Redes neurais e logica formal em processamento de linguagem natural

AUTOR(ES)
DATA DE PUBLICAÇÃO

1993

RESUMO

This dissertation is about Natural Language Processing (NLP). NLP consists of a series of tasks the machine should carry out in analysing a texto In literature, there are papers having different approaches. This work combines two approaches: based on logic and connectionism. The proposed work is divided in three parts. The first makes the parsing, or the syntactic analysis of sentences in the Portuguese language, based on logic. The second part takes care of the semantic analysis, or the verification of the meaning of words in a sentence. This is achieved through artificial neural networks that "Iearn" the binary representation of the words (their semantic microfeatures). This approach is called connectionism. Its major advantage is the ability of generalizing, i. e., it is able to recognize a word even it is not presented to the nets. The third, and last, part of this work is about the use of recurrent networks in text analysis. This kind of network is to "Iink" the words in a sentence because the recurrent net is given memory, which makes it able to "remember" the last word seen in a sequence. This is useful to link the words in a sentence like the subject to the object, the object to the complement, etc. This makes a sentence an entire item to be analysed

ASSUNTO(S)

redes neurais (computação) processamento da linguagem natural (computação) inteligencia artificial

ACESSO AO ARTIGO

Documentos Relacionados