Markov-Based Automatic Term Extraction

Abstract

This paper presents an automatic term extraction method based on Markov process. The method aims to extract multi-word domain terms from English corpora. The paper proves that the extracting term process is a Markov chain firstly, and then gives the steps of the Markov-based method. In order to evaluate our method, we use a corpus related to computer science got by Web crawlers, and extract domain terms by methods introduced in the paper. The experiment data shows that our method out performs other methods.