A recursive descent evaluation algorithm on policy context similarity

Abstract

This paper proposes a new policy text similarity calculation method after the analysis and research of the domestic and foreign traditional text similarity calculation based on vector space model. The method which is based on the policy text pretreatment and recursive descent technology presents a computation method of fitting degree among policy clauses and similarity among policy text. The general classic algorithms are involved in the problem of word frequency. This method doesn't establish word frequency vector directly, but it merges the same words. Then by using the recursive descent method, the consistency problem in policy chapters are analyzed from three different levels of policy clauses, policy paragraphs and policy chapters, reducing the space complexity of the fitting calculation. Experimental results show that compared with the existing typical similarity calculation method, this method can increase the efficiency and precision of consistency verification.