Characterization of data on the Gnutella peer-to-peer network

The paper presents a characterization of the different types of files for which users search on the Gnutella network. The searches are classified into five different categories: audio files; video files; image files; documents and program files; files that cannot be classified as one of the other four types. Audio files are the most abundant type of file for which users search, followed closely by video files. Relatively few documents and program files are queried, and queries for image files are almost negligible. The percentage of queries for X-rated material is found to constitute a significant portion of the total queries. In addition to queries, the paper also quantifies the distribution of distinct files shared by hosts and duplicate files that are shared by more than one host.