Caching in Multi-Agent based Architecture for Distributed Information Retrieval Systems

Caching is an e ective technique for improving performance in Databases and Information Retrieval (IR) systems. Traditional IR systems access the collection indices to perform searches. Such searches on large corpora for queries oft repeated can be computationally redundant. In addition, querying remote sources can be expensive because of large communication overheads and frequent inavailability (if dealing with proprietary databases). One way of dealing with this problem is to cache query results. This can provide results to similar future queries. In this paper, I focus on problems related to caching in Distributed IR (DIR) system. An investigation of research issues like caching of documents and metadata routing is provided. Caching implementation e orts are tested on the CARROT-II DIR system. Caching has been known to have a strong e ects on query response times and throughput. The aim of this project is to provide eÆcient caching mechanism and design proper cache maintenance strategies that will improve performance in a distributed retrieval environment.