MetaSpider, a metasearch engine, is described and evaluated in this paper. The search engine receives queries in the form of simple Boolean expressions, reformulates them, and sends them to various canonical search engines to get back a number of top results. The results are filtered for availability and Boolean expression satisfaction. The results can be presented as a list, or as a 2D map representing a categorization of the pages into concepts.
The authors claim that a major research contribution of the paper is its “new” way of evaluating the system. A number of subjects are asked to compare MetaSpider with two other metasearch systems, based on six queries. Qualitative and quantitative assessments are collected and summarized. The evaluation method is interesting, but its validity is questionable. The systems are not compared under the same conditions (each system employs a different list of input search engines). It is also not clear why these particular six queries were selected. Furthermore, six queries are perhaps not enough on which to base any serious conclusions.
The idea of integrating metasearch and categorization techniques in a single system is interesting and practical. It makes MetaSpider unique for its time, and is a good concept for system developers. In terms of research, however, the paper is weak. The authors should narrow their aims and provide a more in-depth study.