KEYWORDS EXTRACTION: Selecting Keywords in Natural Language Texts with Markov Chains and Neural Networks

摘要

In this paper we show our approach to keywords extraction by natural language processing. We present revised and extended version of previously shown document analysis method, based on Khonen Neural Networks with Reinforcement, which uses data from the large document repository to check and improve results. We describe new improvements, which we've achieved with preprocessing set of words and creating initial ranking using Markov Chains. Our method shows, that keywords can be selected from the text with great accuracy. In this paper we present evaluation and comparison of both methods and example results of keywords selection upon random documents.