A personal search agent system

This paper demonstrates a personal search agent system which enables the users to personalise their search and hence to save their time and effort. The system uses its own ranking algorithm and provides two types of search: a direct search (via a local database) and an indirect search (via third party search engines). The system achieves search personalisation through an initial user interaction that decides on what search criteria should be considered more important than others. The system also attempts to learn the user behaviour through a user feedback mechanism. This mechanism can cause two search attempts with the same query to return two different results; because of changing the user preferences even though the search is run on the same data set. Issues regarding the setup, implementation and experimental results of the system are illustrated here.