Enhancing Web Searches from Concept Map-based Knowledge Models

TLDR

Many public search engines retrieve relevant information, yet the returned page lists are often too large or contain irrelevant content. The study aims to improve search engine results for queries generated while constructing or browsing concept map‑based knowledge models. The authors developed two mobile‑agent algorithms, SAgent and WAgent, that exploit concept maps’ propositional and hierarchical structure to filter and rank search engine results, and evaluated them empirically with six users querying Google, Altavista, Yahoo, and Excite. The agents’ ranking proved promising, showing improved relevance over the original search engine outputs.

Abstract

Although many publicly available search engines can retrieve relevant information reasonably well, the list of the retrieved web pages is still often too large or contains information that has no relevance to the query. Our goal is to improve the results of these search engines for queries generated by users while constructing and/or browsing concept mapbased knowledge models. By exploiting the propositional and hierarchical nature of concept maps, we have developed two algorithms, SAgent and WAgent, for filtering and ranking the results obtained by the search engines. The algorithms were implemented via mobile agents and evaluated empirically. In our experiments, six subjects submitted queries based on a concept map to publicly available search engines (Google, Altavista, Yahoo, Excite), and were asked to rank the relevance of the results; the agents also filtered and ranked the engines’ output. The results are promising, in that the Agents’ ranking was ...