Web directories as a knowledge base to build a multi-agent system for information sharing
Web Intelligence and Agent Systems
| Hi-index | 0.00 |
A neural based multi-agent system for smart html pages retrieval ispresented. The system is based on the E\alphaNet architecture, aneural network capable of learning the activation function of itshidden units and having good generalization capabilities. Systemgoal is to retrieve documents satisfying a query and dealing with aspecific topic. The system has been developed using the basicfeatures supplied by the Jade platform for agents creation,coordination and control. The system is composed by four agents:the Trainer Agent, the Neural Classifier Mobile Agent, theInterface Agent, and the Librarian Agent. The sub-symbolicknowledge of the Neural Classifier Mobile Agent is automaticallyupdated each time a new, not included before, document topic isrequested by the user. The Neural Classifier Mobile Agent interactsalso with the Librarian Agent for retrieving the documents in therepositories and with the Interface Agent for user interaction. Theproposed system is particularly useful for classifying documentsstored in private networked document repositories that, for variousreasons (i.e. privacy, security, and so on), cannot be indexed byan external search engine. The system is very efficient: thepreliminary experimental results show that in the best case aclassification error of 9.98% is obtained .