A Search Engine based on Multi-Agents-System

Abstract

A multi-agents system (MAS) is a system composed of multiple interacting agents. MAS is used to solve difficult and impossible for an individual agent or monolithic system to solve. Usually these problems are decomposable to sub-problems, each one will be the duty of one or more agents.
In this paper a new approach for the search engine is proposed depending on a MAS. The main functional subsystems of the engine such as crawler, indexer, updater, and query builder are implemented as agents. The proposed search engine based on multi-agents system, SEMAS, is designed to provide full text, up to date database, query builder, and general search engine web application. The agents in SEMAS are communicating via simple syntax agent-communication facts (ACF). SEMAS was tested by a population of web documents of 1000 web sites of various subjects. The results show that SEMAS provide accurate ranking depending on the visual effect of the page, the up to date database, and ease of query tuning obtained by the query builder. Furthermore many software agent features are attained such as reusability, distributed problem solving (DPS), distributed artificial intelligence (DAI), and parallelism.