| 摘 要: 为解决视频推荐系统中推荐精度不够精准的问题，提出一种K-means优化的自组织映射(Self-organizing Map, SOM)神经网络视频推荐方法。首先，爬取视频网站的数据并对其进行处理；其次，将处理后的数据输入K-means算法优化的SOM神经网络中，得到聚类结果；最后通过计算归类视频的弹幕数量、点击量、评分等推荐出优秀的视频。文中系统的预期结果为在主界面选择分类并输入关键词之后，通过算法计算，为用户推荐感兴趣的视频，并按评分高低列出视频的超链接。实验结果表明，优化的SOM算法在视频推荐的精度上提升了5%—8%。 |
| 关键词: 视频推荐；K-means；SOM算法；优化 |
|
中图分类号: TP391.41
文献标识码: A
|
| 基金项目: 辽宁省教育厅科学研究经费项目(SYDR202004). |
|
| Video Recommendation System based on SOM Neural Network Algorithm Optimized by K-means |
|
FU Limei
|
(Department of Software Engineering, Dalian Neusoft University of Information, Dalian 116023, China )
fulimei@neusoft.edu.cn
|
| Abstract: Aiming at the insufficient recommendation accuracy in video recommendation systems, this paper proposes a Self-organizing Map (SOM) neural network video recommendation algorithm optimized by K-means. Firstly, video website data is crawled and processed; secondly, the processed data is input into the SOM neural network optimized by K-means algorithm to obtain the clustering result; finally, relatively excellent videos are recommended by calculating the number of bullets, clicks and scores of classified videos. The expected result of the proposed system is that after selecting categories and inputting keywords on the main interface, the videos that interest the users will be recommended by the algorithm after calculation, and the hyperlinks of the videos will be listed according to ratings. Experimental results show that the optimized SOM algorithm improves the accuracy of video recommendation by 5%-8%. |
| Keywords: video recommendation; K-means; SOM algorithm; optimization |
