Search

The New York City Police Department (NYPD), the largest state or local police force in the United States, is charged with securing New York City from crime and terrorism. The NYPD’s Domain Awareness System (DAS) is a citywide network of sensors, databases, devices, software, and infrastructure that informs decision making by delivering analytics and tailored information to officers’ smartphones and precinct desktops. DAS development began in earnest in 2008; since then, the NYPD has used the system to employ a unique combination of analytics and information technology, including pattern recognition, machine learning, and data visualization. DAS is used throughout the NYPD, and the DAS software has been sold to other agencies, bringing in revenue for New York City. Through improving the efficiency of the NYPD’s staff, DAS has generated estimated savings of $50 million per year. Most importantly, the NYPD has used it to combat terrorism and improve its crime-fighting effectiveness. Since DAS was deployed department wide in 2013, the overall crime index in the city has fallen by six percent.

Amid the deployment of algorithmic techniques for security – from the gathering of intelligence data to the proliferation of smart borders and predictive policing – what are the political and ethical stakes involved in securing with algorithms? Taking seriously the generative and world-making capacities of contemporary algorithms, this special issue draws attention to the embodied actions of algorithms as they extend cognition, agency and responsibility beyond the conventional sites of the human, the state and sovereignty. Though focusing on different modes of algorithmic security, each of the contributions to the special issue shares a concern with what it means to claim security on the terrain of incalculable and uncertain futures. To secure with algorithms is to reorient the embodied relation to uncertainty, so that human and non-human cognitive beings experimentally generate and learn what to bring to the surface of attention for a security action.

The goal of predictive policing is to forecast where and when crimes will take place in the future. The idea has captured the imagination of law enforcement agencies around the world. Many agencies are purchasing software tools with the goal of reducing crime by mapping the likely locations of future crime to guide the deployment of police resources. Yet the claims and promises of predictive policing have not been subject to critical examination. This paper provides a review of the theories, techniques, and assumptions embedded in various predictive tools and highlights three key issues about the use of algorithmic prediction. Assumptions: The algorithms used to gain predictive insights build on assumptions about accuracy, continuity, the irrelevance of omitted variables, and the primary importance of particular information (such as location) over others. In making decisions based on these algorithms, police are also directed towards particular kinds of decisions and responses to the exclusion of others. Evaluation: Media coverage of these technologies implies that they are successful in reducing crime. However, these claims are not necessarily based on independent, peer reviewed evaluations. While some evaluations have been conducted, additional rigorous and independent evaluations are needed to understand more fully the effect of predictive policing programmes. Accountability:The useof predictive software can undermine the ability for individual officers or law enforcement agencies to give an account of their decisions in important ways. The paper explores how this accountability gap might be reduced.

This study takes a comparative approach to examining the contemporary police change in controlling crime in American and China. We attempt to achieve two primary objectives. First, we critically evaluate American and Chinese police innovations including the theoretical framework, police operations, and associated effectiveness. Second, we attempt to identify the utility of innovative strategies that can be shared by both sides. Five innovative strategies adopted by American police agencies in the past 25 years are summarized and reviewed. On the other side, the primary strategy utilized in crime control is Chinese style of community policing. After comparison of the policing strategies in the two countries, it is self evident that the implementation of community policing in China is broader and more in-depth than American police. Chinese police organizations can benefit from the implementation of problem-oriented policing accumulated from their American counterpart. Finally, we argue that hot spots policing and predictive policing do not have significant utility in the Chinese setting.

From ‘connecting the dots’ and finding ‘the needle in the haystack’ to predictive policing and data mining for counterinsurgency, security professionals have increasingly adopted the language and methods of computing for the purposes of prediction. Digital devices and big data appear to offer answers to a wide array of problems of (in)security by promising insights into unknown futures. This article investigates the transformation of prediction today by placing it within governmental apparatuses of discipline, biopower and big data. Unlike disciplinary and biopolitical governmentality, we argue that prediction with big data is underpinned by the production of a different time/space of ‘between-ness’. The digital mode of prediction with big data reconfigures how we are governed today, which we illustrate through an analysis of how predictive policing actualizes between-ness as hotspots and near-real-time decisions.

To serve and protect: that is the mandate of the modern police force. However, as they acquire more tools for processing information, the notion of serving and protecting necessarily evolves. What are the digital tools necessary to properly protect the public, and how can we ensure that the use of these tools ensures that the police serve the public and not the other way around?

In 2013, the Chicago Police Department conducted a pilot of a predictive policing program designed to reduce gun violence. The program included development of a Strategic Subjects List (SSL) of people estimated to be at highest risk of gun violence who were then referred to local police commanders for a preventive intervention. The purpose of this study is to identify the impact of the pilot on individual- and city-level gun violence, and to test possible drivers of results. The SSL consisted of 426 people estimated to be at highest risk of gun violence. We used ARIMA models to estimate impacts on city-level homicide trends, and propensity score matching to estimate the effects of being placed on the list on five measures related to gun violence. A mediation analysis and interviews with police leadership and COMPSTAT meeting observations help understand what is driving results. Individuals on the SSL are not more or less likely to become a victim of a homicide or shooting than the comparison group, and this is further supported by city-level analysis. The treated group is more likely to be arrested for a shooting. It is not clear how the predictions should be used in the field. One potential reason why being placed on the list resulted in an increased chance of being arrested for a shooting is that some officers may have used the list as leads to closing shooting cases. The results provide for a discussion about the future of individual-based predictive policing programs.

The “near-repeat” effect is a well-known criminological phenomenon in which the occurrence of a crime incident gives rise to a temporary elevation of crime risk within close physical proximity to an initial incident. Adopting a social network perspective, we instead define a near repeat in terms of geodesic distance within a criminal social network, rather than spatial distance. Specifically, we report a statistical analysis of repeat effects in arrest data for Chicago during the years 2003–2012. We divide the arrest data into two sets (violent crimes and other crimes) and, for each set, we compare the distributions of time intervals between repeat incidents to theoretical distributions in which repeat incidents occur only by chance. We first consider the case of the same arrestee participating in repeat incidents (“exact repeats”) and then extend the analysis to evaluate repeat risks of those arrestees near one another in the social network. We observe repeat effects that diminish as a function of geodesic distance and time interval, and we estimate typical time scales for repeat crimes in Chicago.

International audience; This article present an original study from the research team Louis Josserand regarding the acceptability of predictive policing from citizens, mayors and police officers and how this new technology will impact them. This project was supported by the French ministry of Interior and was never studied before.

We take two professions — police officers and doctors — and place their experiences with big data in dialogue. Policing and medicine, while naturally different in some obvious respects, actually both need to grapple with a lot of the same moral, social, and legal questions that come with adopting big data programs. This because, as we discuss below, both professions generally possess a monopoly over an acute societal vulnerability, be it safety or health, and have accordingly developed a set of settled internal norms to shape individual discretion in service of each respective function. We place the professions side-by-side and try to distill certain insights from the perspective of three key stakeholders — practitioners, policymakers, and the polity.

Community Policing (CP), which refers to communities’ participation in police actions, is an evolving topic having a potential impact on reducing crime and creating safer communities. The next generation of community policing (NGCP) is helping to transform CP into a more robust and active participatory process than what has been previously witnessed. There is a connection between the multilayered complexity of CP that is being fuelled in this engagement through the current global digital revolution, exponentially growing big data, and advancement in technology. In the global scene, NGCP is received differently given the cultural, societal, legal, privacy and security aspects in specific continents, countries, regions, cities and even neighbourhoods. This chapter aims at introducing challenges of NGCP at a global stage and offer options for moving forward. To do this, the chapter sets the scene of NGCP in Europe and North America (with focus on Canada). The role of digitalization, big data in predictive policing and digital community policing are also discussed. In particular, the chapter tries to leverage predictive policing from a community engagement perspective that effectively addresses and promotes the seamless collaboration between police and communities. The “Block Watch” programme in Vancouver Police Department (VPD) and European Neighbourhood Watch Association (EUNWA) in Europe are introduced in the chapter as vital role models in the NGCP. The role of technologies and the interdependency on making the right technological choices with the goal of advancing NGCP and then incorporating human interactivity are explored, specifically, by examining, on the one hand, how the Vancouver Police Department (VPD) has operationalized predictive policing to integrate into the community engagement perspective and, on the other hand, by studying the successful case of Burgernet across all of The Netherlands. European research and innovation project INSPEC2T is also presented as solution for NGCP with contributions from around the globe.

Smart solutions that make use of machine learning and data analyses are on the rise. Big Data analysis is attracting more and more developers and researchers, and at least five requirements (Velocity, Volume, Value, Variety, and Veracity) show challenges in deploying such solutions. Across the globe, many Smart City initiatives are using Big Data Analytics as a tool for doing predictive analytics which can be helpful to human well being. This work presents a generic architecture named Machine Learning in Microservices Architecture (MLMA) that provides design patterns to transform a monolithic architecture of machine learning pipelines in microservices with separate roles. We present two case studies deployed to a Smart City initiative, where we discuss how each component of the architecture applied in specific applications that use predictions with machine learning. Among the benefits of this architecture, we argue prediction performance, scalability, code maintenance and reusability makes such transition a natural trend in Big Data and machine learning applications.

How can we secure an accessible and open democratic debate about police use of predictive analytics when the technology itself is a specialized area of expertise? Police utilize technologies of prediction and automation where the underlying technology is often a machine learning (ML) model. The article argues that important issues concerning ML decision models can be unveiled without detailed knowledge about the learning algorithm, empowering non-ML experts and stakeholders in debates over if, and how to, include them, for example, in the form of predictive policing. Non-ML experts can, and should, review ML models. We provide a ‘toolbox’ of questions about three elements of a decision model that can be fruitfully scrutinized by non-ML experts: the learning data, the learning goal, and constructivism. Showing this room for fruitful criticism can empower non-ML experts and improve democratic accountability when using ML models in policing.

Forecasting when and where crimes are more likely to occur based on years of historical record analysis is becoming a task which is increasingly helping cities' safety departments with capacity planning, goal setting, and anomaly detection. Crime is a geographically concentrated phenomena and varies in intensity and category over time. Despite its importance, there are serious challenges associated with producing reliable forecasts such as sub-regions with sparse crime incident information. In this work, we address these challenges proposing a crime prediction model which leverages features extracted from time series patterns of criminal records based on spatial dependencies. Our results benchmarked against the state of the art and evaluated on two real world datasets, one from San Francisco, US, and another from Natal, Brazil, show how crime forecasting can be enhanced by leveraging Spatio-Temporal dependencies improving our understanding of such models.

In the past years, artificial intelligence has received increased attention in legal scholarship. Topics such as self-driving cars, predictive policing or discriminatory profiling are only a few examples of trending legal scholarship, mirroring similar academic developments in other scientific disciplines – especially in computer science research. Interestingly, and perhaps not surprisingly, several of the same research questions asked in relation to artificial intelligence have been posed in earlier research decades ago, albeit with more emphasis on normativity. As legal research does not often make use of self-standing literature reviews, and given the ever-growing body of legal literature available on artificial intelligence, it is getting more and more difficult to have an overview of existing work on the subject, which may lead to repetitive research. This chapter presents the first part of a project aimed at making such a research overview. It does so by using a dataset of 3931 academic journal articles obtained from HeinOnline referencing the topic of artificial intelligence (hereinafter Corpus). HeinOnline, one of the largest legal databases in the world, indexes sufficient publications to shape a reliable knowledge map of existing research that refers to artificial intelligence: this leaves the possibility for additional observations to be added from other sources (e.g. other databases, other publishers, etc.). In the first section, we describe the Corpus, including the methodology used in obtaining it and the characteristics of the publications therein. The second section comprises visualisations of the Corpus using descriptive statistics, as well as examples of thematic clusters, which are subsequently discussed. Additionally, it briefly explains the statistical model (Latent Dirichlet Allocation Topic Modelling) to be deployed in the second part of the project. The third section reflects upon the meaning and future of legal research on artificial intelligence, and is followed by the conclusion of the chapter. Doing research on research is a remarkable opportunity for legal scholars to zoom out of a given field of law and observe it not only spatially but also temporally. Time plays an important and often overlooked role, especially in research that focuses on the socio-legal implications of complex technologies. By placing legal debates in a broader timeline, it is possible to identify patterns (allowing for a faster understanding of the socio-legal phenomena central to legal research on artificial intelligence), and to provide a checklist for questions both that have and have not yet been answered.

Crime rate is increasing significantly over the years. Crime prevention is the attempt to reduce and deter the crimes and the criminals. The government must go beyond law enforcement and criminal justice to tackle the risk factors that cause crime because it is more cost effective and leads to greater social benefits. The data driven method is used which is based on the broken windows theory, having an enormous impact on the working of the police department. The theory links disorder and incivility within a community to subsequent occurrences of serious crimes. Predictive policing is used by the law enforcement stakeholders for taking proactive measures against crimes. This will help the police departments to efficiently focus their resources on the potential crime hotspots. The model is built to predict the crime rate based on demographic and economic information of particular localities using decision trees, linear classification, regression and spatial analysis.

Recent revelations concerning data firm Cambridge Analytica’s illegitimate use of the data of millions of Facebook users highlights the ethical and, relatedly, legal issues arising from the use of machine learning techniques. Cambridge Analytica is, or was – the revelations brought about its demise - a firm that used machine learning processes to try to influence elections in the US and elsewhere by, for instance, targeting ‘vulnerable’ voters in marginal seats with political advertising. Of course, there is nothing new about political candidates and parties employing firms to engage in political advertising on their behalf, but if a data firm has access to the personal information of millions of voters, and is skilled in the use of machine learning techniques, then it can develop detailed, fine-grained voter profiles that enable political actors to reach a whole new level of manipulative influence over voters. My focus in this paper is not with the highly publicised ethical and legal issues arising from Cambridge Analytic’s activities but rather with some important ethical issues arising from the use of machine learning techniques that have not received the attention and analysis that they deserve. I focus on three areas in which machine learning techniques are used or, it is claimed, should be used, and which give rise to problems at the interface of law and ethics (or law and morality, I use the terms “ethics” and “morality” interchangeably). The three areas are profiling and predictive policing (Saunders et al. 2016), legal adjudication (Zeleznikow, 2017), and machines’ compliance with legally enshrined moral principles (Arkin 2010). I note that here, as elsewhere, new and emerging technologies are developing rapidly making it difficult to predict what might or might not be able to be achieved in the future. For this reason, I have adopted the conservative stance of restricting my ethical analysis to existing machine learning techniques and applications rather than those that are the object of speculation or even informed extrapolation (Mittelstadt et al. 2015). This has the consequence that what I might regard as a limitation of machine learning techniques, e.g. in respect of predicting novel outcomes or of accommodating moral principles, might be thought by others to be merely a limitation of currently available techniques. After all, has not the history of AI recently shown the naysayers to have been proved wrong? Certainly, AI has seen some impressive results, including the construction of computers that can defeat human experts in complex games, such as chess and Go (Silver et al. 2017), and others that can do a better job than human medical experts at identifying the malignancy of moles and the like (Esteva et al. 2017). However, since by definition future machine learning techniques and applications are not yet with us the general claim that current limitations will be overcome cannot at this time be confirmed or disconfirmed on the basis of empirical evidence.

[Extract] In this chapter, we focus upon the post-sentence consequences that have flowed from the growth of preventive and data-driven policing in Australia. The existence of a past conviction can result in entry into a complex web of supervisory mechanisms, including the deprivation of liberty and incarceration, according to opaque procedures of risk assessment. As discussed throughout, Australia’s ‘jurisprudence of risk’ extends from predictive policing of those with an existing criminal record to extensive schemes of preventive detention and supervision. The ‘risk-based’ consequences of conviction also extend to the immigration system, where upon completion of their sentence, offenders also face possible preventive deportation. We argue that the use of opaque risk-centred criteria threatens to displace the principle of proportionality, resulting in a diminution of fundamental rights. The chapter offers an analytical framework for understanding how these systems have taken root and expanded, and how they might be reformed.

Part 2: Social Implications of Algorithmic Phenomena; International audience; This paper reports on an ethnographic study of the use of analytics in police work. We find that the introduction of predictive policing was followed by the emergence of the new occupational role of “intelligence officer”. While intelligence officers were initially intended to merely support police officers by making sense of algorithmic outputs, they became increasingly influential in steering police action based on their judgments. Paradoxically, despite the largely subjective nature of intelligence officers’ recommendations, police officers started to increasingly believe in the superiority and objectivity of algorithmic decision-making. Our work contributes to the literature on occupational change and technology by highlighting how analytics can occasion the emergence of intermediary occupational roles. We argue that amidst critical debates on subjectivity of analytics, more attention should be paid to intermediaries – those who are in-between designers and users – who may exert the most consequential influence on analytics outcomes by further black-boxing the inherent inclusion of human expertise in analytics.

Technological development in every aspect of human life has formed a wider analytical approach to crime. The genesis and structure of crime, its intensity, and dynamics are subjects of intense scientific studies carried out by researchers in various fields of science. Increasing possibilities to track crime events give public organizations and police departments the opportunity to collect and store detailed data, including spatial and temporal information. At the same time, the crowd-sourced open datasets as social media and Internet datasets can be a valuable source of knowledge about various behavior patterns and social phenomena, including those of criminal nature. Thus, exploratory analysis and data mining become an important part of the current methodology for the detection and forecasting of crime development. The ability to use data analysis tools to extract useful information related to criminal events enables law enforcement agencies to more efficiently allocate their resources to specific crime areas. It allows the effective deployment of officers to high-risk crime areas and elimination from areas with a decreasing crime trend as well as developing effective crime prevention strategies. The purpose of this paper is to show the usefulness of analytic algorithms in predicting crimes, however, there are other applications of such analyzes in the area of law enforcement, such as defining criminal hot spots, creating criminal profiles, and detecting crime trends. The most important factor is the accuracy with which one can infer and create new knowledge based on observations from the past that will be useful in the process of reducing the number of crimes (predictive policing) and ensure the security of citizens.

Z napovednim policijskim delom, ki se izvaja s pomočjo sodobnih varnostnih tehnologij, je možno napovedati kdaj in kje naj bi se zgodilo naslednje kaznivo dejanje. Napovedi se izračunavajo na podlagi podatkov o preteklih kaznivih dejanjih in drugih dejavnikov, ki vplivajo na kriminaliteto. Varnostni subjekti po vsem svetu uporabljajo tehnologije za napovedovanje kriminalitete, da bi rešili kadrovske in finančne težave, s katerimi se soočajo že vrsto let. V teoretičnem delu diplomskega dela je predstavljen koncept napovednega policijskega dela, zatem pa še metodologije, ki se uporabljajo za napovedovanje kriminalitete ter normativni okviri v povezavi z varovanjem zasebnosti in varstvom osebnih podatkov. Da bi se varnostni subjekti pri izvajanju napovednega policijskega dela izognili prekomernim posegom v človekove pravice je pri uporabi tehnologij nujno potrebno zagotoviti ustrezno zasebnost in normativno skladno obdelavo podatkov. Teoretični del diplomskega dela nadalje vključuje tudi študije primerov uporabe tehnologij za napovedovanje kriminalitete v praksi. Empirični del diplomskega dela je sestavljen iz večnivojske raziskave. Z anketo izvedeno med splošnim prebivalstvom Slovenije, smo ugotavljali koliko anketiranci poznajo in podpirajo uporabo napovednega policijskega dela v Sloveniji. Izkazalo se je, da so te tehnologije večini anketirancev nepoznane. S kratkimi strukturirani intervjuji strokovnjakov na področju napovednega policijskega dela smo ugotovili, da so tovrstne tehnologije najprimernejše za premoženjska kazniva dejanja, so pa njihova mnenja glede prihodnosti zelo deljena. V zadnjem delu raziskave smo izvedli še intervju z vodjo policijskega oddelka za ocenjevanje varnostnih tveganj, ki je pojasnil, da obstajajo potenciali za uporabo napovedanega policijskega dela v Sloveniji, vendar slovenska policija tovrstnih aktivnosti še ne izvaja. With the help of predictive policing and by applying analytical techniques, it is possible to predict when and where the next crime is supposed to take place. The predictions are generated on the basis of information regarding past crimes and other factors that have an effect on criminality. Law enforcement agencies all around the world are implementing predictive policing technologies in order to overcome the financial and human resource problems, which they have been plagued by for years. The theoretical part of the thesis describes the history and meaning of the concept of predictive policing itself, as well as methodologies that are used to predict crime. In order to avoid excessive interference with human rights, it is necessary to ensure the privacy of people and normative compliant processing of data. The theoretical part of the thesis also includes practical examples of the use of predictive policing. The empirical part of the thesis is composed of a multi-tier research. With a survey conducted among the general population of Slovenia, we determined how many respondents are familiar with predictive policing technologies and whether they support its use in Slovenia. Results showed that the aforementioned technologies are unknown to the majority of the respondents. With short structured e-mail interviews of professionals in the area of predictive policing, we confirmed that crime prediction technologies are best suited for property crimes, but their opinions on the future of predictive policing are divided. In the last part of the research, we conducted an interview with the head of the police department for the assessment of security risks, explaining that there are potentials for the use of predictive policing in Slovenia, but they are not using them yet.

Racial bias of predictive policing algorithms has been the focus of recent research and, in the case of Hawkes processes, feedback loops are possible where biased arrests are amplified through self-excitation, leading to hotspot formation and further arrests of minority populations. In this article we develop a penalized likelihood approach for introducing fairness into point process models of crime. In particular, we add a penalty term to the likelihood function that encourages the amount of police patrol received by each of several demographic groups to be proportional to the representation of that group in the total population. We apply our model to historical crime incident data in Indianapolis and measure the fairness and accuracy of the two approaches across several crime categories. We show that fairness can be introduced into point process models of crime so that patrol levels proportionally match demographics, though at a cost of reduced accuracy of the algorithms.

Communities are adversely affected by heterogeneous social harm events (e.g., crime, traffic crashes, medical emergencies, drug use) and police, fire, health and social service departments are tasked with mitigating social harm through various types of interventions. Smart cities of the future will need to leverage IoT, data analytics, and government and community human resources to most effectively reduce social harm. Currently, methods for collection, analysis, and modeling of heterogeneous social harm data to identify government actions to improve quality of life are needed. In this paper we propose a system, CDASH, for synthesizing heterogeneous social harm data from multiples sources, identifying social harm risks in space and time, and communicating the risk to the relevant community resources best equipped to intervene. We discuss the design, architecture, and performance of CDASH. CDASH allows users to report live social harm events using mobile hand-held devices and web browsers and flags high risk areas for law enforcement and first responders. To validate the methodology, we run simulations on historical social harm event data in Indianapolis illustrating the advantages of CDASH over recently introduced social harm indices and existing point process methods used for predictive policing.

Life using AI becomes more prevalent in the next few decades. Keeping people safecould be the major challenge in coming days. As neural networks are increasinglyutilized to capture and process data, we look at the benefits and dangers of futureAI systems used in policing and security. Predictive Policing is an intelligenceled policing which focus on research based approaches to take another large stepaway from community policing. It provides the relevant information to describe,understand and foresee crime at different scales: operational, tactical and strategicevents. The aim is to simplify decision-making. It makes sure that certain type of Law-enforcements should respond to the potentially lethal crime scenes. This enablespolice agencies to provide strategic deployment of resources to respond effectivelyto anticipated crimes. Predictive policing is a software which uses machine learning algorithm to calculate its predictions. The algorithm’s trained for each new city’shistorical datasets, then it updates every-day with new events received from thedepartment. AI isn’t going to provide a quick fix for our Justice-system but it’s goingto simplify the problems

Settings such as lending and policing can be modeled by a centralized agent allocating a resource (loans or police officers) amongst several groups, in order to maximize some objective (loans given that are repaid or criminals that are apprehended). Often in such problems fairness is also a concern. A natural notion of fairness, based on general principles of equality of opportunity, asks that conditional on an individual being a candidate for the resource, the probability of actually receiving it is approximately independent of the individual's group. In lending this means that equally creditworthy individuals in different racial groups have roughly equal chances of receiving a loan. In policing it means that two individuals committing the same crime in different districts would have roughly equal chances of being arrested. We formalize this fairness notion for allocation problems and investigate its algorithmic consequences. Our main technical results include an efficient learning algorithm that converges to an optimal fair allocation even when the frequency of candidates (creditworthy individuals or criminals) in each group is unknown. The algorithm operates in a censored feedback model in which only the number of candidates who received the resource in a given allocation can be observed, rather than the true number of candidates. This models the fact that we do not learn the creditworthiness of individuals we do not give loans to nor learn about crimes committed if the police presence in a district is low. As an application of our framework, we consider the predictive policing problem. The learning algorithm is trained on arrest data gathered from its own deployments on previous days, resulting in a potential feedback loop that our algorithm provably overcomes. We empirically investigate the performance of our algorithm on the Philadelphia Crime Incidents dataset., The short version of this paper appears in the proceedings of ACM FAT*-19

We present some results from a joint project between HP Labs, Cardiff University and Dyfed Powys Police on predictive policing. Applications of the various techniques from recommender systems and text mining to the problem of crime patterns recognition are demonstrated. Our main idea is to consider crime records for different regions and time period as a corpus of text documents with words being crime types. We apply tools from NLP and text documents classifications to analyse different regions in time and space. We evaluate performance of several measures of similarity for texts and documents clustering algorithms.

Prispevek obravnava izbrane ekonomske in pravne vidike velikega podatkovja in analizira novejše trende oz. mnenja o potrebnih pristopih k pravni regulaciji tega področja, ki je tesno zvezano s hitrim tehnološkim napredkom. Temeljna teza prispevka je, da razmeroma enovitega in podatkov polnega virtualnega sveta oz. kibernetskega prostora ne bo mogoče uspešno regulirati parcialno po posameznih pravnih področjih, ampak bo tradicionalno delitev pravnih področij treba preseči z novim krovnim (mednarodnim in široko sprejetim) pravnim aktom oz. področjem, ki bo določilo osnovna pravila in pravna načela, po katerih se bodo nato lahko ravnale tudi posamezne pravne discipline. Če želi pravo igrati aktivno in konstruktivno vlogo v spremenjenih družbenih razmerah ter biti dolgoročno upošteven dejavnik oz. razumna protiutež tehnološkemu razvoju, je potreben sistemski pristop na globalni ravni. Izziv je torej velik, a tudi težav ne manjka. Ena bistvenih je čas - vprašanje je, ali bo pravo še zmožno obvladovati tehnologijo, ko bo (končno) dojelo, za kaj gre. Article outlines some economic and legal aspects of big data, and provides analysis of recent trends and opinions on necessary approaches to the legal regulation of this area which is tightly connected with fast technological progress. Article sets essential thesis that virtual world or cyberspace - that is relatively uniform and full of data - cannot be constructively regulated partially in each separate legal field. Instead, the traditional division of legal fields or disciplines should be exceeded by a new general (international and commonly accepted) legal act or discipline that will set up basic legal rules and principles to be then adopted in different legal fields. If law aspires to play active and constructive role in new social environment and to be a factor of lasting importance or reasonable counterpoise to technological progress, systemic approach on global level is essential. Therefore, challenges are big, but there are troubles as well. One of the important ones is time – i.e. the question whether the law will be able to master technology when it (finally) realizes what the point is.

While applications of big data have been extensively studied, discussion is mostly made from the perspectives of computer science, Internet services, and informatics. Alternatively, this article takes the big data approach as an institutional innovation and uses the problem of illegal subdivided units (ISUs) in Hong Kong as a case study. High transaction costs incurred in identification of suspected ISUs and associated enforcement actions lead to a proliferation of ISUs in the city. We posit that the deployment of big data analytics can lower these transaction costs, enabling the government to tackle the problem of illegal accommodations. We propose a framework for big data collection, analysis, and feedback. As the findings of a structured questionnaire survey reveal, building professionals believed that the proposed framework could reduce transaction costs of ISU identification. Yet, concerns associated with the big data approach like privacy and predictive policing were also raised by the professionals.

Epistemic injustices are the distinctly epistemic harms and wrongs which undermine or depreciate our capacities knowers. This dissertation develops a theory of epistemic injustice and justice which accounts for excesses in epistemic goods as a source of epistemic injustice. This is a theory of epistemic overload as epistemic injustice. The dissertation can be divided into three parts: 1) motivational, 2) theoretical, 3) applications and implications. First, Chapters 1 and 2 motivate the study of epistemic injustice and epistemic overload. Chapter 1 identifies a gap in the literature on epistemic injustice concerning excesses in epistemic goods as sources of epistemic injustice while canvassing the major themes and debates of the field. Chapter 2 levels an objection to ‘proper’ epistemology, thereby providing an indirect defense of the study of epistemic injustice. Second, theoretical development occurs in are Chapters 3, 4, 5, 6. Chapter 3 initiates the argument for epistemic overload, while Chapter 4 extends the case for epistemic overload, identifying several epistemic injustices arising from excesses of understanding, credibility, and truth. Chapter 5 explains the oversights of prior theorists by developing a more descriptively adequate account of social epistemics that explains the many sites of epistemic injustice. Chapter 6 develops a two-stage contractualist theory of epistemic in/justice to explain the bad-making features of epistemic injustices and generates the duty of epistemic charity. The third part of the dissertation applies the findings of earlier chapters to contemporary practical and theoretical problems. Chapter 7 employs the contractualist reasoning of Chapter 6 to address and ameliorate problems from excesses in the uptake and circulation of hermeneutical resources and true-beliefs. Chapter 8 considers the mutual dependence relations between political phenomena and epistemic in/justice, showing that accounts of political justice depend upon or presuppose epistemic justice. Finally, Chapter 9 applies epistemic overload to the use of big data technologies in the context of predive policing algorithms. An abductive argument concludes that the introduction of the “Strategic Subjects List” as part of a Chicago policing initiative in 2013 introduced understandings which likely contributed to gun-violence in Chicago and which constitutes an epistemic overload. In sum, the dissertation shows the theoretical and practical significance of epistemic overload as epistemic injustice. Ph. D.

Various kinds of media and metadata, such as pictures, videos, and geo-location, can be attached to emergency reports to the police using dedicated platforms, social networking sites, or general communication apps such as WhatsApp. Although potentially a very useful source of information for law enforcement agencies, this also raises considerable concerns regarding surveillance and privacy in public spaces: It exhorts citizens to establish a supervisory gaze over anyone, at any time, and anywhere. This chapter analyses these concerns using theories from surveillance studies. It considers the (surprisingly high) applicability of panoptical theories by Foucault and others to the effects of increased visibility of citizens in public spaces. This analysis importantly reveals how discriminatory tendencies might be introduced and exacerbated. Attention is then paid to Deleuze’s ‘societies of control’ and related notions such as database surveillance, surveillance assemblages, and predictive policing. This analysis shows that the enrichment of emergency reports with media and metadata from smartphones can pressurize people into conformity, erode the presumption of innocence, and diminish societal trust. Furthermore, this process will disproportionality affect already disadvantaged groups and individuals. Policy makers are advised to implement enriched emergency reports carefully.

The spatial and temporal thresholds (K and T) are two key parameters that control the performance of the prospective space-time scan statistical (PSTSS) hotspot method. This study proposes an objective function approach, in which the optimal values of K and T that maximize the mean hit rate (a measure of predictive accuracy), are determined. The proposed approach involves sweeping through a range of values defined for each parameter and monitors their impacts on the mean hit rate. A case study of the crime data sets of the South Chicago area is presented in which 100 one-day consecutive predictions are carried out. Two aspects of the derived results are significant. First, is that there is a trade-off between the predictive accuracy obtainable from the use of PSTSS and the level of hotspot coverage. Second, is that K is found to have more influence on the accuracy than T. As K increases in size, the accuracy level decreases, whereas there is no notable impact of T on the accuracy, particularly when T is 30 days or more. This study also demonstrated the distinctiveness of PSTSS as a hotspot method as compared to other conventional hotspot methods. Lastly, it is argued that the approach demonstrated in this study is not only applicable to crime hotspot prediction, but could also be used in many other domains where the PSTSS technique is used.

VeriPol is an effective text-based lie detection model for police reports.Our model includes feature selection by L1 penalization and heuristic rules.Computational experiments on a real dataset show a validation accuracy of 91.A pilot study shows a lower bound on the empirical precision of 83%, approx.The model analysis provides linguistic insights of how people lie to the police. Filing a false police report is a crime that has dire consequences on both the individual and the system. In fact, it may be charged as a misdemeanor or a felony. For the society, a false report results in the loss of police resources and contamination of police databases used to carry out investigations and assessing the risk of crime in a territory. In this research, we present VeriPol, a model for the detection of false robbery reports based solely on their text. This tool, developed in collaboration with the Spanish National Police, combines Natural Language Processing and Machine Learning methods in a decision support system that provides police officers the probability that a given report is false. VeriPol has been tested on more than 1000 reports from 2015 provided by the Spanish National Police. Empirical results show that it is extremely effective in discriminating between false and true reports with a success rate of more than 91%, improving by more than 15% the accuracy of expert police officers on the same dataset. The underlying classification model can be analysed to extract patterns and insights showing how people lie to the police (as well as how to get away with false reporting). In general, the more details provided in the report, the more likely it is to be honest. Finally, a pilot study carried out in June 2017 has demonstrated the usefulness of VeriPol on the field.

Social sensing plays an important role in crime analytics and predictive policing. When humans play the role of sensor, several issues around privacy and trust emerge that must be carefully handled. We provide a framework for deploying predictive crime models based upon crowd-sourced information (crime reports, tips, Nextdoor posts, etc.) while protecting individual privacy and striving for a high level of algorithmic transparency. For this purpose we introduce a novel online Hawkes process estimation algorithm requiring no event history coupled with an online k-means type algorithm based upon the word movers distance. We illustrate the methodology using synthetic data, crime report data from Los Angeles, and public safety posts from Nextdoor in Indianapolis. In particular, we show that privacy and transparency can be maintained without sacrificing accuracy in space-time models of criminal incidents. Furthermore, our methodology provides a framework for sharing of information between private companies collecting crime tips or public safety information, law enforcement agencies, and the general public.

In this paper, I use The New York Times’ debate titled, “Can predictive policing be ethical and effective?” to examine what are seen as the key operations of predictive policing and what impacts they might have in our current culture and society. The debate is substantially focused on the ethics and effectiveness of the computational aspects of predictive policing including the use of data and algorithms to predict individual behaviour or to identify hot spots where crimes might happen. The debate illustrates both the benefits and the problems of using these techniques, and makes a strong stance in favor of human control and governance over predictive policing. Cultural techniques in the paper is used as a framework to discuss human agency and further elaborate how predictive policing is based on operations which have ethical, epistemological, and social consequences.