A multi-agent softbot to retrieve medical information on Internet

Area of Science:

• Information Science
• Computer Science
• Medical Informatics

Background:

• The World-Wide Web presents challenges for information retrieval due to its unstructured, distributed, multimedia, and multilingual nature.
• Existing search tools like subject hierarchies and general search engines have limitations in precision, multilingual indexing, and information distribution.
• These limitations hinder efficient and accurate access to online information, particularly in specialized domains.

Purpose of the Study:

• To address the limitations of current web search technologies.
• To propose a novel approach for information retrieval using a distributed, multi-agent system.
• To evaluate the effectiveness of this approach in a specific domain, such as medicine.

Main Methods:

• Development of the M.A.R.V.I.N. (Multi Agent Retrieval Vagabond on Information Networks) project.
• Distribution of the information indexing task among specialized agents, each focusing on a specific domain.
• Implementation of a domain-specific search engine, MedHunt, for the medical field.

Main Results:

• Successful application of the M.A.R.V.I.N. approach to the medical domain.
• Creation of a comprehensive medical index and the MedHunt search engine.
• MedHunt currently indexes 50,000 medical sites with over 100,000 monthly accesses, indicating significant user engagement and utility.

Conclusions:

• The distributed, agent-based indexing approach offers a viable solution to enhance web information retrieval precision and multilingual capabilities.
• The M.A.R.V.I.N. project and its medical implementation, MedHunt, demonstrate the practical benefits and scalability of this methodology.
• This approach holds promise for improving access to specialized information across various domains on the World-Wide Web.