Título: Utilização de Redes Neurais Artificiais na Recuperação de Informação em Bases de Dados Textuais
Autores: Lopes, Ramon da Cunha; Barbosa, Eduardo Fernandes; Braga, Antônio de Pádua
Resumo: The Information Retrieval in the scope of this work, was considered as the retrieval of a document from a set of documents gathered at random and organized by keywords and by the context of each document defined by a librarian. The documents used as base of information were initially classified by placing each key found in alphabetic order, in an inverted list. The keys were counted and gathered according to the number of times that they occurred in each document. Starting with a search algorithm using an Artificial Neural Network, a result of this search of key-words was obtained, associated with a specific context in textual database, with a smaller number of access in relation to the traditional search, using the binary tree algorithm. The algorithm was tested with a set of documents obtained in the Internet, in the areas of Electronics, Artificial Intelligence and Artificial Neural Networks. The weights and biases were stored in a hard disk for future simulations and queries. A study is made to characterize the behaviour at random of the distribution of the keys along the text. From this point on, a probabilistic approach is developed to measure the context of the key-words in the set of the documents.
Artigo em pdf: 4cbrn_041.pdf
Arquivo BibTex: 4cbrn_041.bib