Research On Feature Words Extraction And Emotional Tendency Analysis Of Video Commentary

With the rapid development of the Internet in China,people's life has changed greatly.People can learn more about our country and even the whole world of things at a faster and faster rate,and to express their views on these things.In the field of network video,people can find their own favorite video to watch and publish a user comments or messages,and so on.To a certain extent,these comments and messages can express the user's views and attitudes.So they are very valuable for video sites,producers,actors and users.At present,in the field of English,the analysis of emotional tendency has achieved some results.However,in the field of Chinese,because of the particularity of Chinese,the research is still relatively scattered.In this paper,the method of basing on HowNet emotional tendency analysis is used to analyze the video review,the main contents include the following aspects.Firstly,we collect user comments from the website.Using the Chinese text segmentation algorithm for user comments,and tag part-of-speech for the result.Further,the paper put forward a fixed pattern of video review,and pick up the characteristic words from video reviews in accordance with this pattern.Secondly,the paper put forwardavideoreview model.The model is a general framework for video analysis.I proposed the emotional reference words set.This set contains a derogatory word set and commendatory term set.Using the method of basing on HowNet emotional tendency analysis calculates the correlation between the feature words and the set of emotional reference wordsto judge the emotional tendency of the feature words.Thirdly,the paper classified the user commentsaccording to the feature words.All comments of a video are classified into several categoriesaccording tothe commentary angle and the number of this angle in all the comments.Finally,the automatic processing of the feature extraction and sentiment analysis of the video commentary is realized.The final results will be displayed in the form of a histogram.I analyze the final result.