Following the continuation of the previous article , I have finally come to the end of the data governance book. I have been busy during this period and have not updated this part in time. This article mainly describes some relatively mature data governance application practice cases and 6 major preparations for data governance Work, 6 major misunderstandings and 5 technical prospects , I hope it will be helpful and reference for everyone to carry out data governance work and digital transformation work. As emphasized many times before, data governance work is a practice-oriented, continuous iterative, and intricate procedural system engineering. The data governance project cycle is long, the time to effect is slow, and the value is reflected indirectly. It even needs to be done well. The series of work, the classic methodological practice compiled in this book, I think it is very useful for reference, so I sorted it out and shared it with everyone, so as to encourage everyone. Data governance, as the only way for enterprises to digitally transform, is the basis for enterprises to speak with data, manage with data, make decisions with data, and innovate with data in the future. It is also the cornerstone for enterprises to transform from business-driven or process-driven to data-driven in the future. Under the tide of digital transformation in the whole world, the development of enterprise data governance is not a multiple-choice question, but a compulsory question. If enterprises want to improve business efficiency and intelligentize management decisions in future competition, the premise must be data-driven. Therefore, in order to improve enterprise In order to improve market competitiveness and security compliance, enterprises have to carry out data governance work, and regard it as an important business work and normalized daily work of the enterprise, and focus on it.
1. Two typical enterprise data governance practice cases
Carrying out data governance work in an enterprise involves all aspects of the enterprise, including enterprise strategy, organizational mechanism, data standards, management norms, data culture, technical tools, etc., which is equivalent to a complex. Moreover, data governance is the basis for the development of enterprise data projects. Projects such as data warehouse, data middle platform, data asset management, and master data management are all inseparable from data governance. The following are two typical practice cases.
(1) Master data management practice of a wire and cable group company
Including enterprise introduction, construction background (there are problems such as lack of unified planning, different standards, low efficiency of horizontal and vertical collaboration, and difficulty in integration and sharing), master data census to find out the status quo (manpower, customers, material domain), master data management solutions and construction results. The latter two parts will be elaborated below.
1. Master data management solution
1) Management organization construction, as shown in the figure below, including organizational structure, centralized department and management and control mode (centralized or decentralized) 2) 
Master data standard construction, including classification standards, coding standards, naming standards and descriptions specification.
3) Master data cleaning plan, including cleaning scope, precautions for different types of master data cleaning, etc.
4) The single-source master data management solution achieves a single credible view within the group and identifies the source and destination of each master data.
5) The multi-source master data management solution, as shown in the figure below, requires a multi-source data collection interface. 
2. Project construction results
Through project construction, build a master data management platform, carry out master data governance, improve the quality of master data, and realize cost reduction and efficiency increase and management innovation within the enterprise. The specific results are as follows:
1) Promote business digital integration
2) Promote business management standardization
3) Promote business management refinement
4) Promote scientific management decision-making
(2) The data asset management system practice of an energy company
Including enterprise profile, construction background (data diversity, low reuse value density, difficulty in integration and sharing, etc.), enterprise data management status (data is complex and unclear, data quality is worrying, lack of data standards), data asset management solutions and construction results. The latter two parts will be elaborated below.
1. Data asset management solution
1) The overall blueprint of data asset management, see the figure below, including metadata management, data standard management and quality management 
2) Data management maturity assessment, find out the stage of the enterprise, and clarify the improvement content and path etc.
3) Data asset research and sorting, see the figure below, including value chain, business domain, determination of data domain, design of data model, establishment of data catalog and mapping of data sources, etc. 
4) The construction of data asset management system, including organizational structure, standard system, management methods and process construction, etc.
5) Construction of data asset management platform, including data standard management, metadata management and quality management.
2. Project construction results
Through project construction, build a data asset management platform, improve enterprise data management level and data asset use efficiency, focus on data standards, metadata and data quality to carry out application construction, and form a unified visual data asset management platform. Realize unified, automated, and open visual management of data assets. The construction results are as follows:
1) Established a data asset management system, formed a data management solution that organically combines organization, process, strategy, standards, security and technical support, and provided comprehensive supervision of enterprise information construction 2) Established a platform,
from Starting from the main value chain of production, supply and marketing, conduct a comprehensive inventory of enterprise data assets to form an asset map. Through the analysis of the influence of metadata, it supports the tracking and tracing of business indicators and provides support for management decisions. 3) Establish data standards and management systems and clarify
responsibilities Main body, unified planning and standards
4) Establish data quality rule definition, inspection, problem analysis, quality rectification full life cycle model, provide data production, exchange, storage, and control the quality monitoring of the whole link, and improve the level of control.
2. Six preparations that need to be done in advance for data governance
Data governance can not only establish an enterprise-level data consensus, let all employees in the enterprise realize the importance of data and its value to the enterprise, but also use data as assets, gradually revitalize enterprise data assets, and let stakeholders in the enterprise understand Find out what data assets the enterprise has, where they are, how they are used, how they are used, and so on. Therefore, if you want to do a good job in data governance, you need to meet the following conditions in advance:
1. Management's understanding of the value of data governance: decision-making, insight, data protection, security compliance.
2. Reasonably assess the status quo of enterprise data management
3. Select a leader or team for data governance
4. Integrate business and IT together to do
5. Select data governance tools, clarify the focus and urgent needs
6. Data governance consulting and implementation experts
3. Six major misunderstandings in data governance that should be understood early
Data governance, as the only way of enterprise digital transformation, is the integration of enterprise system integration and business collaboration, and is an important means to realize the integration and integration of enterprise business, strengthen group management and control and other management objectives. However, in the process of data governance work, it is easy to form the following misunderstandings:
1. Blind governance led by the technical department: clarify the reasons and core driving forces of governance, focus on the goal, don't be greedy for everything, and strictly control the scope and not be boundless.
2. Partial governance led by the business department: It is easy to cause local and narrow data governance, lacking the overall view of the overall situation.
3. Focus on project construction and neglect continuous operation: daily business needs to be continuously improved, and outsourcing is not necessarily good.
4. Tool-only theory: Is data governance a solution that integrates methods, standards, systems, processes, technologies, and tools? Both are indispensable.
5. Emphasis on results, not process: Governance is a process of building consensus. The process of speaking Mandarin and writing standardized characters.
6. Difficult to adapt due to multiple data sources: process-driven to achieve multi-source data collection and integration, or undertake data access to business systems through the public interface of Zhongtai Thinking.
4. Five technical prospects for future data governance
In the future, with the continuous evolution of data governance, the data field needs to rely on the integration of new IT technologies (big data, artificial intelligence, Internet of Things, blockchain, data lake, middle platform, digital twin, digital transformation, etc.) The improvement of the governance theory system, the development of the technical system and the accumulation of practical experience mainly include the following five technical prospects.
1. Is master data management still necessary in the context of big data: Master data is always the golden data of an enterprise, and it is a link between the past and the future. It will never die and is very important. In the future, the application of master data + artificial intelligence will be sublimated, such as AI automatically identifying master data, cleaning master data, and automatically marking data.
2. Under the big data, how to manage enterprise data: This article puts forward the six-character policy of acquisition, storage, management, viewing, finding and using. Looking at data includes data asset maps, data flow analysis, heat analysis, blood relationship analysis, quality problem analysis, etc. Finding data includes using knowledge graphs to extract entity relationship attributes from various data, etc., to achieve rapid data positioning and precise query. , and finally realize the true value of data when the data is visible, found, manageable and well used.
3. Under microservices, how to manage enterprise data: clarified what, where, and how to manage, including online data processing solutions and offline processing solutions.
4. Blockchain to help enterprise data asset management: the essence of blockchain is a decentralized distributed database, which can carry out the following tasks, based on blockchain data asset rights confirmation, protection of data asset security, and improvement of data quality , Accelerate data asset sharing, etc.
5. Artificial intelligence, plugging wings for enterprise data governance: artificial intelligence includes NLP, intelligent search, machine learning, knowledge acquisition, combination scheduling, pattern recognition, neural network, etc. In the process of data governance in the future, there are many places that need to rely on artificial intelligence, as follows:
1) In terms of data collection, various types of data are automatically collected through images, voice, and NLP, and machine learning technology is also used to collect data from historical data. Automatically discover structural patterns, relationships, entity attributes, etc.
2) In terms of modeling, through kg, machine learning, graph database and other technologies, realize text recognition of different types of data in enterprises, entity relationship recognition, and mine dark data in enterprises.
3) In terms of metadata management, use AI to better manage and integrate metadata.
4) In terms of master data management, AI is used to monitor data sets to make master data management automatic and efficient.
5) In terms of data standards, frequency and popularity are evaluated and optimized.
6) In terms of data governance and management, word frequency analysis, automatic record merging, outlier detection, automatic replacement and completion, and deletion processing, etc.
7) In terms of data security, real-time and dynamic identification, automatic generation and labeling of sensitive data, automatic classification and grading.
8) In terms of data analysis, realize automatic data cleaning, classification and marking, identification of data relationships, connection of related terms, etc.