Research On The Constructing Of Corpus Ont Social Security Audit

The normal operation of the social security system relates to the vital interests of the people, in the time of information explosion how can we use of domain information effectively to guide the adjustment of social security audit system is a pressing problem. The corpus of social security is used for processing of the field of social security auditing language, management information, thereby supporting the optimization of auditing methods.In this thesis, the authors analysis the information of Social Security auditing field and depending on the feature of the information, proposed the way of building the corpus with original corpus, processed corpus and semi-automatic updated management, with the assessment of the information of corpus on the situations of corpus sources and domain spoken material set. Using Web search automatic way to download constantly the updated corpus from the specified areas. Supported by spoken material, the author use Back Forward Traversal with Double Dictionary algorithm to get spoken material continuously, richen the material set, at the same time the organizational structure and the update management of the spoken material is given. When the domain material is descript the time, frequency, circulation, degree and the source level of the material as the feature value, based on support vector classification methods the corpus of field is managed.With the mass domain corpus, this paper, the author compared information of domain corpus with the audit method of social security auditing, according to the updated situation of spoken corpus and the classification of the domain corpus the authors dynamic monitored the updated of field of domain corpus continuously, explored changes of information contained in the domain corpus. With the feedback of the diversification on the domain information the authors can guide the adjustment of social security auditing methodology to support the intelligent auditing.