Under the background of big data era, it is an inherent requirement to use big data to carry out audit work. In the recent audit of the implementation of social assistance policies in Ningbo, the Ningbo Audit Bureau has actively explored ways to use large data technology in audit practice, intensified the comprehensive utilization of data, and improved the ability of using information technology to check problems, evaluate and judge, and analyze macroscopically.
　　First, expand the coverage of big data. In order to find out the implementation of social assistance policies in various departments of the city, auditors have explored the audit mode of "centralized collection, discovery of doubtful points, grouping verification and systematic research" by using large-scale data from provincial, municipal and county levels. They have carried out data on industry, commerce, taxation, public security and provident fund of more than 10 departments, such as civil affairs, education and medical services. In order to ensure that the coverage of the audit coverage is complete, the implementation of the salvage policy is not neglected. Through multi-dimensional and multi-level analysis of massive data, various data analysis models and methodological systems are attempted to discover the rules and characteristics contained in the related data, and to dig out the clues of audit doubts so as to maximize the effectiveness of audit resources.
　　Two is the big data collision analysis of doubt. Auditors use SQL statements through multi-department, multi-field, multi-data, multi-dimensional data analysis, accurate detection of doubtful points, collision analysis between different data, through the current discovery of hundreds of people illegally enjoying the disabled two subsidy, low insurance or special hardship policies, some medical and housing assistance funds management is not standardized. At the same time, a number of difficult personnel failed to enjoy the subsidy for old housing reconstruction, and the phenomenon of rescue should be saved. In view of the problems found during the audit, the relevant departments actively rectify, and have asked the counties (cities) to cite one example and the other as an opportunity to actively carry out self-examination and self-rectification of other relief work, focusing on whether the release of relief funds is in violation of regulations, and whether the policy of benefiting the people is in place to ensure the legitimate rights and interests of the difficult groups.
　　The three is big data analysis to achieve high efficiency. The implementation of the social assistance policy audit time is short, involving a wide range of people, only a sample of the low-security, special hardship personnel have more than 40,000, and related to various types of business data records of more than 10 million items. In order to improve the audit efficiency, the audit team concentrates on the basic data collection and analysis pretreatment before the trial, and establishes more than 50 large data audit models on the audit site, finds tens of thousands of abnormal data and more than 20 audit doubts, thus sums up the overall audit direction to achieve the purpose of accurate audit. On the same day, the audit entry point directly dispatched an audit team to extend and verify the doubtful data, to ensure that "not more than one person, many one person" principle, so that eligible people should enjoy, should make up for, must not open their mouth at will, enjoy the allowance should not be enjoyed, causing social instability, to ensure that the total number of rescued personnel. Enjoy accurate bonus. (Zhou Jie, Hu Shengke, Liang Wenting)