Research on Big Data and the Transformation of Investigation Mode (1)

Research on Big Data and the Transformation of Investigation Mode (1)_ Data Analyst Training

Big data is widely used in the West in presidential election prediction, business marketing, disease prevention, financial analysis, education reform, social monitoring and prediction, public safety management, terrorist attacks and so on.

The application of a large amount of data in criminal investigation and control began at 1994. A new public safety information management system, namely CompStat (Computer Statistics for short), was put into use by new york Police Department. CompStat determines the allocation of police resources, crime prevention and countermeasures by comparing statistical reports [5]. With the advent of the era of big data, the West has vigorously built a crime investigation and control system driven by big data. The crime detection and control system driven by big data uses big data to help the police analyze historical cases and discover crime trends and patterns; Predict crime by analyzing urban data sources and social network data; Use big data to optimize the allocation of police resources, thereby improving the level of social public security [6]. Big data has fundamentally changed the mode of crime investigation and control, and using big data to improve the ability of crime investigation and control is the future development direction.

Guo Shengkun, Minister of Public Security, stressed the need to strengthen the ability and level of maintaining public safety and serving the people in the era of big data [7]. Public security organs at all levels in China have begun to consciously use big data to promote crime investigation and control. However, big data is not only a technical problem, but also brings about changes in the concept, method and mechanism of investigation. The academic research in China mainly focuses on the application research of big data technology, and the research on the change of investigation concept, method and mechanism brought by big data is less and not deep enough, which urgently needs more systematic and in-depth research.

First, the complex crime situation and the digital ecology of crime in the era of big data.

At present, the crime situation is more severe and complicated. First of all, the total amount of crimes is large and the crime rate is increasing year by year. According to statistics, in just 20 12 years, the public security organs filed 655 1440 criminal cases, and the procuratorial organs approved and decided to arrest 680,539 criminal suspects, with a number of 986,056 [8]. In the past two decades, the crime rate in China has increased year by year, and the number of criminal cases has increased by more than 22% annually on average, exceeding the growth of the national GDP. Secondly, the intelligence of crime. Crime is a kind of social existence, and the development of science permeates all aspects of crime, improving its ability and harm. This is manifested in two aspects: first, the crime committed by scientific thinking is mainly manifested in strict criminal thinking, careful deployment and planning before the crime, and scientific thinking and strategies are permeated in the process of crime. The second is science and technology crime, especially digital crime. Take the use of cyber crime as an example. In 20 12 years, public security organs across the country cracked more than118,000 cyber crime cases and arrested more than 216,000 criminal suspects. The Norton Security Report released by Symantec in September 20 12 shows that since July 2012, more than 257 million people in China have become victims of cyber crimes. The direct economic loss caused by cyber crime reached 289 billion yuan, and the direct economic loss suffered by the victims per capita was about 1200 yuan [9]. Third, the complexity of crime time and space. With the development of modern science and technology, the time of crime is nonlinear, the space of crime is absent, and the combination of time and space is multidimensional, diversified and arbitrary [10]. Fourth, the causality of the case is complicated. Compared with the traditional static single society, modern society is a dynamic and complex society. In a dynamic and complex society, the causal relationship is nonlinear, coupled, multifactorial and fractured, and it is often difficult to determine the causal relationship of crime.

With the development of computer and network technology, society has entered the era of big data. The era of big data is first and foremost the era of data recording. In the era of data recording, data recording becomes the default mode [1 1], and human society is under the record of data network, which is composed of ubiquitous sensors and microprocessors. Mobile phones, networks, monitoring probes, radio frequency technology and so on record our behaviors and even our thoughts everywhere. "Going out in the morning, the camera of the elevator records our travel time; Driving to work, the camera on the road records our position and speed; At work, the webpage records our browsing habits and search records, and the telephone records our networking objects and call duration; When we come home from work, shopping records define our professional identity, family background and even personality characteristics, and TV set-top boxes record our viewing habits and value tastes ... "[12]" In the digital world, we all leave electronic footprints or electronic fingerprints. " "We are in a constantly changing but increasingly closely monitored state. In fact, now our every move can find clues in a database. " [ 14] 12

Can cunning criminals become "data hermits" without exception? Being a "data hermit" means that you should be completely divorced from the modern social system, not only can't use digital products, but also can't eat "human fireworks" in a complete sense. Because modern society is almost digital, once you communicate with modern social systems, it is likely to be captured and recorded by data. However, this does not mean any elements or fragments of the criminal's specific crime, such as crime time, crime space, crime behavior, crime tools and so on. , will be directly and completely recorded and stored by data; It means that the criminal information hidden by criminals is always recorded from different sides by massive data. Even if some or even the main or key criminal elements or fragments are missing, the criminal process can be connected, analyzed, spliced or drawn through the relevant massive data from different sides. Therefore, in the era of big data, don't talk about digital crime. Even crimes committed by traditional means can be said to fall into a network recording and storage system of "justice is slow, but not leaking". Digitalization is the realistic ecology of current crime.

Second, the investigation mode driven by big data is an inevitable choice of the times.

Pattern refers to the refined and abstract standard style. The investigation mode reflects the structural relationship and operational logic of investigation elements. Investigation modes can be classified according to different standards. According to whether information technology is used in investigation, the academic circles divide the investigation mode into traditional investigation mode and information-oriented investigation mode. However, from the perspective of information theory, the essential difference between traditional investigation mode and information-led investigation mode is not whether information is used, but the fundamental difference in the way of recording, storing, extracting and analyzing information. According to the recording, storage, extraction and analysis methods of information that can be used in investigation, investigation modes can be divided into traditional investigation mode, investigation mode led by business information and investigation mode driven by big data. Academic circles generally refer to business information-led investigation mode and big data-driven investigation mode as information-led investigation mode, but they not only have different development stages (big data-driven investigation mode is developed on the basis of business information-led investigation), but also have essential differences in information types, information extraction and judgment methods. Most importantly, this difference has brought about fundamental changes in the concept, characteristics and mechanism of investigation.

The traditional investigation mode is not high in science and technology in information storage, extraction and analysis. In traditional society, human beings mainly record and store information by human brain and writing system (traditional society has developed a whole set of writing system to meet the needs of information recording, resulting in many classified writing files collected by time). For the information recording of crime, in addition to brain and text files, the crime scene also records criminal information in the form of material exchange. Therefore, the traditional main investigation methods are investigation interview (extracting information stored in the brain) and asking for written files. ② The characteristics of the storage and extraction of human brain information are: scattered on different people; The accuracy of information is poor, which is influenced not only by the external environment, but also by the information storage's own feeling ability and memory ability. Information lacks stability, and the amount and accuracy of information decay with time; Whether the information can be extracted or not, and the quality of the extraction depends first on whether the person who stores the information can be found, and then on the inquiry skills (experience) of the investigators, the expressive ability, emotions and cooperative attitude of the interviewees. The advantages of writing information in archives are high accuracy and good stability, but it has two major defects: First, it is difficult to extract. If people want to find some useful information, they must browse all the information; Although a library-style catalogue index was later established, it was still time-consuming and laborious to find it. Second, it can't provide direct criminal information. Writing files can't be a real-time record of crimes, but an after-the-fact registration after solving a case. Such files can't provide direct criminal information for crimes that need to be solved. The information analysis and judgment of traditional investigation mainly depends on the experience of investigators, and experienced investigators often become the key to solving crimes. In a word, this model has low technology content and extensive characteristics. Whether it can solve the case depends mainly on the experience and manpower invested by the investigators, not only that, but also on the luck of the investigators. This may adapt to the traditional static and single society and its crimes, but it is almost completely incompatible with the dynamic and complex society and its crimes.

Business information-oriented investigation mode is an investigation mode based on the storage, extraction and judgment of business information under the guidance of information technology. With the development of information technology, various information recording and storage devices are widely used. The recording and storage of information is no longer completely dependent on the human brain and written documents, but electronic recording, and storage devices have become the main way for human beings to record and store information. These devices replace the human brain and write documents to record human behavior and criminal behavior in real time. From the source and storage distribution, the recorded and stored information is formed in different business operations and distributed in different business information bases, such as the consumer information recorded and stored by merchants, the financial transaction information recorded and stored by banks, and the patient information recorded by hospitals. These databases lack integration and form information islands with each other. Information redundancy and information islands have become the basic ecology of information existence. As far as the investigation mode dominated by business information is concerned, its main characteristics are as follows: First, the investigation department relies on the structured database accumulated by the public security platform, which is mainly used for the verification and comparison of people, things and things, and the real-time criminal information is still mainly collected manually. Second, information extraction is still difficult. Undeniably, compared with the traditional investigation mode, the investigation mode based on business information has greatly improved the efficiency of inquiry and comparison for the structured information accumulated by public security organs. However, in the face of more and more accumulated data from different sources and structures, especially a large number of semi-structured and unstructured data, there is neither the technology and mechanism of data integration nor the technical means of information extraction. Structured data is the model before data, most of which are registered afterwards (a few are real-time recorded data such as hotel accommodation). ), so it is difficult to have real-time criminal record information, and its main value lies in the verification of people, things and things; It is these semi-structured and unstructured data from different sources that record the "clues" of crime in real time. Third, information analysis and judgment still mainly rely on the experience of investigators. Business information system is mainly used for simple query and comparison, and cannot be analyzed by intelligent algorithm. Generally speaking, in the face of the current crime situation, especially mobile phone crime and digital crime, this investigation mode is difficult to work.

Big data-driven investigation mode is based on big data and cloud computing platform, and it is an upgrade of information-led investigation mode in the era of big data. In the era of big data, the investigation mode driven by big data is an inevitable choice of the times, which lies not only in the complex crime situation and its digital ecology, but also in the fact that big data technology makes this choice a reality.

First of all, the digital ecology of crime is the realistic basis of big data-driven investigation mode. Faced with the complicated crime situation, people seem to be at a loss. To some extent, crime control is a kind of investigation technology, which has more advantages than criminal technology. However, the development of modernity makes criminals more anonymous and mobile, which once broke the advantages of public security organs, which is one of the reasons for the explosive growth of crime today. However, as a kind of social existence, crime will also provide opportunities for human beings to restrict it when the society reaches the conditions of crime. The digital ecology of crime has fundamentally changed the way of recording and storing criminal information and greatly expanded "social memory". Big data technology will completely change the contrast between investigation technology and criminal technology. Therefore, we must change the traditional investigation mode and adopt the investigation mode driven by big data to control and combat crime.

Secondly, in the era of big data, the data that investigation faces and can handle is no longer small data, but big data. Nowadays, the data that investigation faces and can handle has the characteristics of large amount of data, many types and low value density. The easiest difference between "pond" and "sea" is the scale [15]. In the past, even in the investigation stage dominated by business information, the amount of data faced or processed was equivalent to a "pond". In contrast, the amount of data faced and processed by modern investigations was an "ocean". Moreover, modern investigation faces the diversity of data: structurally, there are not only structured data, but also a large number of semi-structured and unstructured data; From the data type, there are business data, user-generated data and sensor-perceived data; From the form of data expression, there are words, pictures, audio, video, links and so on. Judging from the composition of criminal cases, there are people and their relationships, behaviors, things, time, space and subjective intentional data. The value density of data is low. In the vast amount of data, the relevant crime data is only a small "spray", but it is precious. Take video as an example. In the process of continuous monitoring, the data that may be useful is only one or two seconds [16].

Third, big data technology can extract, analyze, judge and predict the future from massive data. Big data is data whose scale or complexity exceeds that of common technologies, and it is captured and processed with reasonable cost and time limit. Big data technology based on cloud computing can break through the cost and time constraints of conventional technology. Specifically, firstly, big data technology can timely extract, analyze and process multi-structure and multi-source data, especially semi-structure and unstructured data, extract a large number of details, bits and pieces, data and information related to crimes from massive and chaotic data, and connect "data, information points and components in series" [13]29-30. For determining the identity of the suspect, perhaps only four information points are enough. Second, relying on cloud computing, big data can extract and analyze information in a reasonable time. Taking the Zhou case as an example, Nanjing police spent several days and used hundreds of police officers to search video surveillance data, while using big data technology may only take a few hours. Third, the most fundamental breakthrough of big data technology is the ability to use massive data for algorithm analysis and information research, thus helping us to understand the past, analyze the reasons and reveal the law of crime. Finally, big data can find meaningful patterns in analyzing the past, so as to predict the future and provide opportunities for us to optimize the allocation of police resources and combat crime.

Third, the concept change of big data-driven investigation mode.

Hegel pointed out that "idea is the rationality of any knowledge" [17], and thought that idea contains "expected things" and is forward-looking, instructive and designed [18]. The change of investigation mode is first of all the change of concept. The concept in the investigation mode refers to the views, opinions and beliefs that reflect the investigation law and have the ability to guide, dominate and decide the investigation activities. The investigation mode driven by big data is not only a new working mode, but also a new thinking and concept. In the era of big data, the concepts that need to be established in the survey are:

The concepts of online and open. Big data is online data first. Big data is not only massive, but also records the complex dynamic data of society in real time: data generated by users and sensed by various sensors, which are mixed with "clues" of crime. For investigation, the structured data accumulated by the public security platform is important, especially for the verification of people, things and things, but it is difficult to have real-time criminal records. Big data-driven investigation is based on the structured data accumulated by the public security platform, extracting, analyzing and processing the ever-changing data generated by users and perceived by various sensors to obtain information. Therefore, for big data-driven investigation, we must adhere to the concept of online and open data, obtain the massive data we need, and then analyze and process these data.

The concept of data-led survey. In the era of big data, data is the ecology of crime, and the investigation process is the process of data storage, extraction and analysis. Data runs through all aspects of investigation, and "let data speak" has become the basic thinking of investigation. The concept of data-led investigation includes at least the following three aspects: first, all phenomena related to crime can be digitized. Everything can be quantified and digitized [19]25-26. Not only tangible things related to crime, such as time, space, human characteristics (biological characteristics, behavior habits, etc. ), behavior, means, things, etc. It can be quantified and digitized, or it can be intangible things related to crime, such as people's values, attitudes and emotions. Secondly, big data is a basic resource and a toolbox for investigation. Investigation is the mining and analysis of data, and the success of investigation depends on the ability to extract and analyze big data resources to a certain extent; Using all kinds of analysis techniques of big data, we can get the criminal information we need. Finally, in the era of big data, data is the core of the investigation process and dominates the operation of investigation. Crime scene reconstruction, investigation decision-making, investigation approach selection, investigation analysis, data collation and investigation prediction are all carried out around data.

The concept of relevance. Big data determines the correlation by quantifying the mathematical relationship between two data values. Strong correlation means that when one data value increases, another data value is likely to increase by [3]7 1. Traditional surveys collect and analyze data according to the standards of causality and data structure. In the era of big data, we can analyze and use almost all relevant data. We don't have to stick to causality and data structure standards to collect data, but stick to correlation standards, not only collecting structured data, but also collecting semi-structured and unstructured data. Although this correlation can not directly reveal the internal causal relationship, it still has strong practical value for criminal investigation and control.

Relevance allows investigators to think and analyze cases from all directions and angles. Although relevance does not pursue accuracy, it pursues richness, does not refuse any opportunity, and tries to create and use opportunities. Through correlation, seemingly unrelated information can be linked internally, so as to have a more comprehensive understanding of the case. This may help us find clues to solve the case, clarify the thinking of solving the case and delimit the scope of the case.

Correlation can give us further guidance, determine the causal relationship, so as to determine the cause of the crime and prove the crime. Correlation analysis is the basis of causal analysis. Correlation is not necessarily causality, but causality must be highly correlated. Through correlation, we can further explore whether there is a causal relationship, thus proving the crime.

An important value of correlation is that it can monitor the crime situation. As mentioned above, at present, the causes affecting crime are complicated, and it is not easy or even impossible to determine the causes of crime. For investigators, it may be important not to find out the cause of the crime, but to control the crime. Through correlation, the related objects can be determined, and then the crime situation can be monitored, so as to effectively allocate police resources and crack down on crime.

Through correlation, crime can be predicted. The core value of big data is prediction. By collecting relevant data and establishing a big data model, we can predict when, where, who and what types of crimes may occur from a micro perspective, and also predict the crime trend from a macro perspective, which provides a better opportunity for us to prevent and combat crimes.

The concept of combining online criminal investigation with offline evidence. Big data makes it very easy to find and identify criminal suspects. But data is just a mirror image of facts, which does not mean that it is facts; ④ Moreover, there is a difference between the algorithmic logic of big data (emphasizing relevance, only determining a probability, and even causing fatal errors due to noise and other factors) and the legal proof logic (emphasizing causality and excluding reasonable doubt standards). Therefore, criminal investigation needs to be further proved according to the operational requirements of the legal system. Even if it is possible to identify criminal suspects through big data and reach the standard of excluding reasonable doubts, it is necessary to transform the algorithm system of big data into a proof system that meets the requirements of legal norms, and transform data identification into legal identification. However, there is no separation between solving crimes online and giving evidence offline. Big data can guide our proof, help us find evidence and determine causality. Therefore, in the era of big data, we can't abandon correlation, only pursue causality, but also prevent correlation from replacing causality and prediction from replacing facts.

The above is the related content of "Research on Big Data and Investigation Mode Change" (1) shared by Bian Xiao. For more information, you can pay attention to the global ivy and share more dry goods.