This is a machine-translated version of the original paper which is available at http://www.infocomm-journal.com/bdr/article/2023/2096-0271/2096-0271-9-2-00016.shtml
At present, the theory and practice of authorized data operation are still chaotic and controversial. Based on the clarification of what open data means in the Chinese context, this paper challenges the current theory that authorized data operation is either complementary to open data or part of open data, and repositions authorized data operation that implements open data through a market-oriented delegation mechanism. This paper also presents the need to unpack the definitions of two key elements of authorized data operation: data product and data operation, and proposes a novel interpretation of authorized data operation as it is a social-technical system serving the purpose of open data and mainly producing primary data products for reuse rather than just use with support from a rich tiered data operation service ecosystem.
As a new type of production factor , the maximization of the value of data elements lies in the market-oriented allocation of data elements, so as to realize the free flow and extensive development and utilization of factors. In this context, public data - that is, "by state agencies, institutions, organizations authorized by law to manage Public Affairs, and organizations that provide public services such as water supply, electricity supply, gas supply, and public transportation in the performance of public management and Various types of data collected and generated in the process of service responsibilities" , its circulation and utilization in the process of marketization has played a key role in the socio-economic benefits of data elements.
As one of the important breakthrough points of public data circulation, Shanghai, Beijing and other places have taken the lead in exploring public data opening work since 2011. Although 208 local data opening portals have been released nationwide by October 2022 , from the perspective of the overall society In terms of perception, there are still problems of low data value, slow data update and poor data quality in data opening work . The root of the problem lies in two points: One is "can't" , that is, data opening work has a high degree of technical professionalism, and needs to deal with individual needs directly for data users, so traditional public function departments lack the corresponding technical capabilities and service capabilities to do data opening work well; On the other hand, it is "unwilling" , that is, because the fundamental data opening  requires the public welfare of data to be free, and in order to meet social needs and ensure data security and quality, real capital, manpower and material resources are needed, which makes data opening work "both horse running and horse running" Children don't eat grass ", caught in an endless cycle of Incentive Mechanism.
In response to the practical difficulties of the above-mentioned public data opening work, data authorization operation is proposed as a new type of system, trying to break down the obstacles of the implementation process of data opening and reactivate data opening. The Outline of the Fourteenth Five-Year Plan for National Economic and Social Development and the Outline of the 2035 Vision (hereinafter referred to as the "14th Five-Year Plan") first proposed "to carry out pilot operations of government data authorization and encourage third parties to deepen the mining and utilization of public data", and the "Opinions of the Central Committee of the Communist Party of China and the State Council on Building a Data Infrastructure System to Better Play the Role of Data Elements" (hereinafter referred to as the "Twenty Data Articles") further requires "promoting the implementation of the public data confirmation and authorization mechanism In terms of landing practice, the "Chengdu Model"  explored by Chengdu to uniformly authorize the operation data of municipal state-owned enterprise platform companies, and the "Financial Data Zone"  created by Beijing for industry data circulation applications have become typical cases of public data authorization operations, which have received extensive attention and research from academia and industry. In terms of supporting system construction, the Shanghai Data Regulations are the first local regulations to propose a sound data authorization operation system, while Chengdu has drafted and passed the Chengdu Public Data Operation Service Management Measures (but not publicly released) in the process of exploring the "Chengdu Model" in 2020, and Zhejiang Province publicly released the "Zhejiang Provincial Public Data Authorization Operation Management Interim Measures (exposure draft) " in the fourth quarter of 2022.
However, as an emerging system that is still being explored, there are still a lot of questions and controversies around its conceptual connotation, legal nature, operation mode, and regulatory mechanism. Previously, scholars have discussed practical cases [6, 7, 8], legal nature [9, 10, 11, 12, 13], and operation mechanism [14, 15, 16] from the perspective of public management and law under the social system. Some issues have been interpreted and responded, and some scholars have analyzed the practical cases of public data authorization operations from the perspective of technical systems from the perspective of FL technology .
In this paper, we believe that the key to building a data authorization operation system is to clarify why authorization operation (positioning) is needed and what authorization operation should do and what it is (connotation). By reviewing the current academic discussions and practices, we find that there is a common phenomenon, that is, we try to regard authorization operation and data opening as two different data circulation (development and utilization) systems, and believe that authorization operation is a new model of data circulation. In this regard, we believe that it will not only be detrimental to the continuation and implementation of the existing public Data Governance system and regulations for data openness, but will also bring about impacts and confusion in the cognition and practice of various stakeholders to a certain extent. Fundamentally, the key is to clearly recognize that data openness under China's national conditions is no longer a special "point" for the external circulation of data that follows fundamentalist openness, but should be built on the basis of unconditional openness and conditional openness. A broad-spectrum "line" on top of the classification model. Therefore, we believe that the authorized operation does not create a new circulation (open) model (How), but on the basis of retaining the existing data opening mechanism, we turn our attention to solving the problem of "who will implement the opening" (Who) to solve the problem of inefficiency of data opening caused by the "unwillingness or inability" of the original administrative force.
On the other hand, we also observe that in the current discussion of authorized operations, the function of authorized operations is often regarded as "processing and development" of "data-to-data products and services", while ignoring the multi-link work such as pre-planning, demand research, and even later operation and maintenance, promotion and trading. It can be fully or partially included in the operation activities to bring rich authorized operation possibilities. Therefore, we propose the need to refine and discuss the two concepts that are often discussed extensively, operational behavior and operational output, and make a preliminary refined exploration of the two concepts. Based on these two clarifying discussions, we hope that this paper will provide new theoretical support for further constructing the authorized operation system and carrying out the corresponding practice.
In the process of discussing data authorization operations, a core issue that cannot be avoided is how data authorization operations are positioned in the framework of data element circulation and market-oriented configuration, and what is the relationship between it and data openness? This section will sort out the existing theories, further clarify and unify the definition of data openness, and finally build our views.
In the current academic literature and policy discussions, the main idea is to try to place data authorization operation and data opening in the same logical system, and to explore the logical subordination relationship between the two. Under this line of thought, two mainstream theories have been born: the parallel complementarity theory and the inclusion subordination theory . The core point of the parallel complementarity theory [6, 8, 10, 13, 14, 16] is that data authorization operation, data opening and data sharing generally constitute three paths of data circulation, and the authorization operation is regarded as a kind of data opening. A supplementary mechanism; the inclusion theory [12, 15, 17] believes that data authorization operation should be regarded as a part of the data opening system and a new form of data opening.
Scholars who hold the theory of parallel complementarity expound the positioning of authorized operations from the macro-strategic perspective of market-oriented allocation of data elements. Zhang Huiping believes that "government data entering the market is mainly achieved through two paths: data opening and data authorized operations"  based on his research on the "Chengdu Model" , Lu Zhipeng  further pointed out that "public data authorized operations are a market-oriented supplement to public data opening. (Authorized operations) introduce market participants to participate, expanding the space and depth of public data development and utilization." This is further expanded in Xiao Weibing's argument as "government data authorization operation is a supplementary mechanism to make up for the defects of government data opening". In general, the above scholars emphasize that data opening is a public (public welfare) circulation, while the authorized operation is a market-oriented circulation, so the two coexist and complement each other under the data circulation framework. Under this argument, Lu Zhipeng, Zhang Huiping and others all pointed out that it is necessary to balance the relationship between data opening and authorized operation, so as to avoid authorized operation from overly impacting data opening and bringing new obstacles to it.
Scholars who construct the theory of affiliation base their arguments on a basic fact, that is, the national "14th Five-Year Plan" proposes to "carry out the pilot operation of government data authorization" in the section of "Strengthening the opening and sharing of public data" [15, 17], so authorized operation is a special form of data opening. Song Shuo  further pointed out in his paper that "in the" 14th Five-Year Plan to Promote National Government Informatization "and the" Overall Plan for the Comprehensive Reform of the Market-oriented Allocation of Factors ", the authorized operation of government data is also listed as a new measure to improve the data opening and sharing mechanism." Therefore, the authorized operation of data should belong to the scope of the data opening system, which is "a form of government data opening that is standardized in accordance with the laws on the management and utilization of administrative public property, implements market-oriented operation of data as state-owned assets, and has multiple goals of industrial promotion and public welfare maintenance" .
At the micro level, both schools of view point out the difference between authorized operations and "existing mechanisms". The difference is that the "existing mechanisms" they refer to are different: the parallel complementarity theory [8, 10, 16] is data openness, while the inclusion subordination theory  is conditional openness . In Table 1, we summarize the main viewpoints of the two schools, and if we compare the differences between the two schools of thought, it is not difficult to find that the concept of data openness referred to by the parallel complementarity theory obviously does not include conditional Open and only follow the fundamental definition of strict openness ; while the inclusion theory is built on the concept of data opening that includes both unconditional opening and conditional opening. Obviously, the definition of data openness varies in different discussions, so in order to further clarify the views of the school, we realize that the key is to first clarify what is the definition of "data openness" that should be followed under China's national conditions, that is, to unify and compare data authorization operations. On this basis, we can draw a reasonable conclusion on the relationship between data authorization operations and data openness.
Table 1 Summary of the relationship between two mainstream authorized operations and data opening
Explanation of the positioning of authorized operations
Main differences with reference objects
Parallel complementarity theory
And data openness and data sharing belong to data flow;
It is a complementary mechanism for data opening
(1) The difference between supply entities, data opening is provided by public departments and service agencies, while authorized operations are provided by authorized parties;
(2) Differentiation of supply objects, data openness provides data itself, and authorized operations only provide data products and services;
(3) The supply and income are different, the data is open and free, and the authorized operation has a profit-making purpose.
Inclusive subordination theory
Is part of open data;
Is a new form of open data
(1) Distinction between supply objects, authorized operations provide data products and services to the society, while conditionally opening the original data sources provided
(2) The difference in the number of participants, the authorized operation involves three parties, while the conditional opening only involves two parties, and there is no intermediate link
(3) Object distinction, authorized operation is only for limited and qualified subjects, while conditional is for a wider range
(4) Differences in charging mechanisms, conditional opening to retain the right to charge for cost compensation, but authorized operations can adopt pricing charges with a more market-oriented mechanism
The international consensus of data openness emphasizes the openness of fundamentals , that is, (1) technical openness requirements: data should be machine-readable, the data format should not use a dedicated closed format, and the data content should maintain the originality of the data; (2) Legal openness requirements: that is, the use of open licenses to ensure that data can be obtained and used regardless of identity and purpose, and that data can be freely reused and shared; (3) Social openness requirements: that is, free access, use and sharing.
However, we believe that in the process of more than ten years of research and practice of "data openness", a Western import, its conceptual connotation has undergone qualitative changes under the influence of landing practice, and has evolved from a rigorous fundamentalist openness to a relaxed pan-data circulation concept . Through the Internet, we reviewed the regulations and policies issued by some major provinces and cities (a total of 10 provinces and cities, including 3 provinces, 7 municipalities directly under the central government or provincial capitals), and summarized and compared how they define "data openness" - whether it is government data openness or public data openness (see Table 2), trying to clarify the accurate definition of local data openness through this.
Table 2 Definition of "open data" in some provinces and cities
(Note: In each definition statement, the agreed data characteristics are marked by underscores; regulations and policies are sorted by year)
Year of introduction
Whether to consider
Whether to agree
Guiyang Municipal Government Data Sharing and Opening Regulations
The opening of government data as mentioned in these Regulations refers to the act of administrative organs providing government data to citizens, legal persons and other organizations
Interim Measures of Shanghai Municipality on Public Data Opening
The term public data opening as mentioned in these Measures refers to public management and service agencies providing public services with original, machine-readable, and socially reusable data sets within the scope of public data
Shenyang Municipal Government Affairs Data Resources Sharing and Opening Regulations
The opening of government data resources as mentioned in these Regulations refers to the act of government departments opening government data resources to citizens, legal persons and other organizations
Guizhou Provincial Government Data Sharing and Opening Regulations
The opening of government data as mentioned in these Regulations refers to the act of administrative organs providing government data to citizens, legal persons and other organizations in accordance with the law
Qingdao Municipal Measures for the Administration of Public Data Openness
The term public data opening as mentioned in these Measures refers to the public service of data opening subjects providing original, machine-readable, and socially reusable data sets to the society.
Shenzhen Special Economic Zone Data Regulations
The term public data opening as mentioned in these Regulations refers to the activities of public management and service agencies to provide public data that can be read by machines to the society through the Public Data Open Platform
Measures of Wuhan Municipality for the Administration of Public Data Resources
The opening of public data resources as mentioned in these Measures refers to the behavior of government departments and public enterprises and institutions providing public data resources to the society in accordance with the law.
Zhejiang Public Data Regulations
The term public data opening as mentioned in these Regulations refers to the public service behavior of natural persons, legal persons or unincorporated organizations that provide public data in accordance with the law
Interim Measures for the Opening of Public Data in Guangdong Province
Public data openness refers to the public services of public management and service institutions providing public data to the society;
Chongqing Data Regulations
The term public data opening as mentioned in these Regulations refers to the public service behavior of natural persons, legal persons or unincorporated organizations that provide public data in accordance with the law
In view of the above different definitions, if we put aside the differences in the agreed data characteristics in different definitions for the time being, and only focus on the common points of each version of the definition, it is not difficult for us to come to a conclusion, that is, the current consensus in China is The most liberal (or most basic) data opening can be defined as "data opening is an act of providing data to society", that is, its essence is the external circulation of data (In order to facilitate the distinction between data opening with different definitions, we have constructed Table 3, and will refer to different data opening based on this later) .
Table 3 Definitions of different levels of "strict-relaxed" data openness
Whether it reflects openness 
And unconditional open relationship
And conditional open relationships
Open data is the act of making machine-readable and raw data available to anyone, regardless of destination, and ensuring free reuse and sharing through licensing agreements
Does not contain
Open data is an act of providing data to society, and its provision methods, objects, data characteristics and behavioral attributes all follow certain guidelines and requirements
Not sure, may reflect part
Open data is the act of providing data to society
Furthermore, since the "Interim Measures for Public Data Opening of Shanghai Municipality" promulgated in 2019, all the above-mentioned policy documents have introduced "unconditional opening (or general opening), conditional opening (or restricted opening/open upon application), and prohibited opening" under the framework of data opening. Three categories of classification systems. And following this system, all localities have also provided data in the data opening portal according to unconditional and conditional distinctions . The introduction of conditional openness substantially makes data openness in the Chinese context break through the openness constraints of fundamentalism, and extends the upper limit of data circulation behavior covered to cover the full spectrum of data external circulation (that is, loosely defined data openness) behavior ( see, Figure 1, the European and American models vs. the Chinese model from the perspective of data circulation ) .
Figure 1 European and American models v.s. China model from the perspective of data flow
The European and American models are constructed with the data spectrum  of the United Kingdom Open Data Research Institute as an example. From the figure, it can be seen that China's data openness is equivalent to covering a period of circulation behavior covered by data openness (strictly defined) and external data sharing in the European and American models; China's conditional data openness is essentially equivalent to the circulation behavior of external data sharing under the European and American models.
In this process, because all stakeholders have a certain awe and compliance with the original strict definition of data openness, on the other hand, they need to consider practical needs and break through the openness constraints, so there is a "middle definition" of data openness in current practice, and the flexible space of its interpretation depends on the trade-off ratio of strict openness constraints. This is also reflected in the current policy statement (see Table 2 for a summary of the policy): At present, only Shanghai, Qingdao, and Shenzhen have mentioned the characteristics of the data provided in the data opening policy, and have not yet reached an agreement. Both Shanghai and Qingdao propose that data should have the requirement of "originality and machine-readable", while Shenzhen only requires "machine-readable" and no longer emphasizes "originality". Another feature that is often overlooked in the discussion is the characteristic of "for social reuse" mentioned in parallel in the Shanghai and Qingdao policies, that is, it is emphasized that the service purpose of data opening should be directed to reuse (Reuse in English, that is, it is separated from the original purpose of data collection and can be used for new needs through development and analysis) rather than simple direct use (Use in English, that is, directly digesting the information carried by the data through reading, inspection, etc.).
We believe that the existence of the "intermediate definition" is a product of a stage. The flexible space of its interpretation is affected by the awareness of openness of various stakeholders, especially public data management agencies, the blessing of new technologies such as Privacy Computing on the security of data circulation, and the need for data openness in the context of the data element market. The need for change in the economic model of data openness and other factors. Over time, Data openness in the Chinese context will eventually generalize into external data flow behavior (that is, loosely defined), which will evolve into a "line" that includes all external data flow patterns that conform to and do not conform to the fundamentalist openness, rather than as a single "point" that emphasizes openness . It and data sharing in the Chinese context (internal data flow) form a binary classification, collectively called data flow (see Figure 1).
Based on the combing of Sections 2.1 and 2.2, we believe that the current logical affiliation between data authorization operation and data opening is wrongly constructed on the one-sided and selective understanding of data opening. On the premise of respecting realistic practices, data opening in the Chinese context should no longer be equated with strict data opening, but should be regarded as external data circulation behavior (loosely defined). Therefore, authorized operation and data opening are neither parallel, complementary nor include subordination, and do not constitute a logical affiliation .
In Section 2.1, we summarize the views and references of the current two schools of thought, and when we return to the interpretation of data openness in the Chinese context, we find that the logical subordination relationship of the current two schools of thought cannot be established (see Table 4). Therefore, we believe that the two should not actually construct a logical subordination relationship.
Table 4 Problems existing in the current data authorization operation and data open relationship theory
Problems that exist
According to Section 2.2, China's data opening is no longer strictly open, so the parallel complementarity theory based on strict opening as a reference cannot be reasonably established.
Mistakenly applying "conditional opening" to strict openness, thus emphasizing "originality" and "free", but in fact, conditional opening in current policies and practices does not emphasize data originality, and the way data is supplied is not limited to data download (original provision). There are already rich APIs, sandboxes, special areas and other paths, and charging is not prohibited (but the inability of administrative agencies to charge fees institutionally is another topic), so it cannot be established.
Based on the appeal's review of the current relationship theory and the definition of data opening under China's national conditions, we further draw a set of new relationship theories between authorized operations and data opening:
(1) Authorized operations should be regarded as a mechanism for indirectly implementing data opening, which emphasizes the participation of market-oriented entities
We believe that data authorization operations and data openness essentially solve two different types of problems, the former solves "Who" and the latter solves "How". That is, data opening is to solve how (How) to provide data from the data owner to the data user ( e.g. whether to provide the original data source, whether to use technical means such as data out of the field, available and unavailable, whether to establish access qualifications, etc. are all How part ), and the data authorization operation is to solve who (Who) specifically implements data opening to achieve data flow ( that is, to answer the public institution itself or a third party to implement ) . In order to facilitate understanding, we use a popular scene analogy here to explain: the same is cooking, the ingredients can be made into dishes by frying, frying, roasting, frying and other different ways, which is a How problem; and the choice is to do it yourself, Or hire a third-party professional to do it, it is a Who problem.
As mentioned in the introduction, the current dilemma of data opening comes from the "cannot but be unwilling" of the implementing body itself. The problem lies in who implements it, not the specific circulation mode of existing data opening. The idea of authorized operation is to use authorization to allow market-oriented entities with the ability and willingness to replace the original administrative entities to implement data opening. This point has also been echoed in the discussions of Changjiang and Zhang Zhen  on the legal nature of authorized operation: "... Public data authorized operation belongs to the'public service concession 'in the context of concession. That is, in the provision of public services, due to the constraints of objective reasons, the government cannot directly provide perfect services to the society, but introduces a third party in the society, and transfers all or part of the services that should normally be responsible by it to the third party in the form of a contract (Of course, the third party is not a narrow single subject). " Therefore, as a framework system for the external circulation of public data, public data opening should be divided into two types: direct implementation of opening and indirect implementation of opening from the perspective of solving who opens (Who). The department itself carries out administrative implementation, and indirect implementation of opening is to carry out market-oriented implementation through authorized operations . Indirect implementation of opening is also an effective means to separate public data production and supply, enhance multi-subject participation, and enhance market-oriented configuration . Its characteristics and advantages lie in the introduction of market-oriented professional forces to do professional affairs. It is worth pointing out that the inclusion of affiliation theory can be interpreted as a new form of data opening, which is a logical relationship based on how to open (How). Here, based on who opens (Who), it is divided into direct implementation and indirect implementation, which should be understood as authorized operation is a new form of implementation of data openness rather than a new form of openness. The classification logic of the implementation method should not be equivalent to that authorized operation and data opening constitute a membership relationship.
(2) Authorized operation is the realization path of "conditional opening" from "passive service" to "active service"
We believe that the core service object of authorized operations is "conditional opening", but this does not mean that authorized operations only involve operations with conditional open data If public institutions really think it is necessary, third-party market forces can also be introduced to operate unconditional open data (emphasizing the processing of original data sources into technical data products such as APIs), or data that is unconditionally and conditionally integrated. At the same time, it should be pointed out that both the concepts of authorized operation and conditional opening involve "authorization" (for example, "providing conditionally open public data resources to authorized service providers" is mentioned in ), but the two concepts have different meanings: Conditional open authorization refers to the public data management that has authorized qualifications to grant "permission and rights to access and use data" to data users that meet the access conditions. Generally, it stipulates data utilization, reuse, and redistribution rules through, for example, a Licensing Agreement. In authorized operations, it should mean that the public data management party grants a market-oriented third party "to implement accordingly." Rights and responsibilities of data opening work ". That is, the objects authorized by the two are authorized with different rights.
On the other hand, at present, conditional opening is used as an "application" mechanism, and we believe that it is in a passive service mode, that is, waiting for the data user in demand to contact the public department to obtain the right to access and utilize the data. In this model, it is easy to fall into a vicious circle of "chicken or egg first", that is, without sufficient applications and demand, public institutions will not be better able to supply data through technical products; without better technical supply, the market will naturally There will be no incentive to apply for data (because data is very troublesome to use). Therefore, we believe that the introduction of market-oriented capabilities by authorized operations can allow operators to actively transform data into technical products or services through market-oriented self-driving force, and actively supply them to the society in a mature way, which can turn passive into active and accelerate data. The user's early adopters of data, and then improve the quality of data supply through demand .
In Section 2, we clarify the positioning of data authorization operation, which is the mechanism for indirectly implementing data opening. In this section, we try to further sort out the connotation of data authorization operation. In order to answer this question, we will first sort out and summarize the definition of data authorization operation in current policies and regulations and academic papers, and based on this, we will further finely disassemble and interpret the key elements of authorized operation to clarify its connotation.
The current data authorization operation is defined by different perspectives, different understandings, and different granularities in different policies and regulations, as well as in the literature discussed by the academic community (see Table 5). In our summary, we cover 1 national policy, 2 provincial policies and regulations, 1 local technical standard, and 5 academic papers, all of which give a relatively complete definition of data authorization operation. In their Horizontal comparison, we analyzed and compared the definition elements:
Authorized subject : As for who has the right to authorize third parties, it is basically divided into two types of definitions. One is the coarser granularity defined as "government" but there is no clear institutional direction, and the second is finer granularity, which is also relatively mainstream, that is, the competent department of public data is determined as the authorized subject.
Authorization object : who can be authorized, the current basic definition is divided into four categories. The first category is that the rough granularity is defined as "society (subject) ", or it is "legal person or unincorporated organization" according to the type of subject; the second category is to further add binding conditions on the basis of the first category, trying to explain the subject qualifications, but the current specific conditions are relatively broad and vague, such as "specific", "meeting the prescribed conditions", "trusted market entities", etc.; the third category is to further refine the direction of the constraints relative to the second category, such as "institutions with professional operating capabilities", "meeting the prescribed security conditions", etc.; the fourth category is directly pointing to a specific category of conditions The main body, such as "state-owned enterprises" .
Authorized object : What is the content object authorized by the authorized subject, in the current definition version, there are three different methods. The first type emphasizes the type attributes of data, that is, it is "government data" or "public data" or "conditional open data", which is the most common definition; the second type emphasizes the value attributes of data, thus defining it as "data that is closely related to people's livelihood, has urgent social needs, and has significant commercial value-added potential"; The third type jumps out of the framework of data resources as an authorized object, and believes that the authorized object should be a data-related right, which can be "data utilization right"  or "market operation right" .
Operational behavior : As for the operational activities that the authorized entity should carry out, the basic definitions of each version point to "development and utilization" and its extended series of activities, including processing, value mining, development and formation of products, etc. But in addition, only a few definitions will further extend to propose "market-oriented service methods... to meet demand"  and "organize upstream and downstream related institutions of the industrial chain... (development and utilization) "  and other behaviors. This extension is to jump out of the single framework of "development and utilization", and the second is to propose the function of non-authorized subjects but industrial chain coordination, which is worthy of further discussion.
Operational outputs : As for the specific outputs after operational activities, there are currently two different ways to define them: the first category, the most common and more specific, is generally "data products or services"; the second category is relatively more abstract, such as "promoting the use of external subjects" , "meeting the needs of business innovation and public services" , etc., relative to its orientation is not limited to data products or services, and its connotation imagination is larger but relatively vague.
Other relevant elements : In our review of the Horizontal comparison of definitions, we found that individual definitions raised more critical but currently not generally mentioned elements. First, with regard to the definition of "operating income", that is, whether the definition clearly states the compensability of authorized operations in providing services to the outside world. At present, the local standards of Taizhou, Zhejiang and Xiao Weibing  have clearly included this point in the definition; Second, regarding "platform constraints", that is, whether the authorized operation as a whole must be constrained to be carried out on a specific technology platform, at present, only the China Software Evaluation Center  has clearly included the statement "on the basis of building a secure and controllable Development Environment" in the definition. Although Zhejiang Province has not made it clear in its definition, its management measures that are soliciting opinions clearly put forward the concept of "authorized operation domain" and require specific operation activities to be carried out in this domain.
Table 5 Definitions of data authorization operations in policies, regulations and academic literature
(Note: The definition elements have been marked with underscores in the original definition text, and the corresponding element order is given a letter label)
[please note the table is too large to display here; please visit the original publication site for the table content]
Based on the comparison of appeals, we believe that in the current different definitions of data authorization operations, a certain consensus has been basically reached on the subject, object and object of authorization. The main difference lies in the rigor of expression and the determination of the scope. The further determination of these three elements depends on the further clarification of the following two elements:
First, operational behavior . The current definition of it is relatively simple, too prominent and limited to "development and utilization". In fact, as we discussed in 2.2, authorized operation is an indirect path to implement data opening, which should cover all or part of the behavior of data opening. That is, it should potentially include non-technical work such as demand research, open planning, marketing activities in data opening, as well as technical work such as Data Governance, product development, and technical maintenance. Therefore, we believe that further refinement and clarification of operational behavior is one of the keys to clarifying the definition of data authorized operation.
The second is operational output . The description of it in the current definition of each version is still vague. The common one is to refer to "data products and services", but the definition of the concept itself is not clear. If it is an end point-oriented applied product, it runs counter to the goal of data opening itself. Although the directions of abstract goals such as "promoting the use of external entities" and "meeting the needs of business innovation and public services" are more in line with the original purpose of data opening, they also need to be further clarified for concrete outputs. Therefore, we believe that it is necessary to further sort out operational outputs.
Given the length of this article, we naturally cannot do an in-depth and detailed study of operational behavior or operational outputs. But we will try to reveal in the following sections: Even with a preliminary refined interpretation of these two elements, we can see the limitations in the current discussion of authorized operations. Therefore, it is expected to pave the way for further in-depth refined research on operational outputs and operational behaviors in the future.
Let's start with the definition of operational outputs. In order to answer this question, we think that the corresponding is to clarify what exactly "data products and services" (abbreviated as data products) are?
In fact, although the term data product has repeatedly appeared in various policy documents and local regulations, its definition is still vague and lacking unity, and the academic community has not reached a consensus on its precise definition  . From the macro perspective of data science, data products can be broadly defined as "deliverables derived from data or supported and driven by data, which can be in the form of some kind of discovery, prediction, service, decision-making, model, paradigm, system, etc. The final value of data products will be reflected in knowledge, intelligence and decision-making" , correspondingly, in the "Interim Administrative Measures for the Development and Utilization of Public Data Products in Hainan Province", public data products are correspondingly defined as "The public data products referred to in these Measures refer to the processing, analysis and research of public data resources or integrated social data resources, which can give full play to the value of data Products, including data models, Data Analysis reports, data lake visualization, data indices, data engines, data services, etc. ". From the perspective of data transaction circulation, Pei J  believes that data products are a measurable and tradable data set. From the perspective of the" White Paper on Standardization of Data Product Transactions ", it is more broadly believed that" data products are a commodity with definable property rights and tradable, and are the main trading objects and targets of the data element market ". It further classifies data products, namely primary data products and advanced data products. Primary data products include data API (Application Programming Interface), data Cloud as a Service, technical support, offline data packets, etc.; advanced data products include solutions such as visualized Data Analysis reports, data application systems and software for specific business scenarios, and integration with cloud Various Big data technology products, etc. Considering the length and purpose of this article, we do not intend to derive an accurate data product definition by ourselves. Rather, based on the above-mentioned limited literature, the definition and classification system of primary data products and advanced data products are selected as a staged work framework for our discussion. We believe that the binary classification of primary and advanced responds to the dual purpose of the use and reuse of open data to a certain extent (see the discussion in 2.2), that is, primary data products serve the purpose of reuse (for further processing and development by third parties), while advanced data products serve the purpose of use (for third parties to directly digest information and assist decision-making). To make a popular analogy and treat data as vegetables (such as potatoes), primary data products can be roughly processed potatoes such as shredded potatoes and potato chips, which are mainly used by third parties for further processing into dishes. While advanced data products are finished potato foods such as French fries, shredded cold potatoes, and mashed potatoes made directly from potatoes, which are mainly for direct consumption by consumers.
In addition to the above theoretical discussion, we try to identify specific products of data products from practice. We have summarized some authorized operation cases from the literature and sorted out their output content (see Table 6). It should be noted here that we apply the above-mentioned binary classification of primary and advanced data products to the classification of output types; and when we classify service objects, in order to avoid confusion of terminology and unclear meaning, we refer to Wang Weiling  The role classification in the authorized operation ecological chain described by , that is, it consists of data producers, data operators, data analysts, data consumers, data authorizers and data regulators.
Table 6 Operational outputs in some typical authorized operating cases
Type of output
Air travel vertical and horizontal [7, 14]
Entrusted by Zhonghang Square letter, the company develops and utilizes civil aviation operation and passenger air travel data to form professional products or services.
Advanced Data Products: Apps
State-owned enterprise intelligent review 
Chengdu authorized Chengdu Big data Group to operate government data. By establishing a trusted Appcloud (secure multi-party computing and data sandbox service), the smart review company can access the service through the computing node, and obtain the model under the premise that the data is available and invisible. Output results.
Primary Data Products: Secure Multi-Party Computing and Data Sandbox Services
Financial District 
Beijing authorized Beijing Financial Holding Group to establish a financial public data zone to provide joint modeling, information query and interface call services to Financial Institutions through the SaaS service platform
Primary data products: services such as joint modeling, information query and interface call
Qingdao Metro App 
Qingdao Metro App is authorized and operated by Chengjiao Big Data (Qingdao) Co., Ltd., and is committed to creating a comprehensive App product that helps the development of smart cities
Advanced Data Products: Apps
Through the analysis of the above cases, it is not difficult to see that At present, there are basically two "product type-product object" combination modes of "primary data product-Data Analyzer" and "advanced data product-Data Consumer" , that is, for end point data consumers, what they need is advanced data products or services that exist in the form of App applications, visualizations, reports, etc. to transform data into information again, so as to facilitate them to make corresponding decisions and actions; and for Data Analyzers, what they need is Data products should be used as a means of production so that they can easily Further development and utilization and mining value, processing into the final advanced data products to serve data consumers.
So what kind of data authorization operation is expected to produce? Based on our positioning of authorized operation, we believe that since authorized operation is a mechanism for indirectly realizing data opening, it should meet both the external use (Use, that is, direct application) and reuse (Reuse, that is, development and utilization) of data after data opening.), but "reuse" should be the main purpose, thereby activating the data element market. This also means that, Authorized operations should mainly produce "intermediate" or primary data products or services (For example, Lu Zhipeng  proposed that one of such intermediate products is a data element); Under specific circumstances or market demand, such as lack of willingness or space for market reuse, authorized operators can also directly produce advanced data products serving data consumers, but should also take into account the provision of primary data products to meet potential demand For example, the construction of my country's transportation market is mainly based on a state-owned system. From the perspective of providing transportation service information and building intelligent transportation MaaS services, there may be less space for third-party intervention in the market. Therefore, from the perspective of efficiently developing and utilizing traffic data and releasing value as soon as possible, it should be the main responsibility of authorized operators to directly develop the applications required by end point consumers. At the same time, however, it should not be ruled out that third parties may need traffic data in the transportation field or non-transportation fields (such as urban planning, commercial site selection), so both the development and supply of primary consumer products should be taken into account to meet the needs of specific data reuse.
The operation behavior of authorized operation should be sorted out based on two points. One is based on the operation goal, that is, it is clear that authorized operation serves data opening, and it is to introduce market-oriented forces to realize data opening, so its goal is naturally to successfully open data., release the value of data; the second is based on operational output, that is, operational activities should be mainly carried out around outputs, and ultimately achieve operational goals through the operation of outputs.
In Section 3.2, we make it clear that the main output of authorized operations should be primary data products, that is, intermediate data products, and special cases are direct production of high-level data products. Therefore, an indispensable operational behavior is naturally the "development and utilization" of data, which is also in line with the consensus of most definitions in Section 3.1. As we have emphasized earlier, operational behavior should not be limited to development and utilization, but should be defined around the life cycle of data products. Therefore, we propose a new concept here, namely the operational behavior chain, which responds to different operational needs at various stages by providing different operational behaviors according to the life cycle of data products. Here we construct a preliminary example of the operational behavior chain in Table 7 , but please note that, as we have mentioned earlier, due to space and purpose limitations, we cannot further in-depth, scientific and detailed discussion of this concept. This will be left to follow-up research, and this example is only used here to illustrate the complexity of operational behavior. .
Table 7 Examples of operational behavior chains
Data product lifecycle
Management and protection
Operational behavior (example)
Management of operations;
Deployment of products;
Tracking of results;
In the above example, we simply divide the data product lifecycle into four links: planning, development, service and management. The "development" link is the core link of data products, which extracts the value of single data or multiple types of data through the process of Data Governance and mining, and finally develops and tests to form data products or services that can be supplied and traded externally. And its front, that is, the " planning" link , which investigates the needs of data products. It should be pointed out that the demand research for different types of data products has different degrees of complexity. If the target is advanced data products, the demand research is more based on the end point. It is derived from the research of consumers, but if the target is a primary data product, it means that a common intermediate product demand needs to be summarized for different advanced data Product Research & Development needs of Data Analyzers. Any data products and services do not stop at development and deployment, but further need to go to the market, which is the "service" link . In this link, the necessary marketing activities need to be carried out, the Demand side needs to complete the delivery of products or services, and provide necessary technical support. In this process, if any form of access control needs to be given to the Demand side for data products and services, corresponding access control working mechanisms need to be matched. For example, if the application mechanism is set, there should be audit; for example, if the limit is set, there should be technical audit and dosage control. In the final "management and protection" link , there are both technical work such as product iteration, and non-technical work such as understanding customer feedback, tracking and mastering the social and economic benefits of customers using data products, etc., so as to facilitate the authorized subject to understand the corresponding data. The overall benefit of the entire chain process from authorization to market entry.
We believe that the current discussion of data authorization operations often has a one-sided understanding of its connotation. Based on our sorting out the current multi-version and multi-perspective definition of authorized operations, groundbreaking refined discussions on operational outputs and operational behaviors, and clarification of the positioning of authorized operations, we draw the following conclusions:
Authorized operation is a mechanism for indirectly implementing data opening, and its authorized object should be rights rather than data
Among the current definitions, we note that most of the definitions are based on the logical affiliation between authorized operations and data opening, so the authorized object is naturally regarded as data, which leads to the authorization of public resources themselves. Discussions on rationality, profitability, etc. If we accurately grasp the positioning that authorized operation is a mechanism for indirectly implementing data opening, then its The accurate authorization object should be to grant the authorized subject all or part of the "rights-responsibilities-interests" of the authorized subject to perform or implement data opening . Therefore, it may be more reasonable and appropriate to use "market-oriented operation right"  as the object of authorization in many current definitions , but it also needs to further clarify the scope according to the specific operation behavior required by the introduction of third parties. Under the grant and transfer of "rights-responsibilities-benefits", the corresponding data resources are granted to authorized operators to access and use as accessories.
Further development, The granting of this right is accompanied by the granting and allocation of corresponding rights and interests, and is in line with the "separation of rights" proposed in the "Data 20": that is, the right to hold data resources, the right to use data processing, and the right to operate data products Correspondingly, in the process of authorized operation, the authorized party should not actually hold or own the data, that is, the holding right of data resources should not be transferred and authorized. The authorized party should be accompanied by the acquisition of the right to operate, and correspondingly enjoy the right to process and utilize data, thereby recognizing and ensuring that its development investment in data products or services enjoys corresponding revenue rights. Further, regarding the operation work carried out by the authorized party for the processed data products or services (that is, the non-development stage work such as planning, service and management in the corresponding operation behavior), including but not limited to marketing activities, transaction delivery, technical support, etc., it has the corresponding right to operate data products, so that it can further enjoy reasonable returns on the investment in this stage. In the process of authorized operation, the authorized party's authorization expires and withdraws or should be breached during the authorization period, etc., resulting in changes in the authorized subject, or the authorization of operation work is segmented (for example, A is responsible for the processing and development of the original data source to the data product, B is responsible for the processing of the data product). In the case of the operation of the processed data product), the above-mentioned separation of ownership can also ensure the reasonable attribution of the income.
Authorized operation involves a complex chain of operational behavior and requires the support of a classified and hierarchical operator ecology
As we discussed in Section 3.3, Operational behavior is not a single development behavior. It follows the life cycle of data products to form a complex chain of behaviors, and complex operational behavior chains naturally correspond to different types of service providers . This is also in line with the concept of vigorously developing the data element circulation and transaction service ecosystem advocated by the "Data Twenty", that is, expanding the "intermediary industry". This can include product developers who specialize in providing development services, data intermediaries who specialize in providing data products and services marketing activities and dealmaking, product operation and maintenance providers who specialize in providing customer-facing technical support for data products, Client Server, etc. Of course, for an authorized operator, it can only have a single behavioral function, or it can shoulder multiple behavioral functions (ie, complex operators).
Further, according to the summary of current academic discussions, there are three modes of authorization construction between the licensor and the licensee in practice : industry-leading mode, regional integration mode and scenario traction mode. The three can actually be further merged into two types of authorization modes: comprehensive authorization or scenario authorization. Scenario authorization authorizes specific operators around the specific application requirements of a specific scenario (ie, advanced data products), while comprehensive authorization is in principle a vertical field (ie, industry dominance) or a region (ie, regional integration). Various types of data resources within the jurisdiction are classified into the scope of operating rights granted to third parties.
Under the comprehensive authorization model, we believe that authorized operators should also establish a hierarchical management system based on the classification of operational functions and behaviors . One possible form is that the authorized party directly authorizes and manages a composite operator, which is similar to the system integrator in an informatization project. In terms of business capabilities, it should have strong planning capabilities, that is, it is good at extensively investigating social needs and discovering common primary data product requirements. On the basis of demand sorting, it can further cooperate with the authorized party on Data Governance and do a good job in the development and operation planning of data products. At the same time, the operator (the licensee) should do a good job in the management and coordination of the secondary operator, that is, according to the product planning plan reached with the licensee, organize, coordinate and manage the secondary operator to complete the development of corresponding data products, and to the market. Further, the composite operator should do a good job in sorting out the final utility output evaluation and user feedback, and gather with the authorized party to carry out targeted Data Governance and corresponding data product planning iterations based on the above evaluation and feedback summary. Work. Of course, we believe that there should be other possible forms under the hierarchical system, which need to be further discussed in combination with the current governance and jurisdiction of the data.
The main operational output of licensed operations is primary data products, which are "reused" for the purpose of enabling third-party analysts and developers to further develop advanced data products. In Section 3.1, we point out that some of the definitions of authorized operations clearly involve the concept of a unified technology base such as a public authorized operation platform or authorized operation domain, and combined with the fact that primary data products are the main operational outputs of authorized operations, We further believe that this is a figuration of the concept of "city as platform": that is, public institutions build and govern urban public data bases, and their original data sources are transformed into primary data products through direct operations or authorized operations (that is, they constitute the external interface on the base), and finally The advanced data products born in the city are plugged in through these interfaces like building blocks Above the base of the city.
We believe that the authorized operation from the perspective of "city as platform" is not only a social system, but also a part of the technical system, that is, authorized operation is an abstract social-technical system, which should be built and run on a unified technical base . How to understand it? First, from the basic requirements of data security, any development and utilization of authorized operations should be based on a unified technical base, that is, "no number of domains"; secondly, all outputs of authorized operations should be operated and retained in a unified technical base, that is, "maintain the loose coupling of data products and authorization relationships", which ensures the stability and continuity of any data product without any interruption due to changes in authorized subjects. Therefore, we agree with Changjiang  and Xiao Weibing  The assertion that authorized operations should build contracts through PPP; furthermore, in terms of the quality of public data, rooted in The unified platform for Data Governance and the development and operation of primary data products is conducive to the traceability of all stakeholders to any products and services developed based on public data, thereby ensuring credible public data circulation and laying the foundation for the credible operation of the data element market.
In this paper, we challenge the currently commonly recognized logical subordination between data authorization operation and data opening. By sorting out and unifying the definition of data opening in the Chinese context, we believe that data opening has been generalized into external data circulation behavior in China, and therefore data authorization operation should not be regarded as a new form of external circulation behavior, but should be regarded as a response to the current administrative power "unwilling can't" bottleneck and then solve "who will open up".
We believe that the proposal of this new relationship theory has three advantages:
First, to ensure the continuity of existing policies and regulations : it ensures that the current policies and regulations have set the data opening system to continue, without the introduction of authorized operations and fragmentation conflicts; at the same time, do not artificially equate authorized operations with licensed development (that is, licensed Objects obtain data only for the development of specific applications), thus reversing history and destroying the achievements of data opening.
Second, it is conducive to the consistency of cognition of all parties in society : it ensures that public departments and data users have a consistent understanding of the external circulation of public data, without deliberately distinguishing between opposing data opening and authorized operations, resulting in cognitive and practical confusion;
Third, it is conducive to the independence of institutional exploration experiments: it ensures that public departments explore the opening of experimental data (especially conditional opening) and authorized operations have their own independence, that is, departments with the ability and willingness can choose to directly carry out conditional data. The system construction and practice of opening up, rather than requiring authorized operation; and the experimental exploration of authorized operation will not affect the existing open system.
After clarifying the positioning of authorized operation, we pointed out that the current general discussion of authorized operation lacks a refined discussion of authorized behavior and authorized output, and initially expanded the discussion on these two points based on the existing limited literature basis. Although it is impossible to construct authorization behavior and authorization output in a comprehensive and detailed manner, our discussion can basically reveal the current simplified and single understanding of authorization operation, that is, it is mainly carried out around the "processing and utilization" of "data to data product services", which has corresponding limitations. At the same time, by introducing the perspective of social-technical systems, we also preliminarily reveal the limitations of the current discussion of data authorization operations from a single social science or a single purely technical perspective. Based on this, we propose that the next step should be from the perspective of social-technical systems. Discuss and experiment on the institutional design of authorized operations, especially its operating logic and regulatory mechanism.
Finally, we suggest that for the construction and practice of the system of data authorization operation, we should pay close attention to the practice and research under the same logic at home and abroad, including but not limited to data trust, data space, etc., and summarize experience from it. Further research on the theory, through theoretical summary, reverse guidance research, so as to ensure that authorized operations serve data openness well, rather than mistakenly opening the historical reversal to a single-point, monopolistic data circulation and utilization mode.
State Council. Opinions of the Central Committee of the Communist Party of China and the State Council on Building a More Perfect Market-based Allocation System and Mechanism for Factors [Z] .2020-04-09.
Shanghai Municipal People's Government. Interim Measures for the Opening of Public Data in Shanghai [Z]. 2019-08-29.
Digital and Mobile Governance Laboratory of Fudan University. China Local Government Data Openness Report-Provincial Index (2022) Open Forest Index [R/OL] .2023-01-10. http://ifopendata.fudan.edu.cn/report
Davies,T. & Fumega, S. Global Data Barometer Report (First Edition) [R]. 2022.
Open Knowledge Foudation. Open Definition [OL]. https://opendefinition.org/. Last Accessed on 2023-01-04.
Zhang Huiping, Gu Qin, Xu Zhongbo. Research on the Implementation Mechanism and Internal Mechanism of Government Data Authorization Operation - Taking Chengdu as an Example [J]. E-Government, 2021 (05): 34-44. DOI: 10.16582/j.cnki.dzzw.2021.05.004.
China Software Evaluation Center. Research Report on Public Data Operation Model [R] .2022
Wang Weiling. Authorized operation of government data: practice dynamics, value network and promotion path [J]. E-Government, 2022 (10): 20-32. DOI: 10.16582/j.cnki.dzzw.2022.10.002.
Yuan Qiang, Shi Xiaobing. Analysis of the legal nature and system improvement of government data authorized operation [J]. China Informatization, 2021 (07): 8-10.
Xiao Weibing. Analysis of legal issues on government data authorization and operation [J]. Journal of Beijing Institute of Administration, 2023 (01): 91-101. DOI: 10.16365/j.cnki.11-4054/d.2023.01.008.
Chang Jiang, Zhang Zhen. On the characteristics, nature and legal regulation of authorized operation of public data [J]. Research on the Rule of Law, 2022 (02): 126-135. DOI: 10.16224/j.cnki.cn33-1343/d.20220217.001.
WU Liang. Legal Perfection of Government Data Authorization Operation Governance [J]. Law Forum, 2023, 38 (01): 111-121.
Zhao Yehu, Yu Kai. Analysis on the legal mechanism of authorized operation of government data [J]. Chinese Lawyer, 2022 (01): 82-84.
Lu Zhipeng. Exploration of Public Data Authorization Operation Mechanism [J]. Cyber Security and Data Governance, 2022, 41 (07): 30-35. DOI: 10.20044/j.csdg.2097-1788.222.01.005.
SONG Shuo. Constructing Public Data Open Utilization Mechanism Based on Authorized Operation [J]. Legal Science (Journal of Northwest University of Political Science and Law), 2023, 41 (01): 83-94. DOI: 10.16290/j.cnki.1674-5205.2023.01.014.
LIU Yangyang. Public Data Authorization Operation: Generation Logic, Practice Picture and Specification Path [J]. E-Government, 2022 (10): 33-46. DOI: 10.16582/j.cnki.dzzw.2022.10.003.
Hu Yefei, Chen Meixin, Zhang Yimeng. Balancing Value Co-creation and Data Security: Research on FL-based Government Data Authorization Operation Model [J]. E-Government, 2022 (10): 2-19. DOI: 10.16582/j.cnki.dzzw.2022.10.001.
Zhang Huiping, Ma Taiping, Sun Lishuang. Government Data empowering digital economy upgrade: authorized operation, Privacy Computing and scenario reconstruction [J]. Intelligence Journal, 2022, 41 (04): 166-172.
Open Data Institute. Data Spectrum [OL]. www.theodi.org/about-the-odi/the-data-spectrum/. Last Accessed on 2023-01-04.
Xia Yikun. Discussion on the market-oriented model of government commercial information resource acquisition and utilization [J]. Library and Information Work, 2009, 53 (07): 121-124.
Zhang Huiping, Gu Qin. Government Data Flow: Mode, Practical Dilemma and Collaborative Governance [J]. Governance Research, 2022, 38 (03): 59-69 + 126. DOI: 10.15944/j.cnki.33-1010/d.2022.03.011.
Ye Yazhen, Zhu Yangyong. Boxed Data: A Data Product Form Based on Data Box [J]. Big data, 2022, 8 (3): 15-25.
CAO L B .Data science [J]. ACM Computing Surveys, 2018,50(3): 1-42.
PEI J . Data pricing:from economics to data science [C]// Proceedings of the 26th ACM SIGKDD International Conference on Knowledge Discovery ＆ Data Mining. New York:ACM Press, 2020: 3553-3554.
Zhijiang Laboratory. White Paper on Standardization of Data Product Transactions [R]. 2022.