Weibo sentiment analysis (II.)

The previous article mentioned a few questions about sentiment analysis, and perhaps these questions can inspire you, but before we look at these questions, we need to know more about the characteristics of Weibo and whether it will have an impact on our emotional analysis.

      In addition to some very cow x people, in fact, most of the microblogging users are grassroots, and even like the author, the grassroots are not, just a little cock silk ~ Weibo owners rarely useful real names, most of the network name, because it is a virtualized environment, so comments can be reckless The spread of Weibo is often dandelion-style, that is, a microblog owner issued a compelling microblog, will form a theme, reflected in the label, the other microblogging will be around this topic comments, the next round of Weibo may also be for the previous round of Weibo comments, and so on, Thus, a topic-based microblog group with a distinct theme was formed. The size of a topic-based microblog group depends on how much attention is paid to the topic. Here I have a question, is this dandelion-style spread. If we are going to do emotional analysis, as the depth of communication increases, does the emotional outcome change? Will the user who commented later be affected by the previous comment? This may be a point.       Unlike the general microblogging, the text structure of the topic-based microblog determines its linguistic features: short sentences, negative tendencies, strong verbal and emotional expressions, strong expression of emotions, and dilution of rational evaluation, in which the object of evaluation does not appear directly in the sentence, the language is not standardized and so on. Let's talk about these characteristics separately.       (1) A short sentence, a simple word multi-bo has a limited number of words, so often short. Topic-based microblogging and because there is a clear topic, people are around this topic to express opinions, write micro-bo, as long as the expression of mood, attitude, few words can also, a word is also OK, so compared with the general style, topic-based micro-blog sentence relatively short, the majority of simple sentences.       This may return to the previously discussed topic, the first question, a few words can express the whole emotion? If the expression is not omnipotent as the user's emotional analysis? Are the results of the analysis effective? The second problem, because the topic of micro-blog has a clear topic, so, most of the microblog will lack of subject and object, which will undoubtedly bring difficulties to our analysis. I think the first situation can be solved by the idea of a collinear chain, that is, the user's comments on the microblog are captured in a unified and unified emotional analysis. The advantage of this is that while users may not be able to express their emotions in a single tweet, they can express different emotions in multiple tweets. That in this case, also leads to the second question, with the micro-blog forwarding, the user's feelings will change? This problem can also be achieved with collinear results.       (2) The formation of multi-topic micro-blog is based on certain social topics and social events, and the topic of concern in today's society is more negative, which makes the topic micro-blog in the expression of views, the majority of negative tendencies. This is what happens in other social media, too., people's feelings are broadly divided into joy, anger, sorrow, music, love, evil, fear, you can see, negative emotions accounted for the vast majority. In these emotions, each emotion is divided into different levels, and at each level there are different levels, so it may be necessary to establish an effective, comprehensive negative emotion dictionary. A good emotion dictionary can clearly get the degree of emotional vocabulary, and a good negative emotion dictionary will get an accurate and comprehensive analysis result.       (not to be continued)

Weibo sentiment Analysis (ii)