Research On Short Video Recommendation Algorithms Based On User Behavior Analysis

In the context of big data,information overload is serious,and traditional search engines can no longer satisfy users' demand for information.The user wants to obtain information conveniently and quickly,and the information producer also wants to pass the information to the target user,and the intermediary between the two is the recommendation system.Short video recommendation is one of the applications of the Internet recommendation system,because short video itself has some characteristics such as uneven video quality,less affiliate information,less content tags,and shorter video time.The traditional recommendation algorithm is not suitable for the short video recommendation.Collaborative filtering algorithm is the most mature algorithm in recommendation system.For short video data loss and video content information,it is difficult to extract features.Based on collaborative filtering,this paper proposes a short video recommendation algorithm based on user behavior analysis.This article starts with the short video user behavior,mainly analyzes what behaviors users have when watching short videos,and whether these behaviors are related to the preference for short videos.This article analyzes the dominant and implicit behavior of users.Clean the behavior information of short video users and establish a User-Video scoring matrix to compensate for sparseness of date.Solve the cold start problem of the recommendation system by establishing the "User basic information registry" and "User interest preference table".The user's interest in short videos will shift over time,but the user's interest is basically stable in a short period of time.The recently added user behavior has a huge impact on the recommendation prediction,so we use time weighting to adjust the time variable.Finally,the short video that the user may like is recommended to the target user according to the KNN recommendation algorithm.In the field of short video recommendation,through experimental research,it is found that the short video recommendation algorithm based on user behavior analysis in this paper exceeds the traditional general recommendation algorithm in the accuracy,recall and F values.The algorithm proposed in this article can make short video users get the video they want to watch more quickly,and understand the user's needs more than the user.