Abstração de provas para explicação na web

AUTOR(ES)
DATA DE PUBLICAÇÃO

2008

RESUMO

In order to answer queries to human users on the web, intelligent agents use a reasoning mechanism to process information and access services available on the web. Every step of this reasoning process form a proof and the information presented in these proofs facilitate generating explanations from proofs on the web. However, these automatic generated proofs are not ready to be presented to human users and they need to be transformed and simplified into explanations closer to human language. In this dissertation, we present an architecture for abstracting proofs in order to make them more appropriate for human explanation. This architecture uses abstraction patterns, which are proof fragments that can be replaced by rules that are more meaningful for people. Our abstraction approach consists of using the IWAbstractor algorithm, which has been developed by researchers of Stanford University, along with a set of strategies to abstract automatic generated proofs using abstraction patterns. This way, proofs become simpler and more comprehensible for people. Keywords: Artificial Intelligence, Semantic Web, Knowledge Representation, Abstraction, Explanation.

ASSUNTO(S)

representaÇÃo de conhecimento - dissertaÇÕes internet - dissertaÇÕes sistemas de informacao inteligÊncia artificial - dissertaÇÕes web semÂntica - dissertaÇÕes

ACESSO AO ARTIGO

Documentos Relacionados