Resumo

RICO SULAYES, Antonio. Belief functions in computational linguistics: a reconsideration of belief modes. Tóp. Sem [online]. 2016, n.36, pp.139-161. ISSN 2594-0619.

Proposed as a core element for a probabilistic model, belief functions are aimed at representing the beliefs an artificial agent or subject produces as it observes the world. Although such a construct has been successfully applied in a number of scientific fields, most of its applications in the design of artificial agents have exploited the modality of believing using only a few of all of its possible modes of expression. This has recently changed with a number of current computational linguistics applications which employ all modes of believing for this cognitive modality. The expansion in the use of the modes of believing for the design of artificial agents has resulted in an improvement of the performance by these computational linguistics applications. These results confirm the current validity of Greimas' warning as to not to ignore this cognitive modality.

Palavras-chave : theory of evidence; artificial agent; speech recognition; conversational agents.

        · resumo em Espanhol | Francês     · texto em Espanhol     · Espanhol ( pdf )