Spam characterization and detection in peer-to-peer file-sharing systems
Proceedings of the 17th ACM conference on Information and knowledge management
| Hi-index | 0.00 |
Peers in peer-to-peer file-sharing systems cannot effectively share their files if they are poorly described. Terms one user employs to describe an instance of a file may not be those that are commonly associated with the file, making this instance difficult to locate. To alleviate this problem, a server can ask its peers for help in improving the description of files they have in common. We consider the design of a fully distributed, automatic system for the exchange of descriptive metadata. Experimental results show that the proposed techniques are effective in improving search accuracy with reasonable cost.