How does universities build a audit big data management system?

On February 18, 2021, the Zhejiang Provincial Party Committee and Provincial Government made a clear conference at the Digital Reform of the province that the digital reform of Zhejiang Province must adapt to the wave of digitalization, establish digital consciousness and thinking, cultivate digital capabilities and methods, build a digital governance system and mechanism, and take the initiative to take the initiative to take the initiative Lead the leap of global digital changes.

In order to better promote digital reform, the meeting requested to open up data assets to the greatest extent, promote the application of data associations, and stimulate the enlarged, superimposed, and double -increase effects of data production factors. In accordance with the requirements of "unified planning, unified support, unified architecture, unified platform, unified standards, unified construction, unified management, and unified operation and maintenance", integrated data catalogs are adopted, and public application support components are used Research and judge evaluation, promote scientific decision -making and efficient implementation.

Zhejiang University of Finance and Economics's audit big data management integrated system was established based on the successful practice of the school's "data center". This system enables the normalized audit data reporting mechanism and model to innovate and break through, solve the audit data reporting in the traditional model, and effectively improves the quality and efficiency of audit informatization work.

Audit data management integrated system architecture

System design ideas

The construction of the integrated system of auditing data management of Zhejiang University of Finance and Economics, in accordance with the universities that cover 20 business domains (including education, assets, libraries, personnel, scientific research, academic, finance, procurement, etc.), which are proposed by the Zhejiang Provincial Department of Education in 2020 The requirements for auditing data submission requirements, relying on the existing "data center" construction results of the school, the unified collection of audit data, regulating storage, and safe storage have been achieved. The data is collected from the source of the business until the full -link closed -loop management of the submission and analysis. The overall design of the project is as follows:

1. Relying on the existing "data center" of the school for incremental development and iterative upgrading. The construction of audit data management integrated system is based on the design of the new generation of global data centers and data governance results in the school, in accordance with the requirements of auditing data reporting, and the design of incremental development and iterative upgrades. In this way, the school's early construction results can be reused to achieve the horizontal expansion of existing data architecture. While the system greatly reduces the overall development workload, it also guarantees the continuous operation of the existing data system.

2. For different data sources, complete data unified collection, centralized management, and standardized output. The audit data submitted data covering the data of the 20 business areas. Some of the data exist in the school's business management system, and some data have no business management system to support it. In response to this situation, the system needs to use different collection architectures for different data sources for unified collection. Business data is uniformly collected through the "data center" to uniformly enter the warehouse and offline data. Concentrated management, including data standard management, data model management, etc.; Build audited data standard warehouses and audit data stations, docking data interfaces of the Education Department, and providing data standardized output.

3. Construct a data verification model, the quality of the entire process and visual monitoring data. The audit data needs to be checked before the submission of the relevant data. Data quality problems can be divided into two situations: structural and content: for structural data quality problems, after the data is unified into the warehouse, the relevant software tools of the "data center" are used for structural cleaning; for content data quality problems, It is necessary to concentrate on the verification data in the form of reporting, and check the inter -table relationship of the data, build a comprehensive data verification model, and remind the data with an abnormal data. Review, modify, and confirm. Through such a complete set of visual data quality monitoring closed -loop process, the high -quality audit data special library is finally generated to complete the data report.

4. Multi -dimensional visualization of audit data through reporting tools and guide school -related decisions. In addition to the "collection -verification -standardization -storage -reporting" full -link closed -loop management work of auditing data management system, you can also conduct multi -dimensional and deep levels of the annual audit data through the reporting tools. The analysis of visualization, strengthen the comprehensive application of audit data in terms of correlation analysis, trend forecasting, etc., effectively promote the digital transformation of the audit business, and improve the level of accurate management and scientific decision -making of the school audit department.

System technical architecture design

The data audit inspection system constructed based on big data computing technology can be divided into four parts: the collection layer, sorting layer, audit inspection layer, and audit submission layer according to the data architecture level. The details are shown in Figure 1.

Figure 1 College big data audit inspection platform architecture

1. Collection layer. Uniformly collect data sources, offline electronic meter data, and file text non -structured data requirements for various business system systems, offline electronic meter data, and file text, and use different types of collection architectures for different types of data sources. At the same time, considering the normalized audit data submission work requirements, the collection method can be configured to two modes: full or incremental. The business data is concentrated by the collection layer to the full amount of the data. The full volume library is a big data storage architecture based on Hadoop. It supports distributed deployment and supports a variety of types of data sources.

2. Sort the layer. This layer is an important level of auditing data asset management, which can help realize the standardization of auditing, structural quality improvement, and data structure integration of data. After integrating the data submission requirements of the audit data, it is uniformly stored to facilitate subsequent query and analysis. The standard data warehouse is the core of data set storage and integration to help realize the integration and sharing of data, and avoid repeated storage and processing of data between systems. The audit market library mainly stores the original data and derivative data provided by the standard data warehouse. Corresponding processing is processed according to the business needs of the audit data, and the required results data are provided. 3. Audit check layer. Through big data visual analysis platform, the results of the results in the audit market library are developed for visual verification interface development, which is convenient for auditors to audit and verify the data content. If there are content data quality problems, the auditors can report it to the data source department in a timely manner for data correction. After the review is completed, the data will be transferred to the audit fixed manuscript library. The data structure in the library is completely consistent with the data structure required by the audit data. At the same time, this layer will also generate a backup of historical data of the year and place it in the audit historical database for historical data retention and subsequent inspection. The integrated data can also be performed through the platform, and the data processing results are displayed in different ways, including fixed reports, multi -dimensional analysis, flexible queries, etc.

4. Audit submitting layer. According to the requirements of auditing data, customized the development of lightweight API interfaces, text uploads, and database direct connectivity to support data reporting business.

System function design

The overall function of Zhejiang University of Finance and Economics's audit data management system is divided into five major modules, which are business system data collection modules, data filling modules, data warehouse management modules, data market modules, and data report modules. The specific design is shown in Figure 2 Show.

Figure 2 The overall function module of the system

Business system data acquisition module

It mainly completes the uniform collection and gathering of data in the school business management system, and includes functions such as interface management, storage management, and task scheduling management.

1. Interface management is the management of custom interfaces and developer interfaces of business system data collection. Specifically includes: custom interface module is used to manage the created ETL interface, which can display the currently created ETL interface situation, support support According to the interface name/ description, interface type, interface status, the ETL interface is retrieved; the developer interface module is used to manage the interface created by the developer, which can display the interface information created by all developers, support the development of the developer name to retrieve the search according to the name of the developer name Essence

2. Storage management, support the management of the storage procedures of accessible data, and support the retrieval of the storage procedure name, description instructions, and database types.

3. Task scheduling management can manage the current data collection and scheduling tasks, support the creation, classification, delete the collection interface task, and support the scheduling name, associated interface name, scheduling type, scheduling status; at the same time, it also supports viewing through the log view to view Interface task running status details.

Data filling module

It mainly completes the unified reporting, importing and management of offline Excel data supported by business management systems, including data source management, data table management, user management and other functions.

1. Data source management, which mainly completes the data source management function of the data filling module. At present, it supports structured databases such as MySQL, MSSQL, Oracle, etc., which can directly display the current connection status and connection information of the data source, and support fast retrieval.

2. Data table management is the most important function of the data filling module. Support users to import the Excel electronic form file data in accordance with the uniform data specifications and data standards; and support users to fill in and edit data online (real -time saved to the database without worrying about data data Lost), can also be downloaded to the local editor, and then uploaded to the database for unified storage management.

3. User management mainly realizes the management of user permissions, and supports different operating permissions for different users, including data viewing, editing, and deleting; and can meet the business needs of multi -person coordinated work. You can differentiate the different tables of the same table. The field gives different users with different editing permissions, and editing authority includes functions such as addition, deletion, modification and investigation.

Data warehouse management module

It mainly includes three sub -modules: data standard management, data model management, and data quality management.

1. Data standard management to regulate the structure and format of audit data to achieve the accuracy and consistency of the data, including standard retrieval, code set management, coding rules management and other functions. Standard search is to quickly match the corresponding standard details in the audit data standard information item. Users browse the details in the results according to the needs. The program searches hot words through algorithm memory to reduce the complexity of the retrieval operation.

Code collection management is used to uniformly manage the various code sets required by audit data, which can be cited by the field of public attributes and data market lists. When viewing the fields that are bound to the code set, support the code that is bound or downloaded and binds. set. The coding rules are managed to manage the coding rules automatically generated in accordance with the setting rules. The encoding rules can be bound to the public attributes of the audit data, and the data in the physical table can be detected. Essence 2. Data model management, the core goal of the integrated system construction of auditing data management is to open up the data islands and build a unified audit data warehouse that meets the standards of the Department of Education. The data model management module provides the data model -related visual modeling function. It uses a combination of top -down and bottom -up to use the model design to assist design standardized audit data models, including data model management, data model design, physical tables, physical tables Management and other functions.

Data model management supports the historical version management of auditing data models, and realizes the process and transparency of the entire process of data model construction. It can compare the models of different versions, check the changing details, and assist auditors to manage the models of different versions. It can back back to the model design status of any node to strengthen the model management and control capabilities; the data model design conducts standardized data model design according to the audit data standard to ensure the consistency of the data from the source. The entity table management needs to generate a data entity table in the standard audit data warehouse after the data model is established. The physical table management function is used to manage the physical table generated by the data model. Essence

3. Data quality management. In order to achieve comprehensive management and intelligent monitoring of the quality of audit data, the "rule settings" in the data quality management module can be set to complete the setting of relevant data quality inspection rules. The data objects of the quality inspection are bound to complete the quality testing of the data. It mainly includes functions such as data quality rules and reports. Data quality rules management can configure the quality detection rules of the audit data, and support the creation of a variety of data structural quality detection rules, including value domain rules, enumeration rules, regular rules, and unique rules.

Data quality report management is mainly to generate data quality reports regularly after data quality testing. Through rich and diverse icon display forms, multi -dimensional statistics can be performed to make the boring data quality more vivid and help all departments to be responsible for it. Related data Establishing a unified perspective of understanding, looking for optimization space for the quality of audit data, and providing a basis and guidance for the formulation of quality improvement plans.

Data market module

As the core of the data set storage and integration, the audit standard data warehouse can realize the integration and sharing of data, and avoid repeated storage and processing of data between systems. The data market module can form a special audit market library, which mainly stores the original data and derivative data provided by the standard data warehouse, and processed accordingly according to the business needs of the audit data, provided the required results data, and completed the required results. Data reporting. This module includes functions such as data market lists, departmental data catalogs, and data open management.

Data report module

In order to meet the statistical analysis of audit data, to ensure the accuracy and credibility of the reporting data, the data report module supports the visual display of the reporting data in the audit data market through the form of the report; The fields are verified, and the data of the verification is reminded and identified; supporting the report data to browse, review and confirm. At the same time, the audit data can be analyzed through the reporting tools to allow audit data to survive to guide school -related decisions.

System highlights and difficult points

At the end of 2020, on the premise of taking into account the safety and accuracy, the audit data management integrated system successfully supported 14 functional departments of Zhejiang University of Finance and Economics, and completed 27 business systems, 59 tables, about 900, and 12 million audit data. Regular reporting of information.

As the first batch of innovation pilot projects in the digital reform in the field of education in Zhejiang Province, the integrated system of auditing data submitting the integrated system and the school's "New Generation Global Data Center" enriched the application of the school's digital cross -scenario, and further promoted the digital reform of the school. Provide reference to the digital transformation of the audit data management business of Zhejiang Province.

System bright spot

1. With "data center" as the system support. The integrated audit data management system built by the school is not essentially an independent new system. On the contrary, in order to avoid the disadvantages of a single system "chimney -type" management, the system rely on the school's global data center to make horizontal direction of the data central platform on the data center platform. Function expansion, construction data filling and data report module. On the basis of the full data governance results of the school, the system realizes the unified collection of school audit data, standardized storage, security storage, auditing verification, centralized reporting, and statistical analysis.

2. Promote "tube" with "use". The construction of the school's data level, while completing the establishment of the data base and management system, also enhances the user's perception of data construction results. In a sense, the audit data of the Department of Education provides a very typical data application scenario, which realizes the full -link closed -loop management of auditing data "collection -verification -standardization -storage -reporting". Through such a typical data application, it can promote the improvement of school data management capabilities. At the same time, the construction model can be used quickly to other application scenarios, and it has now covered application scenarios such as school ideology, party building, academic and engineering and other application scenarios. 3. Combination of services, digital empowerment. The system combines management with services. In addition to completing the data management work, it will further improve the service capabilities, use a unified visualization platform to conduct multi -dimensional, deep, and visual analysis of audit data; The correlation analysis of cross -departmental data can be "let the data tell the truth", timely grasp the overall situation of the school's business operation, find the weak links in the business and improve it, and use the digital reform of the digital enable audit business.

Challenges encountered in system promotion

In the advancement of the entire system construction, not only made large progress and breakthroughs, but also encountered some difficulties and challenges. There are mainly the following aspects:

1. Audit data belong to the confirmation of the department. The audit data of the Zhejiang Provincial Department of Education submitted the data of the 20 core business areas of the school, including personnel, education, assets, assets, academic work, one card, procurement, scientific research, finance, enrollment, employment, continuing education, campus management, campus traffic management , Dormitory management system. Confirm the belonging department and system of these data, including which data from which department, which system (or no management system for data online maintenance), offline data introduction and filling, who can conduct data and responsibility claims, etc. It is a very complicated job. It is necessary to set up a special class from the organizational structure to coordinate and coordinate the school's departments in order to effectively promote the work of this work.

2. Establishment of audit data standard models. The establishment of the audit data standard model. Under the existing "data center" system in the school, based on the existing school -level data standards, the standardized field conversion and mapping must be performed in accordance with the audit data standard specifications of the Education Department. Effective support for the manpower and platform tools of the technical department.

3. The horizontal expansion of the "data center" capacity. The integrated system of auditing data management is an incremental development and iterative upgrade based on the school's "data center". The newly constructed data filling module and the data report module need to be expanded in horizontal function expansion on the original "data center" capacity architecture. It is necessary to meet the requirements of auditing data management, but also maintain the consistency with the original data system. This puts forward a lot of challenges for the development and construction of the system, including product function compatibility design, data security design, and user permission system construction.

*Fund Project: This article is the research results of Zhejiang's philosophy and social science planning project "Innovation and Practical Research on Data Governance Models in Digital Reform" (Project Number: 22ndjc108yb); Research and Application of Data Center -based college big data audit inspection systems "(topic number: KT2021390) research results.

Author: Kong Linjun, Zou Zhixin (Information Technology Office of Zhejiang University of Finance and Economics)

Editor -in -chief: Chen Rong

- END -