Abstract
Chat rooms and newsgroup on the internet is a valuable, and often free of charge, source of information. In this paper, a design of smart chat room bots that automatically retrieve and filter on line messages is proposed. The design is based on internet technology and Bayesian Networks. Technical details of connecting to and retrieving data from web based chat room servers are presented. A Naive Bayesian network classifier is implemented using frequency of the keywords that mostly appear in the selecting messages as input features. A prototype of such a message classification system has been implemented. It has been trialed on detecting investment related messages from four Australian chat room sites.
| Original language | English |
|---|---|
| Pages (from-to) | 569-574 |
| Number of pages | 6 |
| Journal | IFIP International Federation for Information Processing |
| Volume | 228 |
| Publication status | Published - 2006 |
| Externally published | Yes |
Keywords
- Bayesian network
- Information retrieval
- Web mining