Towards Holistic Web-Based Information Retrieval: An Agent-Based Approach

This paper presents an agent-based system for bolstering holistic information retrieval via the WWW. In Ellis’ holistic model of information seeking behaviors, the information seeking activities include: selection of sources, browsing and differentiating, monitoring as well as extraction. Through the use of a query processing agent (QPA), information filtering agents (IFAs) and information monitoring agents (IMAs), these activities can be automated. By establishing sub-class relations among (key)words the query processing agent (QPA) expands a query with a list of sub-queries to select appropriate URLs. Using three relevance metrics: word relations, frequency and nearness of keywords, the IFA is used to determine the relevance of a page. Additionally, IMAs can be used to track changes in the content of selected pages, paragraphs or tables in websites. Empirical results demonstrated that the QPA can find appropriate number of websites, and IFAs are effective in filtering relevant information. As part of an on-going work, an Information Extraction Agent is currently being designed and developed.