[author]LIN Xifen, LI Z

[content]

*Author Lin Xifen, 

Professor at Kaiyuan Law School of Shanghai Jiao Tong University and Vice Dean of China Academy of Law and Society

*Author Li Zheng, 

PhD student at Kaiyuan Law School, Shanghai Jiao Tong University

Abstract: With the advancement of the construction of new liberal arts, computational law is becoming a hot emerging discipline and research paradigm in recent years. The current problem with Chinese computational law is that there is no consensus on the development trajectory of computational law, especially in terms of the research scope and development path of the discipline. The article systematically sorts out, analyzes, and compares relevant literature on computational law research worldwide in the past two decades using various statistical methods and research tools, reproducing the evolution process and differences in research stages, scientific research cooperation, research themes, and knowledge structures of computational law research at home and abroad from the beginning of the 21st century to recent years. Compared with foreign countries, Chinese computational law may undergo differentiation in different research areas. Advocating problem oriented computational law research is beneficial for sorting out the relationships between different research paths and promoting the orderly development of the discipline. The academic community needs to attempt to integrate the research path of computational law from different dimensions of problem awareness, research ability, and research methods, and explore the Chinese path of the development of computational law.

Keywords: computational law; Legal empirical research; Computational science; artificial intelligence; Bibliometrics

1.Research origin: How to empirically observe the development of computational law

With the steady advancement of empirical research, the organic integration of intelligent technology, and the gradual enrichment of data resources, computational law is becoming one of the hot topics in the current field of law. On the one hand, the national policy has increased support for interdisciplinary studies in universities, promoting the construction and development of computational law and related disciplines. Emerging disciplines named after computational law, digital law, artificial intelligence law, etc. have successively landed in many mainstream universities. Subsequently, highly interdisciplinary topics such as data rights, legal big data, and judicial artificial intelligence have emerged as new academic growth points in the field of law. On the other hand, compared with previous empirical research based on sporadic data or qualitative information, empirical research based on large-scale data such as judicial documents and supplemented by quantitative statistics is increasingly advocated and promoted, producing a series of academic achievements with strong data science attributes, the latter of which is an academic research paradigm with stronger computational attributes. In addition, academic forums related to computational law, jointly established by the legal community, other academic circles (such as the computer science community), and industry such as the China Computer Society's Computational Law Branch, have promoted the discipline construction of computational law and the paradigm transformation of legal research to varying degrees.

The concept of computational law was not first introduced in China, but originated from the development of related disciplines outside the country and was translated and introduced by domestic scholars. At present, several scholars have made efforts from the perspective of disciplinary theory, such as Zuo Weimin, Ji Weidong, Shen Weixing, Ma Changshan, Qu Maohui, etc., who have conducted a comprehensive review and outlook on computational law from the basic concepts, disciplinary categories, and system construction of computational law; Yu Xiaohong, Deng Jingting, and others have provided a detailed exposition on the application of computational science in legal research from a methodological perspective. Zhang Ni, Liu Dongliang, and others discussed the function and role of computational thinking in computational law. These studies have made fundamental contributions to the rise and development of computational law in China. However, as they mainly adopt a perspective of theoretical refinement and disciplinary advocacy, it seems necessary to introduce more computational methods to restore the global development network of computational law and explore the development path of Chinese computational law.

This study is based on nearly 20 years of literature in the field of global computational law, following computational thinking and using bibliometric methods and tools, aiming to comprehensively diagnose the current development status of computational law at home and abroad. It attempts to explore the following core issues: firstly, what are the similarities and differences between China's computational law and those outside the field in terms of research stages, research cooperation, and research themes? Secondly, the discipline of computational law has its own interdisciplinary nature. Will scientific cooperation between different disciplines promote the output of computational law achievements? Thirdly, is there a difference in the evolution dimensions of the knowledge structure of computational law in China compared to outside the country? If there are differences between the two, what are the root causes? Fourthly, is there a research on computational law from a macro perspective, and how can we better clarify the relationships between different research paths in computational law? The author hopes to reflect on the development path of computational law in China by answering the above questions.

2. Data sources, calculation methods, and operational tools

2.1 Data source

The concept and scope of computational law have not yet formed a unified opinion, and there are continuity, inclusion, and intersection relationships between it and related disciplines and concepts such as legal informatics, quantitative law, data law, and artificial intelligence law. From the current macro concepts and connotations proposed by the domestic academic community, relevant research can be divided into three aspects: ontology, rule-based theory, and instrumentalism. It is the combination of computational science and legal thought at the levels of computational thinking, computational technology, and computational methods. Therefore, at the level of abstract computational thinking, this study selected keywords such as "computation", "automation", and "machine" to reflect the basic concept of using computational science to solve problems; At the specific level of computing technology, keywords such as "artificial intelligence" and "big data" were selected to reflect the most concerned technologies in the field of computing science; At the micro level of computational methods, keywords such as "machine learning" and "algorithms" are used to represent the introduction of computational science methods in legal research and practical applications. At the same time, in addition to focusing on the challenges and empowerment of computational science and technology on the rule of law, computational law also emphasizes the introduction of computational science in research methods, with quantitative empirical research as the main feature and judicial documents as common research samples. Therefore, keywords such as "judicial documents" and "empirical research" were selected to supplement and improve the search function. After combining the above keywords separately and searching in the WoS core database and CSSCI database, the disciplinary scope was limited to the field of law, and a total of 5060 original research and literature reviews published between 2000 and 2020 were collected. To ensure the accuracy of the research results, a combination of software automatic deduplication and manual exclusion was used to remove duplicate data and irrelevant information from the sample.

2.2 Calculation method

This study is conducted from two levels: basic situation and knowledge structure. At the basic level, descriptive statistics and curve fitting are performed on the number of literature. The exponential fitting formula is y=α e β x, and the linear fitting formula is y=α x+β; In the analysis of the research subject's situation, collaborative network analysis was introduced, and the point degree centrality of each author in the largest subgroup was calculated. In this undirected graph, the degree centrality d (ni) of a node ni can be expressed as:

Among them, ki represents the number of edges connected to node i, and N-1 represents the number of edges connected to node i and other nodes. To further reveal the relationship between the author's publication volume, professional background, and their position in the internet, Pearson correlation analysis and independent sample t-test were used for verification; In the analysis of research topics, geodesic distance was used to reduce the dimensionality of keywords, and Euclidean distance was used for multidimensional scale analysis. The two points where distance needs to be measured are represented by i and j respectively, and the calculation formulas are:

At the level of knowledge structure, this study adopts literature co citation analysis. Through computer processing, the original citation network of the literature is matrixized, and the co citation matrix is obtained based on matrix operations for further statistical processing.

2.3 Operating tools

In order to facilitate understanding and reading, this study used various software and methods to visualize the results. Among them, statistical analysis and visualization were completed using SPSS 25.0, author sub network analysis and visualization were completed using Bicomb 2.0 and Ucinet 6, and knowledge structure evolution analysis was completed using Citespace 5.8.R3.

3. The Current Research Status of Computational Law in China: A Comparative Study with Foreign Countries

3.1 There are significant differences in research stages both domestically and internationally

In the past two decades, the number of research papers published in the field of computational law by various countries has shown an increasing trend. The number of research literature in countries such as the United States, the United Kingdom, and Italy has shown a fluctuating increase, with different periods of rapid growth observed in each country. The first important literature in our country appeared in 2001, and the annual publication volume has steadily increased to 30 in 2013. Subsequently, there were two explosive increases in the number of domestic research literature. It first appeared in 2014, doubling the number from the previous year and rapidly growing at an average annual rate of 18% until 2017. The following year, the number of literature exploded again, doubling year-on-year. By then, China's important scientific research output in this field had far exceeded the total of other countries, and remained stable at around 300 publications per year (Figure 1A).

The evolution process of a certain discipline from its birth can be divided into four stages: discipline birth period, discipline development period, discipline maturity period, and discipline saturation period. The literature growth curves of each stage show four forms: difficult to determine expressions, stable exponential growth, evolution towards linear growth, and horizontal axis. Analyzing the changes in the number of publications in China and the United States over time, it can be found that the trend of changes in the number of publications in CSSCI database (R2=0.853), WoS database (R2=0.912), and the total number of publications in both databases in China over the past two decades is consistent, showing a significant exponential growth overall (R2=0.917), while the linear fitting results are poor (R2=0.577). The trend of changes in the number of publications in the United States is slightly higher in the goodness of fit of the exponential curve (R2=0.906) than in the linear curve (R2=0.889), but the difference in data between the two is not significant (Figure 1B, Figure 1C). From the results, it can be seen that China and the United States are at different stages of development in the field of computational law research. The two explosive growth events since 2014 have led to a period of disciplinary development in China's computational law research, while at the same time, computational law research in the United States is shifting from a period of disciplinary development to a mature stage.

3.2 There is a significant lack of domestic research cooperation

With the increasing complexity of scientific problems, the output of high-quality research results relies more and more on team collaboration. At present, countries have developed close cooperative relationships in the research of computational law. There is extensive cooperation between the United States, which has the highest publication volume, and developed countries in Europe and America, accounting for 11% of the overall research results. Although the proportion of international cooperation achievements published in foreign journals in China has reached 31%, it only accounts for 1% of the total number of domestic and foreign journals. At the level of research institutions, a research collaboration network has been formed abroad, led by institutions such as Harvard University, Stanford Law School, and Oxford University. Although domestic universities and research institutions have established research cooperation relationships with overseas institutions, they have not carried out research cooperation with leading institutions in this field. This result once again reflects indirectly that the quantity and quality of China's international cooperation in this field need to be improved.

At the individual level, research collaborations dominated by European and American scholars are more common in computational law research under the tool theory path, while domestic scholars have a relatively low proportion in the network. The author further extracted the largest subgroups in the network and generated two largest subgroups of the author co-occurrence network, covering the main scholars engaged in legal informatics and quantitative law research, representing the origins of the two main disciplines of computational law (Figure 2A, Figure 2B). There are 33 authors in each of the two networks, and the more collaborative relationships they have, the more central their position in the network is. This position can be reflected by the degree centrality d (ni), and the higher the value, the higher their position in the network. In the first subgroup, T. J. M. Bench Capon is at the center with d (ni)=52; H. Prakken and F Bex and 9 others are in important positions, with d (ni)>20; K. D. Ashley and G Sileno et al. 16 times, d (ni)>5; D. Walton and L Van der Torre and 7 others are in a similar position, with d (ni)<5. In the second subgroup, Stephen J. Choi is at the center with d (ni)=12; Michael Heise and Kuo Chang Huang followed closely behind, with d (ni)>5; Tom Ginsburg and Adam S. Chilton, among 15 others, had d (ni)>3, while Lior Jacob Strahilovitz and Geoffrey P. Miller, among 8 others, were in a second place with d (ni)=1.

The Pearson correlation analysis was used to test the relationship between the number of articles published by the author and the degree centrality. The results showed that the correlation coefficients between the two were greater than 0.6 (R=0.771, P<0.01); R=0.657, P<0.01), there is a strong positive correlation between author status and the number of publications in both subgroups. In the first subgroup, there are 33 scholars engaged in research related to legal informatics, mainly computer scientists, but 18 of them all have a background in law. Among the 33 authors engaged in quantitative law research in the second subgroup, 14 of them have professional backgrounds outside of law. Among them, 8 have degrees in economics, 2 have degrees in statistics, and the remaining 4 have degrees in political science, philosophy, psychology, and science respectively. According to their professional backgrounds, the authors in the two subgroups were divided into two groups. The independent sample t-test results showed that regardless of which subgroup, authors with composite backgrounds had significantly higher degree centrality than authors with single disciplinary backgrounds (P<0.05).

3.3 Convergence of research themes and dimensions both domestically and internationally

Using Bicomb to extract keyword fields from two databases, merging synonyms, and selecting the top 10% as high-frequency keywords, 41 Chinese keywords and 33 English keywords were obtained. The top 10 high-frequency keywords in Chinese are artificial intelligence, big data, copyright, personal information, big data era, cloud computing, algorithms, privacy, intellectual property, and legal regulation. The top 10 high-frequency keywords in English are artificial intelligence, big data, machine learning, privacy, ethics, deep learning, neural network, data protection, law, and GDPR. The above keywords represent the hot topics in computational law research at home and abroad in the past two decades. To further investigate the intrinsic connections between keywords, Bicomb 2.0 was used to generate a 41 × 41 Chinese co word matrix and a 33 × 33 English co word matrix. It can be observed that Chinese keywords are divided into 5 categories. Artificial intelligence occupies a central position in the network, forming a separate cluster. Words such as "personal information protection" and "right to be forgotten" form a second cluster, "right ownership" and "artificial intelligence creations" form a third cluster, "legal personality" and "autonomous driving" form a fourth cluster, and "judicial big data" closely related to "artificial intelligence" and "big data" form a fifth cluster. English keywords are divided into four categories. Among them, words such as "Big data" and "GDPR" form the first cluster, occupying the center of the network, words such as "Cloud computing" and "Internet of things" form the second cluster, words such as "Data mining" and "Social media" form the third cluster, and words such as "algorithms" and "surveillance" form the fourth cluster. After importing the co-occurrence matrix into SPSS 25.0 and converting it into a similarity matrix, the data fit was excellent, revealing discourse convergence in important research fields both domestically and internationally.

4. The Knowledge Genealogy of Chinese Computational Law: Cluster Comparison and Evolution

Based on the co citation analysis of literature, a cluster analysis was conducted to compare the research results of China and foreign countries. The results are shown in Figures 3A and 3B, which respectively illustrate the knowledge structure of computational law research in China and abroad. The color of the nodes from dark to light represents the change in time from far to near, the size of the nodes represents the citation frequency, and the dark circles outside the nodes indicate that the article is a highly prominent literature with a sharp increase in citation frequency within one year, which can reflect the changes in research hotspots in different periods.

From the overall structure of the co cited network, in terms of time dimension, foreign research started earlier, and the knowledge structure has undergone multi-level evolution as a whole. However, China has only started large-scale related research in the past five to ten years, and the knowledge structure is more flattened in the time dimension; In terms of spatial dimension, foreign research content is more abundant, including legal empirical research, research on computational law methodology, artificial intelligence governance and other aspects. In China, more emphasis is placed on the research of new legal issues such as personal information leakage, autonomous vehicle infringement brought about by the development of computing science, and other dimensions are less involved.

4.1 The Evolution of Knowledge Structure in Computational Law Research in China

Through cluster analysis, 2182 domestic literature were divided into 7 independent clusters, automatically named as Big Data, Personal Information Protection, Product Liability, Criminal Liability, Copyright, Legal Artificial Intelligence, and Algorithm Regulation (Table 1).

The first clustering in the field of computational law in China was formed in 2013, focusing on the fundamental research of personal information and big data. The distinction between personal information and privacy during this period, as well as the recognition of the potential property value of personal information, have strong guiding significance for subsequent research. With the widespread application of big data technology, the property attributes of personal information have significantly increased. The phenomenon of secret collection, secondary processing, and illegal resale of personal information has attracted the attention of the academic community. Based on the understanding of the value of personal information property, the academic community has further conducted research on its legal nature and protection. By 2016, the relevant research results began to form a second cluster, and the focus of research gradually shifted from individuals to relevant entities such as enterprises and administrative departments. Due to the limited effectiveness of traditional privacy protection frameworks, academia has begun to pay attention to the personal information protection reform and institutional design in European and American countries where big data technology was applied earlier, and has attempted to seek balance among various stakeholders, providing theoretical guidance for the legislative work of China's personal information protection law.

Around 2017, the domestic research heat continued to rise, and multiple research directions were formed in a short period of time. Cluster 2 is formed from a general exploration of legal issues surrounding the characteristics of artificial intelligence technology, focusing on the potential risks of technology, legal challenges, and legal responses, resulting in two types of conclusions. The first approach is based on a highly probable understanding, viewing the application of technology as a systemic risk, and believing that bounded rationality laws are difficult to cope with the development of artificial intelligence technology, and that laws should respond in a timely manner. The second approach is based on a realistic understanding, believing that existing legal theories and frameworks are sufficient to address legal issues arising from artificial intelligence. Cluster 3 is composed of research achievements in criminal law, shifting from the study of general issues to specific scenarios, focusing on the criminal responsibility ability of artificial intelligence and criminal law regulation issues. It continues the characteristics of the two assertions in Cluster 2, forming two distinct positions of "affirmation" and "negation", and discussing the control and recognition ability of artificial intelligence, the possibility of assuming criminal responsibility, and the necessity of legislative regulation. Cluster 4 is also a continuation of general problem research in the field of intellectual property, but unlike Cluster 3, the two viewpoints focus more on the properties of artificial intelligence generated products and pay less attention to whether artificial intelligence should become the subject of legal relationships. On this basis, further discussions and reflections on rights protection have been formed.

Corresponding to the vigorous theoretical discussions in academia, practical agencies have shown even greater enthusiasm for computing technology. While introducing relevant technological innovations to work methods, problems continue to emerge, forming the main content of Cluster 5. The application of computing technology by investigative agencies started earliest, generating a large amount of data while meeting urban security needs. As a result, traditional investigative models have shifted towards integrated and predictive investigations driven by big data. However, two important issues still need to be addressed urgently. One is the problem of excessive collection of citizen information, and the other is the legitimacy of using technological means for the purpose of discovering clues. At the same time as the digital transformation of investigative agencies, computing technology has swept through judicial agencies, forming distinctive research on judicial artificial intelligence. Some logical and legal scholars continue the research context of foreign legal artificial intelligence, attempting to improve the theoretical paradigm and practical path of legal artificial intelligence from the perspective of legal reasoning, and further proposing a theoretical idea of the integration of legal logic and data-driven new generation artificial intelligence. Some legal scholars, starting from China's judicial practice, on the one hand, combine the practical forms of intelligent justice and digital prosecution to construct an ideal picture of judicial artificial intelligence application. On the other hand, they focus on the specific application scenarios of legal artificial intelligence and point out the hidden concerns in the digital construction of judicial organs. In this context, following judicial laws and constructing a human-machine collaboration mechanism with judicial personnel as the main body to achieve limited application of judicial artificial intelligence has gradually become a consensus in the academic community.

In the study of general issues in computing technology, the core position of algorithms has quickly attracted attention from the academic community. Related research has derived research on technical characteristics from discussions around concepts and phenomena, returning from specific problems to discussions of general issues, further enhancing the depth of research. In 2018, Cluster 6 was formed with algorithms and their regulations as its research content. It is precisely because of this profound understanding of algorithms that the threat of technology to privacy and freedom, the resulting discrimination and bias, and the capital monopoly it brings seem to have found the real reasons, which have become the foothold for the vast majority of scholars to advocate regulating computing technology. At present, three types of models have been formed in research: legal regulation, ethical regulation, and technical regulation. One of the mainstream views currently is to regulate the use of models through legislative means, such as requiring algorithm disclosure, empowering personal data, creating algorithm interpretation rights, assessing risks, and limiting applications. Secondly, it is required that algorithms must ensure that the results of their operation can safeguard the interests of humanity during the writing process, and that laws and ethics are integrated into algorithms. Some countries are exploring this model. Thirdly, by writing code to embed programs with regulatory functions into algorithms, data review, dynamic adjustment, and compliance review can be achieved during the training and application stages of the model. Its essence is to enable algorithms to achieve self-regulation, and the computer science community is actively exploring this approach.

4.2 The Evolution of Knowledge Structure in Computational Law Research Abroad

Through cluster analysis, 2878 articles in the WoS database were divided into 8 independent clusters, each automatically named as empirical legal studies, federal court, corporate litigation, circuit court, computational methods, local enforcement, big data, and algorithmic accountability (Table 2).

The first two clusters of foreign research were formed in 2008, which are the phased achievements of legal scholars represented by the United States in the field of judicial politics in the past two decades. Cluster 0 focuses on the Supreme Court, while Cluster 1 focuses on the Federal Court of Appeals. These studies focus on treating judicial decisions as raw data, using empirical research methods to analyze the data, attempting to depict the relationships between variables in judicial decision-making, proving or refuting specific hypotheses about these relationships, in order to gain a clearer understanding of the operational process of judicial decision-making. In recent years, research in judicial politics has not stopped at revealing behavioral patterns, but has begun to use machine learning algorithms such as Markov Chain Monte Carlo algorithm, Bayesian model, support vector machine, etc., introducing more variable factors, hoping to construct robust models that can predict behavioral patterns. Cluster 2 was formed in 2013, focusing on empirical research in law and economics, and can be summarized into two main aspects. One is the analysis of the economic motives behind litigation behavior, and the other is the study of the impact of law on economic governance. Cluster 3, formed in the same year, focuses on empirical research on constitutional and administrative law, and focuses on the issue of power balance between legislative, judicial, and administrative institutions. Through empirical data testing, it examines the viewpoints or theories formed in previous research, mainly focusing on topics such as the interpretation methods of enacted laws and the judicial review of legislative authorization.

By 2016, foreign countries began to summarize and innovate various types of computational law research methods, forming the main content of Cluster 4, which is the latest achievement of applying artificial intelligence technology to legal practice and legal research. Overall, there are two forms of application of computational science from a legal perspective, namely Law as Code and Law as Data. The former has evolved into legal technology applied in practice, while the latter represents quantitative research in applied computational science. In the past, legal technology research in law journals was mostly focused on the application prospects from a legal perspective, using it as an auxiliary work or research tool, or viewing it as a technical foundation for changing existing judicial practices. Recently, there has been a trend towards the integration of the two in some studies, and the newly formed legal technology has become an important tool for legal research. The data and models obtained through quantitative research have also contributed to the development of legal technology in judicial practice.

The research on privacy and personal information protection in foreign countries is more closely related to emerging computing science and technology, mainly focusing on their practical applications and regulations. Around 2011, relevant research results formed Cluster 5 and Cluster 6, with the regulation and privacy protection of legal technology applications in public power agencies as the main research content. Two types of personal information collection and utilization behaviors are often studied in the research. One is intelligence led law enforcement applications, represented by DNA databases and predictive law enforcement systems; The second is the application of voter databases in political elections. The regulatory exploration during this period began with the anonymization of data, but the processed data can still achieve personal identity recognition after matching with a large amount of publicly available non personal information data. At the same time, the repeated use of data has significantly increased the systemic risk of privacy infringement for the public. The loss of the right to know, unfairness, and discrimination caused by the application of artificial intelligence algorithms have emerged in large numbers, further exacerbating long-standing racial issues. Due process has become another option for algorithm regulation. By 2016, relevant research had formed Cluster 7, which extensively discussed applications represented by automatic decision-making systems and the EU's General Data Protection Regulation. The effectiveness of a series of measures such as algorithm transparency and improved interpretability in practice was questioned, especially when the pursuit of algorithm transparency caused further privacy infringement and the accompanying "illusion of control". Therefore, two more typical governance schemes have been further formed. One is technical regulation, which advocates considering issues such as value judgment and due process at the technical level when building artificial intelligence systems. Based on static code checks, the entire process of data input, output, and calculation is fixed to enhance interpretability and compensate for the lack of regulatory understanding caused by technical characteristics. The second is management regulation, which is no longer limited to public authorities, but advocates the formation of an industry self-discipline mechanism through strengthening the review and supervision of private enterprises, third-party evaluation, self correction, and internal personnel reporting, and even reducing the risk of product discrimination through internal management methods such as changing the gender ratio of personnel. This diversified governance concept and personal data empowerment have initially formed a binary governance model, which is further summarized as "collaborative governance" and seen as the future path of algorithmic regulation.

5. The future direction of Chinese computational law: differentiation or integration?

From the comparison of computational law between China and foreign countries mentioned above, it can be found that research in China and abroad is generally similar in some dimensions, but each has its own characteristics. On the one hand, there is a convergence in the research of important issues, and China is not lagging behind foreign countries in the research of cutting-edge issues. On the other hand, foreign research started earlier, with relatively mature choices of research subjects, application of methods, and elaboration of viewpoints. Researchers have diverse disciplinary backgrounds and engage in extensive scientific research cooperation, leading the development of disciplines in China. The prominent differences between China and foreign countries are reflected in the research of the tool theory path. Econometric law and legal informatics, as the main branches of computational law research abroad, occupy half of the knowledge landscape of computational law. They have produced excellent results in different research areas such as legal empirical research, legal technology research, and legal regulation research, and have formed relatively close connections. In recent years, the concepts of related disciplines have begun to be accepted by the domestic academic community with the rise of social science law, the emergence of quantitative legal research in criminal law, and the arrival of the data age. Different views have emerged on the origin of computational law, such as the continuation of quantitative law, the intersection of quantitative law and information science, the branch of legal information science, and the branch of computational social science. Furthermore, different disciplinary connotations and corresponding research categories have been summarized. Among them, the majority of legal scholars hold a narrow view of computational law that focuses on quantitative research, and an increasing number of computer scientists and logicians are attempting to approach computational law research from the perspective of legal informatics. However, from the analysis results of the knowledge graph, it can be seen that the instrumental path foundation of computational law research in China is still relatively weak, and there are preliminary signs of fragmentation between different research dimensions of computational law.

5.1 Chinese computational law moving towards differentiation

There are multiple paths in the research of computational law, and the tool theory path advocating the use of computational technology for empirical research is one of them. In the data of this study, there is always a certain amount of empirical research results in Chinese law. However, unlike the knowledge structure of computational law in foreign countries, there is no co citation network with other research in this direction in China, reflecting the lack of legal empirical research in the knowledge structure of computational law in China. In recent years, thanks to the abundance of data resources, the popularization of computing methods, and the improvement of computing power, research costs have been reduced. However, quantitative research methods have not become the mainstream paradigm. Compared to foreign countries, China's legal research does not have a tradition of quantitative research, and there are very few literature using big data research. At the same time, the current empirical research methods in law are somewhat immature and not closely related to computational science. The vast majority of quantitative research results are often questioned whether they can be used as empirical research in the context of computational law. The shift of research paradigm cannot be achieved overnight, but when quantitative research methods were not yet widely used in China, big data research methods were highly praised in academia. On the one hand, it promoted the output of high-quality research results, but on the other hand, it also reduced the recognition of small sample data research, indirectly raising the threshold for empirical research.

Another tool based approach proposed by academia in computational law research is the study of legal technology. From the perspective of relevant research results in China, computer scientists and legal scholars are still relatively independent and have not yet achieved deep integration at the theoretical level of disciplines. The legal technology research led by computer scientists aims to serve the informationization construction of legal practice work, mostly in the form of application programs, data platforms, smart terminals and other service products. The theoretical research subject of legal technology is mainly legal scholars, who conduct in-depth discussions from multiple levels such as legal theory, judicial practice, and judicial reform, and the results are mostly published in legal journals in the form of papers. The former focuses on the attributes of applied research, which determines the difficulty of forming theories from research. For the latter, although the research on legal technology oriented towards computational science is highly popular, it has not yet been accepted as mainstream legal research by the legal community. When there are significant differences in the scientific problem attributes, achievement presentation forms, and research assessment orientation between the two, the enthusiasm for interdisciplinary cooperation in theoretical research is weakened. At the same time, the academic community, journals, and activities of computational law in the Chinese speaking world are still in the process of formation, and do not yet have the space for publishing achievements that can compete with foreign ministers. This further restricts the formation and presentation of theoretical discussions, resulting in a differentiation between theoretical research and practical applications of legal artificial intelligence.

With the development of the new liberal arts construction, the differentiation of computational law is directly reflected in the establishment of disciplines in universities. In recent years, computer science has emerged as a cross disciplinary field with other disciplines such as computational biology, computational linguistics, computational finance, and computational physics. The common feature of these disciplines is the adoption of a tool based approach to discipline construction, which regards computational science and technology as research methods to enrich and develop the theory of the discipline. This is also the development direction of computational law advocated by many scholars in China. Compared to other secondary disciplines that intersect with computational science, the vast majority of computational law disciplines in China, especially those that excel in law, adopt an object perspective based on legal theory, viewing technological applications as research objects to ensure the orderly development of science and technology. This disciplinary approach reflects the rightful purpose of legal research. As different aspects arising from the intersection of computational science and law, there is a significant knowledge gap between the tool perspective and the object perspective in computational law, which weakens the overall effect of computational law. At present, computational law cannot be considered an independent interdisciplinary field, but rather a cross disciplinary study of computational science and law. If Chinese computational law wants to form an independent interdisciplinary field from interdisciplinary studies, it still needs to further integrate multiple perspectives of computational law into problem oriented computational law, and achieve a positive interaction between different research paths.

5.2 Problem oriented Chinese Computational Law

Firstly, the integration of Chinese computational law is based on problem oriented computability thinking. The study of mathematical logic has formed the fundamental concept of computability in recursive theory through a series of discussions such as the Erbron Gödel computable functions and the Church Turing thesis. With the development of computing technology, the connotation and extension of this philosophical concept continue to expand, becoming the source of development for computational science. In mathematics and computational science, if there is a feasible process to solve a problem, then the problem is called "computable". From the perspective of computational law, computability should refer to the dialectical thinking of the computability of law and whether law should be calculated. Whether starting from the common characteristics of law and computer systems, the logical structure of legal reasoning, or the thriving development of legal artificial intelligence, the computability of law as a description of social order can be proven. Whether the law, as a means of social governance, should be calculated is a question that computational law should also answer: on the one hand, the digital connections formed by computational science reconstruct individual social relationships, and on the other hand, the algorithms relied upon by computational science lead to imbalances between power and power, power and power. In this process, the application of public power in legal technology as an algorithm for law, and the application of private power in digital technology as an algorithm for invading law, both require a return to the study of new legal issues to the revelation of social phenomena and the discussion of basic legal issues, and a response to computable law.

Secondly, the integration of computational law in China follows the development law of problem oriented disciplines. When setting up disciplines in universities, in order to emphasize the organization and boundary sense of disciplines, a certain discipline or academic research category is artificially divided. However, the actual change of this category is not subject to human will, but follows the general laws of knowledge development. From the source of knowledge, whether it is Plato in the West or Kong Lao in the East, the multiple disciplines involved in their works that can be subdivided were considered as a whole at that time. Since modern times, with the deepening of scientific research, disciplinary differentiation has become the mainstream of disciplinary development, but it has also caused isolation and disconnection between disciplines. In recent years, disciplinary integration has become a trend, quickly filling the gaps between disciplines and forming a single interdisciplinary group represented by physics and chemistry and an interdisciplinary group represented by information theory. The inherent need for scientific progress in the study of scientific problems has led to the continuous separation and integration of disciplines, forming the general development laws of disciplines. The development of computational law should also follow this trend.

Once again, the integration of Chinese computational law should be able to address the challenges of a problem oriented digital society. In the reality of deeply integrating algorithms into social order, participating in social management, and causing social problems, human civilization has undoubtedly entered the digital society, and the governance of the digital society inherently requires the digital transformation of social governance. The different research paths based on computational science, such as quantitative legal research, legal technology research, and research on new legal issues, essentially respond to problems from different perspectives of discovery, resolution, and optimization, and form a mutually supportive, interdependent, and common development relationship. The research on legal technology can provide support for data-driven quantitative research in law. The technology, data, and computing power mastered by the former are not possessed by the latter. The development of legal technology to a certain extent determines the height of quantitative research in law. Compared to pure technical research or other application scenarios, legal technology research provides opportunities for legal scholars to deeply engage with computing technology, and a large number of practical applications enable Chinese scholars to have sufficient research samples. Empirical research in law can provide a more comprehensive research perspective and approach for the study of legal issues related to computing technology. Through the correlation relationships existing in massive data, it can explore the real problems brought by digital technology and provide a foundation for standardized research and value judgment.

5.3 The Contradictions and Paths of Integrating Chinese Computational Law

Problem oriented computational law research is more conducive to mutual promotion among various directions, but at the same time, it is limited by two practical difficulties: one is the reality of Chinese legal education with a relatively single professional knowledge background, and the other is the reality of Chinese legal research lacking substantial scientific research cooperation. Computational law is an interdisciplinary research field that involves multiple disciplines such as departmental law, information technology, data science, and statistics. This places higher demands on the professional knowledge background of researchers. From the results of author collaborative network research, it can be seen that authors who occupy key positions in the network often have professional education backgrounds, research experience, or professional experience outside of law, which is significantly related to the quantity of high-quality results they produce. The reason for this may be related to the general legal education model adopted abroad. If the United States regards legal education as graduate education, and Australia implements a dual degree system in university legal education, this diverse educational background gives scholars a clear advantage in interdisciplinary research. In recent years, China has advocated for the cultivation of versatile talents, such as the establishment of a Master's degree in Law (non law), which has to some extent changed the specialized training mode of legal education in China. However, the cultivation of this professional degree aims to provide talents for the practical field, and graduates often enter judicial practice positions, which makes it difficult to quickly form a positive effect on current academic research. Meanwhile, interdisciplinary research often requires researchers to have a deep understanding of two or more disciplines. Therefore, in scientific research, complementary professional knowledge is generally achieved through interdisciplinary scientific cooperation. In recent years, interdisciplinary collaborations between life sciences, physics, chemistry, and mathematics have generally resulted in co published academic papers. In traditional Chinese legal research, collaboration among scholars in the form of joint publications is relatively rare. Usually, the entire process from literature search to paper writing is completed by the author alone. This situation is caused by various reasons such as the characteristics and traditions of the discipline, differences in author viewpoints and positions, and publication requirements of mainstream journals. In legal research within the same department or across secondary disciplines, such impact may not be significant. However, under the trend of highly specialized disciplines and extensive interdisciplinary research, it may have a significant negative impact on the interdisciplinary field of computational law.

To this end, integrating different dimensions of computational law research, we should attempt to further clarify and resolve the relationships and contradictions from a macro perspective of computational law, achieve positive feedback between computational science and law, and explore the development path and model of Chinese computational law. In the construction of disciplines, we should gradually try to break through the traditional boundaries of the legal discipline and avoid the interdisciplinary misconception of computational law moving towards the combination of law and computational science. In scientific research, it is necessary to deepen substantive interdisciplinary cooperation and take the lead in research directions to prevent falling into the trap of technologism. Furthermore, to achieve the integration of computational law in China at multiple levels such as macro, meso, and micro. At the macro level, integrating problem awareness, incorporating computational thinking throughout the entire process of computational law research, focusing on the study of true propositions in real-life situations, using data to discover legal phenomena and laws, refining research perspectives, and making computation the foundation of the discipline of law; At the meso level, integrate research capabilities, innovate legal talent training models, encourage multidisciplinary legal researchers to participate in computational law research, support interdisciplinary research cooperation and the formation of results, and enable dialogue between data and legal issues; At the micro level, integrate research methods, introduce mature methods from other disciplines based on the different needs and orientations of empirical research, achieve full process coverage of data collection, processing, and analysis, and empower legal research with technology.