Design And Implementation Of Pan-Video Recommendation System Based On Online Comments

With the explosive growth of video information on the Internet,people enjoy a huge amount of information,but also feel confused about the choice of video.With the fierce competition in the video field,whether the company can provide better services for users will directly affect the development of the company.Therefore,there has been a personalized recommendation technology to provide users with accurate video recommendations based on user data,such as basic user information,user interest,browsing history,etc.There are many types of recommendation algorithms and methods,but the user's movie review information is used insufficient by the company,how to apply high-quality review information to the video recommendation field becomes the main problem to be solved in this paper.Firstly,relevant data information is obtained through the web crawler,including: basic information of 1920 movies of Douban Net,500 reviews of each movie;9103 music information of NetEase Cloud Music;1768 novel information of 555 Chinese net;17173 game network 774 game information;Secondly,the shortcomings of the traditional collaborative filtering algorithm are analyzed,and a recommendation algorithm based on user review information is designed based on the crawled data set.A distributed computing framework for Hadoop and Spark is built.Based on this framework,the CBOW model in Word2 Vec is used to generate a vector representation of the movie names in the corpus.The movie name can be approximated because the vector generation process is a context-dependent form.The vector is used instead of the movie.Then the Euclidean distance is used to calculate the distance between the movies;Thirdly,according to the idea of the MVC design pattern,the final implementation of the pan-video recommendation system is in the form of a website.This system can make video recommendation for users according to their browsing records,the final result of this paper can provide not only video recommendation services,but also some video-related contents,such as music,novel and game.A highly experienced recommendation system is realized.