Unsupervised acquisition of a markov model for word correction using Wikipedia
Adquisición no supervisada de un modelo Markov para corrección de texto utilizando Wikipedia
Contenido principal del artículo
Resumen
ONTARE. REVISTA DE INVESTIGACIÓN DE LA FACULTAD DE INGENIERÍA
This paper presents a work in progress on the area of automatic acquisition of corpora for spelling correction. Wikipedia contains a high quantity of information including relationships between concepts and named annotations. However, it also contains linguistic information such as misspellings written by many of the Wikipedia collaborators. In this paper, we propose an efficient method to analyze the link structure of Web-based dictionaries to construct a list of misspelled words and their corrections. The method is currently being researched and applied to the Wikipedia as a corpus.