Cognitive Multi-agent Systems for Integrated Information Retrieval and Extraction over the Web

Frederico Luiz Gonçalves de FreitasGuilherme Bittencourt

In the Web, there are classes of pages with similar structuring and contents (e.g., call for papers pages, references, etc), which are interrelated forming clusters (e.g., Science). We propose an architecture of cognitive multi-agent systems for information retrieval and extraction from these clusters. Each agent processes one class employing reusable ontologies to recognize pages, extract all possible useful information and communicate with the others agents. Whenever it identifies information interesting to another agent, it forwards this information to that agent. These "hot hints" usually contain much less garbage than search engine results do. The agent architecture presents many sorts of reuse: all the code, DB definitions, knowledge and services of the search engines. We got promising results using Java and Jess.

Caso o link acima esteja inválido, faça uma busca pelo texto completo na Web: Buscar na Web

Biblioteca Digital Brasileira de Computação - Contato:
     Mantida por: