基于描述文本和实体标签的网络视频分类算法

Web Video Classification Algorithm Based on Description Text and Entity Tag

目前,各大社交平台和视频点播网站的网络视频数量出现了爆炸式的增长,如何快速准确地对这些网络视频进行归类和管理成为了研究的热点问题﹒为了较好地解决这种分类任务,文中提出了基于描述文本和实体标签的网络视频分类算法,该算法结合了描述文本内容和知识图谱中的实体标签来构造文档-特征矩阵﹒实验结果表明使用了实体标签的视频分类算法性能更好,平均精确率和平均召回率以及平均F1值比未使用实体标签的视频分类算法要高2%以上﹒

At present there has been an explosive growth in the number of web video on major social platforms and video on demand web sites. How to quickly and accurately classify and manage these web videos has become a hot spot of research. In order to solve this classification task, a web video classification algorithm based on description text and entity tag was proposed in this paper. The algorithm combines the description text and the entity tags in the knowledge graph to construct a document-feature matrix. The experimental results show that the video classification algorithm using the entity tag shows better performance, and the average precision and average recall and the average F1 value are higher 2% than the video classification algorithm of the unused entity tag.