PrediÃÃo de palavras baseada em modelos ocultos de Markov / Word prediction based on hidden Markov models




Social inclusion search has been promoting the technology development extending people with special needs computerâs use. Amongst the innumerable auxiliary tools, words prediction is an important accelerator that diminishes the number of actions to be executed and, consequently, time and effort to do it. Word anticipation in a text being typed, called word prediction, can be developed from statistical considerations such as occurrences counting, its predecessors and successors. One of the most used techniques for stochastic analysis is the Hidden Markov Models, being the most indicated to word prediction in a collection of writings or recorded remarks used for linguistic analysis sets, called corpus. In such a way, this work added developed software for users with special necessities to the prediction tool based on Hidden Markov Models efficacy. The developed word prediction algorithm was associated to a freeware keyboard simulator, aiming to offer facilities and consequently better performance to the users. The resulting program offers a list containing the next ten most probable words, considering the grammatical class and frequently word inside classes groups. The algorithm is based on a corpus of infantile texts carefully chosen. It was developed in Visual Microsoft C++ 6.0Â based on Viterbis Algorithm and presents an innovative form to deal with its Matrix Occurrences. Digitizing tests of corpus texts were carried out in order to verify the algorithm uses; tests involving preparation of texts out the corpus, for testing the capacity of prediction and literal consistency; and tests with two groups of volunteers, one of them consisted of children with special needs, in order to verify its functionality and applicability. The prototype presented excellent performance, showing the facilities promoted by developed prediction technique uses. The innovative analysis and the Hidden Markov Models application had become the prediction process fast and accurate. Thus, the prototype reduces the effort in text production, collaborating for digital inclusion e, consequently, social inclusion of people that many times are pointed as disabled


sistemas de recuperaÃÃo da informaÃÃo informÃtica na educaÃÃo hidden markov models processamento da linguagem natural (computaÃÃo) markov, modelos ocultos de information storage and retrieval systems inclusive education engenharia biomedica educaÃÃo inclusiva natural language processing (computer science) educational in informatics

Documentos Relacionados