Résultat de la recherche
Change points detection in crime-related time series: An on-line fuzzy approach based on a shape space representation
The extension of traditional data mining methods to time series has been effectively applied to a wide range of domains such as finance, econometrics, biology, security, and medicine. Many existing mining methods deal with the task of change points detection, but very few provide a flexible approach. Querying specific change points with linguistic variables is particularly useful in crime analysis, where intuitive, understandable, and appropriate detection of changes can significantly improve the allocation of resources for timely and concise operations. In this paper, we propose an on-line method for detecting and querying change points in crime-related time series with the use of a meaningful representation and a fuzzy inference system. Change points detection is based on a shape space representation, and linguistic terms describing geometric properties of the change points are used to express queries, offering the advantage of intuitiveness and flexibility. An empirical evaluation is first conducted on a crime data set to confirm the validity of the proposed method and then on a financial data set to test its general applicability. A comparison to a similar change-point detection algorithm and a sensitivity analysis are also conducted. Results show that the method is able to accurately detect change points at very low computational costs. More broadly, the detection of specific change points within time series of virtually any domain is made more intuitive and more understandable, even for experts not related to data mining.
Des données aux connaissances, un chemin difficile : réflexion sur la place du data mining en analyse criminelle
Le "data mining", ou "fouille de données", est un ensemble de méthodes et de techniques attractif qui a connu une popularité fulgurante ces dernières années, spécialement dans le domaine du marketing. Le développement récent de l’analyse ou du renseignement criminel soulève des problématiques auxquelles il est tentant d’appliquer ces méthodes et techniques. Le potentiel et la place du data mining dans le contexte de l’analyse criminelle doivent être mieux définis afin de piloter son application. Cette réflexion est menée dans le cadre du renseignement produit par des systèmes de détection et de suivi systématique de la criminalité répétitive, appelés processus de veille opérationnelle. Leur fonctionnement nécessite l’existence de patterns inscrits dans les données, et justifiés par les approches situationnelles en criminologie. Muni de ce bagage théorique, l’enjeu principal revient à explorer les possibilités de détecter ces patterns au travers des méthodes et techniques de data mining. Afin de répondre à cet objectif, une recherche est actuellement menée en Suisse à travers une approche interdisciplinaire combinant des connaissances forensiques, criminologiques, et computationnelles
The CriLiM Methodology: Crime Linkage with a Fuzzy MCDM Approach
Grouping events having similarities has always been interesting for analysts. Actually, when a label is put on top of a set of events to denote they share common properties, the automation and the capability to conduct reasoning with this set drastically increase. This is particularly true when considering criminal events for crime analysts; conjunction, interpretation and explanation can be key success factors to apprehend criminals. In this paper, we present the CriLiM methodology for investigating both serious and high-volume crime. Our artifact consists in implementing a tailored computerized crime linkage system, based on a fuzzy MCDM approach in order to combine spatio-temporal, behavioral, and forensic information. As a proof of concept, series in burglaries are examined from real data and compared to expert results.
An Intelligent Process-driven Knowledge Extraction Framework for Crime Analysis
In this research, we attempt to study the contribution of data mining techniques in crime analysis and intelligence. It is an interdisciplinary project, combining forensic, criminological and computational methods. We search to develop a frame in which data mining techniques, driven by crime analysis and forensic processes, take an active part to data interpretation and information analysis (in order to extract knowledge). Realized in collaboration with the cantonal police forces of Vaud in Switzerland, the first phase of this project consists to focus on residential burglary data. The sample is provided by the Concept Intercantonal de Coordination Opérationnelle et Préventive (CICOP) database, which is the regional center for crime analysis in French-speaking part of Switzerland. The CICOP analysts use phenomenon codes to define a particular crime situation. These CICOP codes are directly related to the situational approaches in criminology. Concretely, we have three main purposes: residential burglary classification, new phenomena discovery, and series or trends detection. That brings, in first hand, to formalize processes identified in crime analysis with the help of a standard notation called Business Process Modeling Notation (BPMN). Then, different data mining techniques are tested on data, and assessed by confronting them with phenomena identified by police forces analysts. Finally, we make a criminological analysis on the results to check the consistency with main situational theories in criminology. Accuracy and results relevance exam is an important step, because the different data mining algorithms can generate trivial and unexplainable rules. We note then the need of a human interpretation, and in this case, of a criminological interpretation. The first results are hopeful and classification algorithms are effective to classify residential burglaries like CICOP analysts did it.
Crime Linkage: a Fuzzy MCDM Approach
Grouping crimes having similarities has always been interesting for analysts. Actually, when a set of crimes share common properties, the capability to conduct reasoning and the automation with this set drastically increase. Conjunction, interpretation and explanation based on similarities can be key success factors to apprehend criminals. In this paper, we present a computerized method for high-volume crime linkage, based on a fuzzy MCDM approach in order to combine situational, behavioral, and forensic information. Experiments are conducted with series in burglaries from real data and compared to expert results.