem, the users will not be able to reflect the user's true emotional attitude, second, because the user emotional expression is not comprehensive, Perhaps the user's emotional microblog has become the user's subconscious first emotion, and at this stage may also need to the user psychology, behavioral analysis, which is beyond the "microblog sentiment analysis" of the scope. Therefore, if you want to really dig out the user's emotional inclination, the user's psychology, personality and habits should have a great influence ratio.       A little deeper, because only 140 words can be entered, the user must express their attitude in a limited space, the user will not appear in most of the comments on the common line of the word cluster? If you can find the user comments on the word line chain, I think this can be a user's character and psychology to make a certain analysis. Because the text is different from the expression, the expression often reflects a person's character, while the text can reflect a person's psychological dynamics, in fact, often psychological dynamics can determine the behavior of people. Topic back to Weibo. In WeiboThere is a label, because the existence of the label, leading to a large number of micro-blogging subject and object, such as: "#汽油涨价 # I would like to say dirty words," or "#明星整容" Disgrace! The first sentence is missing the second object, and the second sentence is missing the subject. In the analysis of the microblog, the proportion of the label is a guide, I think this is a relatively good topic.

Weibo sentiment Analysis (i)