Typical Opinions Mining based on Douban Film Comments in Animated Movies

Abstract The film comments data contains a huge amount of mining research value, and text mining analysis of the animated film’s comments can objectively reflect the quality of the animated film presentation and the problems generally expressed by the audience. However, these film comments are often mixed. The existing well-known film reviews websites have not excavated typical reviews on the user’s film comment text, so neither the audience nor the animation creators can analyze and apply the comments.This paper presents a general framework for mining typical opinions of film comments and uses crawler technology to obtain network review data, extract comment keywords based on the TF-IDF algorithm, and convert comments segmentation into word vectors trained by a neural network through Word2Vec. Then, using certain extraction rules and the K-means algorithm, the typical opinions with the same semantics but different expressions are aggregated together, and the typical opinions of the animation review of “Monkey King: Hero Is Back” are excavated. From the excavated information, we find out the production problems of the animation, so as to provide a certain reference to the creation of animated movies.