中文语义特征属性扩展的视频推荐系统设计

Design of Video Recommendation System with Extended Attributes of Chinese Semantic Feature

推荐系统在视频服务中广泛应用,设计了一种基于中文语义特征属性扩展的视频推荐系统,使用中文jieba分词系统对视频的剧情介绍进行语义分析提取关键词并排序,以此扩展视频特征向量的维度,通过相似度计算进行评分预测,比仅用元数据特征效果更好。在此基础上,结合视频服务测量系统的分析结果,在推荐系统中加入网站的页面加载速度以及带宽大小两个因素,可以为用户推荐服务质量更好的视频网站。

The recommendation system has been greatly applied in video services. In this paper,we design a recommendation system with semantic feature extensions. We extract keywords by using Chinese word segmentation system jieba word segmentation system to semantic analysis on the video storyline. Then,we sort the keywords and expand the dimensions of the video feature vector. We found that scoring predictions by similarity calculations are better than using metadata features alone. Based on the analysis results of the video service measurement system,we add two factors including the page loading speed and the bandwidth size of the website in the recommendation system,which could provide users a video website with a better service quality.