Associative Text Representation and Correction
Autorzy/Authors: Adrian Horzyk, Marcin Gadamer
Wydawnictwo/Publisher: Springer Verlag, LNAI, 2013
Abstract:
Linguistic communication takes a major role in human communication and information exchange. Information is usually transferred in a text form - sentences. Text descriptions allow us define new terms, gather knowledge and learn more quickly thanks to the associative mechanisms that work in our brains. Automatic and intelligent text processing and compression are very important nowadays. This paper introduces a novelty associative way of storing, compressing and processing sentences. This paper describes an associative linguistic habit neural graphs (ALHNG) that are able to store and activate various important associative relations between letters and words simultaneously in many sentences. These graphs enable us to semi-automatically define various terms and contextually process text corrections after the knowledge collected from previously read texts. The ALHNG construction has a linear computational complexity. The association and triggering interconnected elements in any given context have a constant computational complexity. It also compresses sentences in a very effective way.
Keywords:
linguistic habit graphs, associative linguistic habit neural graphs, associative neurocomputation, text compression, associative artificial intelligence AAI, bio-inspired techniques for text mining.