Structured Processing Research And Functional Implementation Of Chinese Insurance Clauses

The main work of the paper is to analyze the data structure and characteristics of the insurance contract terms and to achieve the structured processing function of insurance contract clauses.The interpretation of traditional insurance contract terms mainly depends on manual processing,which means these policies are interpreted by insurance salesmen and are transmitted to policy-holders.However,due to the different professional levels of insurance salesmen,the ambiguous explanation for insurance contracts often cause disputes between companies and policyholders.Besides,the sharp increase in the number of insurance products also increases the cost of the policyholder's choice of insurance product.Therefore,this paper intends to carry out the structured processing of insurance contract by natural language processing technology and achieve the storage and classification of structured data in key-value form.Firstly,the thesis studies the text data of health insurance contract terms and combines with the insurance law rules to study data of hierarchy and the data characteristics for the insurance contract terms and designs the processing procedure of insurance document structured data,and implements the insurance document structured processing functions,completes the data parsing and stores on insurance contract.At the same time,in order to meet the functional requirements,this paper also completes the automatic acquisition of insurance data and the multi-format text data preprocessing module.The specific work contents are as follows:(1)Analyzes the characteristics and structure of insurance contract data and summarizes itsdata has features such as “multiple attributive”,"sentence similarity" and "professional"and so on.Based on the data characteristics,this thesis designed the treatment scheme.(2)Proposes an attribute template extraction method based on parameter statistics.Firstly,narrowing the scope of attributes in each item by the single-pass clustering algorithm,then calculating the parameters of IDF and IC-value and combining the professionalphrase library established in this thesis to filter candidate attribute names.Finally,aninsurance document RDF attribute extraction template is established.(3)Designes a structuring process for insurance contract terms,which mainly includes threemodules: insurance contract terms pretreatment module,the terms of the contract theRDF template extraction module and the insurance contract document structuredprocessing module,and the thesis also explains the structured process and function ofeach module.(4)Adds constraint rules of structured data and calibration function.The module canoptimize extraction template and the extraction rules by improving thesaurus,developing library data mark and checking feedback,and also enhancing the availabilityof the structured processing results.Finally,this dissertation uses the real insurance contract document data obtained by network to test the structured processing method.The results show that the proposed method can achieve the desired goal.Base on medical insurance domain thesaurus,the thesis breaks the restriction of common word segmentation software on domain words and achieves automatic data extraction.Also,it provides data support for future insurance product recommendation and article information retrieval.