Search Query Generation with MCRDR Document Classification Knowledge
The MCRDR (Multiple Classification Ripple-Down Rules) Classifier was developed to classify documents incrementally. A knowledge base of MCRDR-Classifier consists of two types of rules (refining and stopping rules), categories into which documents are classified, and cornerstone cases used for creating new rules. As document classification knowledge reflects user’s preference for documents, it can be used to generate search queries to retrieve relevant web pages from public search engines. This research aims to propose various query generation methods using MCRDR knowledge base and evaluates them to choose the best one. For this purpose, search queries were generated from ten users’ knowledge bases using the proposed query generation methods and then they were submitted to MSN web search service to retrieve search results. Search results were evaluated with discriminative power (how search results are distinctive?) and domain similarity (how search results are similar to the user’s interest?) criteria to select the best query generation methods.
KeywordsMCRDR Knowledge Reuse Search Engines Search Query Generation
Unable to display preview. Download preview PDF.
- 1.Park, S.S., Kim, S.K., Kang, B.H.: Web Information Management System: Personalization and Generalization. In: The IADIS International Conference WWW/Internet 2003, Algarve, Portugal (2003)Google Scholar
- 2.Kang, B., Compton, P., Preston, P.: Multiple Classification Ripple Down Rules: Evaluation and Possibilities. In: 9th AAAI-Sponsored Banff Knowledge Acquisition for Knowledge-Based Systems Workshop, Banff, Canada. University of Calgary, Menlo Park (1995)Google Scholar
- 3.Park, S.S., Kim, Y.S., Kang, B.H.: Web Document Classification: Managing Context Change. In: IADIS International Conference WWW/Internet 2004, Madrid, Spain (2004)Google Scholar