Abstract: Based on the characters of Web and Peer-to-Peer（P2P） systems，an algorithm of auto-generating index of media files for P2P systems from Web is proposed.This method can efficiently solve the problem about lack of descriptors of media files，which causes the low recall and precision in key words searching.Besides，to satisfy the increasing information of Web and the changing popularities of people，an auto-updating strategy for peers is applied.From the experiments，it shows that this method can expand the describing information for media files and significantly improve the searching capacity of P2P systems.

摘要： 基于环球网（Web）的特点和用户在点对点（P2P）系统中搜索的习惯，提出了一个在P2P系统中对媒体文件自动生成索引的方法。该方法有效地解决了媒体文件描述符不足所带来的查询精度低的问题。同时，提出了一个在P2P系统中节点信息的更新策略。实验表明，描述符扩展后，媒体文件查询结果的准确率得到了显著的提高。