Online Search Scope Reconstruction by Connectivity Inference
Michael Chan, Stephen Chi-fai Chan, Cane Wing-ki Leung · 2007
To cope with the continuing growth of the web, improvements should be made to the current brute-force techniques commonly used by robot-driven search engines. We propose a model that strikes a balance between robot and directorybased search engines by expanding the search scope of conventional directories to automatically include related categories. Our model makes use of a knowledge-rich and wellstructured corpus to infer relationships between documents and topic categories. We show that the hyperlink structure of Wikipedia articles can be effectively exploited to identify relations among topic categories. Our experiments show the average recall rate and precision rate achieved are 91% and between 85% and 215% of Google's respectively.