Retrieving information from the invisible web using mobile agents

This study proposes a model of information retrieval on the invisible Web by using the mobile agent paradigm. The developed architecture uses the power of a search engine to provide a list of sites of the invisible Web which are likely to be relevant and launches a dynamic search on these sites, thanks to mobile agents. To compare and experiment in real conditions, two versions were implemented: a version using the traditional client/server paradigm and a version using mobile agents. Client/server tests on actual Websites generated satisfactory qualitative results. A series of comparative experiments of the two versions implemented were carried out using a test site. Results show that the mobile agent version generates much less traffic and is thus faster than the client/server version, especially with low bandwidth. Moreover, as the mobile agents carry out calculations on the server rather than on the client's site, this approach relieves the resources of the client terminal. Thus, the mobile agent approach seems particularly advantageous in the case of weak resource terminals such as PDAs.

Uncontrolled Keywords

Information retrieval, Mobile agent, Search engine, Invisible web