Abstract: In China’s judicial practice, traditional forms of evidence and evidentiary procedures are gradually proving insufficient to meet the demands of solving complex and challenging cases. As a result, big data evidence has become a key factor in cracking such cases. Whether big data can be admitted as evidence and used as a basis for establishing the facts of a case requires an in-depth examination of its evidentiary characteristics. It is crucial not to sever its connection with traditional types of evidence, nor to remain rigidly bound by the review rules applicable to conventional evidence. Currently, neither the academic community nor the practical legal community in China has granted this new form of evidence legal status. Therefore, the issues arising in judicial practice deserve further scholarly investigation.

Keywords: Big data evidence; probative value; relevance; examination and judgment

Introduction

Currently, the big data forensics process is invariably embedded in the external environment of the big data era. This necessitates using machine-learning-based source code analysis to assess security and, based on this analysis, determine the relevance and authenticity of the big data forensics process. At the same time, it is essential to establish a foundational system for big data forensics review by optimizing the data classification framework, perfecting electronic information authentication mechanisms, and building a comprehensive evidence-admissibility system for big data.

I. Raising the Issue

In recent years, big data mining and artificial intelligence technologies have been thriving. A growing number of scientifically validated findings generated through methods such as big data analytics and AI are gradually moving from “behind the scenes” to the “forefront.” As a result, fact-finding is increasingly trending toward automation and智能化. Big data evidence refers to evidence materials—obtained through the collection, preservation, and analysis of massive amounts of information using big data techniques—that can help establish the truth about an event. Such information cannot be effectively studied and analyzed by human beings alone. Big data evidence represents a brand-new type of data that has emerged in the context of this new technological revolution. Consequently, big data evidence is characterized by its reliance on vast quantities of information. Basic , powered by intelligent algorithms Basic characterized by subjective decision-making. Currently, big data evidence has begun to appear in judicial practice abroad as well. For example, in a murder case in Connecticut, USA, a man claimed that the woman’s husband had been shot while trying to fend off an intruder who had broken into their home. However, the FitBit worn by the wife at the time confirmed that she was actually taking a walk in the suburbs, thus contradicting the man’s account. Subsequently, the man was charged with murder. In China, some people’s courts have incorporated the results of biometric systems into their judicial documents. For instance, one court judgment stated: “Based on the facial recognition technology and advanced computational methods provided by Zhejiang Hikvision Digital Technology Co., Ltd.’s system, if the facial similarity rate exceeds 95 percent, the two faces can be deemed to belong to the same person, thereby identifying Wang Moumou as the suspect in this case.” Nevertheless, the factual verification of big data evidence also raises certain issues, primarily the following two: First, human error. The accuracy of big data analysis and evidence collection largely depends on the machine learning program codes that generate such evidence. Since these computer programs are created and used by human programmers, their actual operation is driven by human commands. However, what they reflect are the factual assumptions and functional orientations embedded by the language designers. For example, in the COMPAS sentencing system in California, USA—a system frequently cited in discussions about smart justice—its exposure of racial bias has drawn particular criticism from us. Second, linguistic distortion. Due to the autonomy of machine learning technologies, the information they process can be reprocessed, leading to information loss. As some experts in the tech industry have pointed out, even photos taken with Huawei’s P30 Pro may undergo AI-based image processing, which can reduce the accuracy of large volumes of data presented in audio and video formats. Such information loss can become a source of errors in fact-finding. However, if we wish to properly apply big data forensics and thereby reduce the likelihood of actual identification errors, we must first conduct an in-depth study of the fundamental technical principles underlying big data forensics. This will enable us to identify the basic types of evidence involved in big data forensics and, on this basis, establish a fundamental review system and corresponding regulatory framework for big data forensics.

II. Technical Principles of Big Data Evidence

(1) Description of the principles of big data technology

Currently, the development of computers is largely driven by vast amounts of data. As big data continues to grow, the behavioral outcomes of artificial intelligence will increasingly resemble human behavior. Moreover, big data processing technologies have endowed computers with powerful capabilities for information storage and computation. Consequently, some scholars have even begun referring to big data information technology as “artificially intelligent information.” However, to ensure consistency with the current understanding of big data—both in practical applications and academic discourse—we have deliberately avoided defining it specifically as either “big data analytics technology” or “artificial intelligence technology.” Instead, we uniformly refer to it as “big data mining technology.” The advancement of big data is also inseparable from the interplay between big data technologies and computational power: the former serves as the “fuel” that underpins the development of modern big data processing technologies, while the latter acts as the “engine” driving these technologies forward. Together, they form the technological infrastructure upon which modern big data processing relies. Furthermore, from a computational perspective, machine computation (algorithms) belongs to the specialized fields of mathematics and machine design. At its core, an algorithm is essentially a process designed to solve a specific problem—but this process is inherently characterized by finiteness and determinacy. Different algorithms choose different directions for data processing; thus, the primary goal or content of any powerful data-processing technology is ultimately determined by the computer itself.

(2) Supporting Mechanisms for the Review of Big Data Evidence

The review system outlined above represents the ideal model for reviewing big data evidence in China. However, to translate this conceptual framework into practical implementation, a corresponding policy framework will also be needed.

The establishment of a system for classifying types of evidence serves as the logical foundation for China’s new rules on evidence. Consequently, before the issuance of the “Interpretation of the New Criminal Procedure Law,” such forms of proof as product appraisal and audit reports, as well as financial audit reports—widely used in judicial practice—did not fall within the scope of legally recognized evidence and thus could not be incorporated into the evidentiary framework. The emerging big-data forensic techniques spurred by the current technological revolution are no exception; they too face the same challenge in determining the appropriate legal classification of evidence types. Therefore, to overcome this hurdle, we must break free from the current constraints imposed by the established system of legal evidence types and explore new approaches to big-data forensics that better reflect the realities of judicial practice.

First, it is essential to place great importance on establishing and perfecting data review standards, thereby preventing an indiscriminate focus on the types of data involved in large-scale digital forensics. To achieve this goal, we should take as our approach the development of evidentiary theories within the civil law system. In the civil law system’s method of proof, “proof” encompasses both the evidentiary materials and the methods of proof. Evidentiary materials refer to all information or raw data that could possibly be relevant to the facts to be proven, whereas the method of proof represents one among various verification techniques. Therefore, in evidence law, “statutorily defined” facts should not be understood merely as the totality of evidentiary information carried by a single evidentiary method. Rather, given that every person, place, and event has the potential to become consultative information or data directly or indirectly related to the criminal facts under investigation, such sources must serve as the primary investigative tool for examining various forms of evidence. Second, we should adopt judicial administrative measures to create reasonable channels for regulating big-data evidence. While assessing the reliability of big-data evidence may involve delving into the source code of machine learning algorithms, current regulations governing big-data review do not yet provide any stringent provisions for examining such source code. Nevertheless, this issue can still be addressed through judicial interpretations and guiding judgments.

China is still in the developmental stage of its evidentiary system. Faced with this new paradigm of proof emerging from the technological revolution, we must squarely acknowledge that while previous challenges remain unresolved, new difficulties have already arisen. This reality—where old and new problems overlap—is precisely what calls for adjustments and innovations in our big-data evidence review system. On the one hand, big-data evidence cannot exist in a vacuum, detached from today’s electronic information society. Given the close interconnections among electronic data, existing electronic information review systems must be comprehensively enhanced. At the same time, the encroachment of emerging technologies such as machine learning on the judicial authentication industry inevitably brings about social and ethical issues—including the risk of losing big-data evidence. Therefore, we need to reflect upon and explore new approaches to auditing the source code of machine-learning algorithms, including various verification methods such as data classification, evidence disclosure, and courtroom cross-examination, so as to establish a comprehensive and robust review system for the authenticity of big-data evidence.

Conclusion

In summary, as the gap widens between what people perceive through their senses and the true reality that they can only glimpse with the aid of tools beyond the reach of their sensory faculties, the role of human senses in the real world is gradually diminishing. Today, numerous medical evidences—such as DNA sequencing and blood-alcohol-content tests—that once dominated courtroom proceedings have given way to computer programs that transcend human sensory experience. Moreover, fact-finders increasingly rely on various “truth machines” to ascertain the truth about events.

References:

[1] Li Yuanyuan. A Study on the System for Collecting Electronic Data Evidence in Criminal Proceedings [D]. Guangxi Normal University, 2022.

[2] Cheng Long. On the Formalization and Substantiation of Adjudication of Big Data Evidence[J]. Politics and Law, 2022(05):96-114.

[3] Ni Chunle, Chen Bowen. A Study on the Mechanism of Criminal Procedure Application of Big Data Evidence [J]. Journal of the People’s Public Security University of China (Social Sciences Edition), 2022(02):37-49.

[4] Liu XiuHua. A Study on the Application Rules of Big Data Evidence in Criminal Justice [D]. Southwestern University of Finance and Economics, 2022.

[5] Wu Feixue. A Study on the Legal Control of Big Data Investigation [D]. Southwest University of Political Science and Law, 2021.