Research On Mining And Validation Method Of Policy Genome Based On The State Space Reduction

Along with the gradual development of legal institution in our nation,policies in all domains are consummated further,new policies are constantly promulgated while old ones are still valid.Consequently,the analysis of the policy text are drawing more and more attention in the domain of policy research,becoming an urgent and important problem to be solved in the progress of legal institution in our nation.Based on the analysis and summary of current achievements on policy research and text similarity calculation both domestic and foreign,this thesis puts forward a method to mine and validate the policy genome by the reduction of the state space.Aiming at the problem of the sparseness of high dimensions in the traditional vector space model,firstly this method preprocesses policy text by the natural language processing technology,and sets up the reasonable expression dimension of policy text by automated-summary-based state space reduction method.During the process,this method solves the unstable weight and low efficiency problems of the characteristic words due to the estimation by the field experts.Meanwhile,in order to solve synonymy relation between characteristic words in policy text,an influence-based vocabulary replacement algorithm is proposed.Then,based on the state space reduction of policy text,this thesis introduces the concept of policy lineage,and carries out the relevant definition and acquisition of the policy genome by the combination of the nature of biological gene in genetics.Finally,this thesis uses the dominant genes of the policy to calculate of text similarity.When the difference between dominant-gene-based policy text similarity data and traditional policy text similarity data exceeds a certain threshold,then policy latent gene will be mined,and this thesis puts policy dominant gene and latent gene together as a part of policy genome,thus to achieve the goal to calculate the similarity with policy genome instead of policy text.This study solves the problem of low efficiency caused by high complexity during the similarity computation in the analysis process of the massive policy text,and providing a necessary foundation to analyse the policy text efficiently and precisely.In the end,this thesis carries out repeated experimental verification on the set of policy texts.The experiments test and verify the availability of the method by the analysis andcomparison between the proposed method in this thesis and the traditional method.