Research And Application On The Data Lineage Of Audit Doubt Based On The Schema Mapping

With the constantly strengthen of the informationization degree and the development ofthe Internet technology, data integration technology got unprecedented development. Usersneed to face how to judge the accuracy of the integrated data resource and how to judge itssource and so on, when they get the integrated data resource richly. In practice, theseproblems also exist in the process of pension audit. How to verify the doubtful data found inthe process of audit when the auditors audit the integration of audit data. By studying thetheory of data lineage, we provide an establishing method of mapping relation and designRDSA algorithm, thereby automatically solve the problem of doubtful data verification.The first stage is to establish the mapping relation between the origin model and thestandard model in the process of pension audit. It can be divided into two processes, whichare mapping relation of table level establishing and mapping relation of data attribute fieldlevel establishing. The mapping relation of table level can be divided into three steps. Firstly,we formalize the pension audit experience by the production representation, and build up anaudit experience knowledge base. Then, based on the knowledge base, we design a table ruleextraction algorithm, and extract a rule set from every table. Lastly, we establish the mappingrelation of table level between two models. Through constructing two kinds of distributionmodel vector of attribute field and the corresponding classification algorithm, we establish themapping relation of data attribute field level on the base of SMDD method.The second stage is to design RDSA algorithm. On the basis of audit method analysis,we puts forward the strategy which combines reverse pattern mapping with connected graphand corresponding algorithm. In the pension audit system, audit method on the origin modelcan be generated from standard model automatically, so as to solve the problem of doubtfuldata lineage tracking and verify the doubtful data.Finally，based on the actual pension audit data model, we validate the algorithm which isconstructed during study by experiment. Meanwhile, we track the lineage of doubtful data,find the origin data, and realize the doubtful data verification.