Abstract. Result merging is a key component in a metasearch engine. Once the results from various search engines are collected, the metasearch system merges them into a single ranked list. The effectiveness of a metasearch engine is closely related to the result merging algorithm it employs. In this paper, we investigate a variety of resulting merging algorithms based on a wide range of available information about the retrieved results, from their local ranks, their titles and snippets, to the full documents of these results. The effectiveness of these algorithms is then compared experimentally based on 50 queries from the TREC Web track and 10 most popular general-purpose search engines. Our experiments yield two important results. First, simple result merging strategies can outperform Google. Second, merging based on the titles and snippets of retrieved results can outperform that based on the full documents. 1.

A metasearch engine is a system that provides integrated access to multiple existing search engine mechanisms. Once a query is executed on a metasearch engine, the system passes the query to its participating component search engines, collects the individual results and merges them into a single ranked list. Metasearch engines increase the search coverage of the Web, help solve the extendibility issues in searching the internet, and improve the retrieval effectiveness, and consequently the relevance, of results. Result merging is a key constituent of metasearch engines. When results from several search engines are collected, the metasearch system has to merge them into a unified list. The effectiveness of the metasearch mechanism and the relevance of the result set are closely related to the result merging algorithm used. The purpose of this research is to build a flexible, general purpose metasearch framework and explore a content-directed result merging approach to rank results. Here the content-direction is provided to the framework by the user in the form of documents or text artifacts. A modular metasearch application programming interface (API) based on java has been implemented. The API framework provides interfaces and utilities to develop components of a metasearch system like segregators, scheduler, aggregators, and search service providers. A prototype metasearch engine has been built based on this framework to study the content-directed result merging algorithm.