Intelligent Semantic-Based System for Corpus Analysis through Hybrid Probabilistic Neural Networks

摘要

The paper describes the application of hybrid probabilistic neural networks for corpus analysis which consists of intelligent semantic-based methods of analysis and recognition of word clusters and their meaning. The task of analyzing a corpus of academic articles was resolved with hybrid probabilistic neural networks and developed word clusters. The created prototypes of word clusters provide the probabilistic neural networks with possibilities of recognizing corpus clusters. The established corpus comprises 1376 articles, from specialist leading SCI-indexed journals, and provides representative samples of the language of science and technology. In this paper, a review of selected issues is carried out with regards to computational approaches to language modelling as well as semantic patterns of language. The paper features semantic-based recognition algorithms of word clusters of similar meanings but different lexico-grammatical patterns from the established corpus using multilayer neural networks. The paper also presents experimental results of word cluster semantic-based recognition in the context of phrase meaning analysis.