Fosca Giannotti

A classification-based methodology for planning audit strategies in fraud detection

Planning adequate audit strategies is a key success factor in a posterion’ fraud detection, e.g., in the fiscal and insurance domains, where audits are intended to detect tax evasion and fraudulent claims. A case study is presented in this paper, which illustrates how techniques based on classification can be used to support the task of planning audit strategies. The proposed approach is sensible to some conflicting issues of audit planning, e.g., the trade-off between maximizing audit benefits vs. minimizing audit costs. A methodological scenario, common to a whole class of similar applications, is then abstracted away from the case study. The limitations of available systems to support the identified overall KDD process, bring us to point out the key aspects of a logic-based database language, integrated with mining mechanisms, which is used to provide a uniform, highly expressive environment for the various steps in the construction of the considered case-study.

Returners and explorers dichotomy in human mobility

The availability of massive digital traces of human whereabouts has offered a series of novel insights on the quantitative patterns characterizing human mobility. In particular, numerous recent studies have lead to an unexpected consensus: the considerable variability in the characteristic travelled distance of individuals coexists with a high degree of predictability of their future locations. Here we shed light on this surprising coexistence by systematically investigating the impact of recurrent mobility on the characteristic distance travelled by individuals. Using both mobile phone and GPS data, we discover the existence of two distinct classes of individuals: returners and explorers. As existing models of human mobility cannot explain the existence of these two classes, we develop more realistic models able to capture the empirical findings. Finally, we show that returners and explorers play a distinct quantifiable role in spreading phenomena and that a correlation exists between their mobility patterns and social interactions.

A planetary nervous system for social mining and collective awareness

AbstractWe present a research roadmap of a Planetary Nervous System (PNS), capable of sensing and mining the digital breadcrumbs of human activities and unveiling the knowledge hidden in the big data for addressing the big questions about social complexity. We envision the PNS as a globally distributed, self-organizing, techno-social system for answering analytical questions about the status of world-wide society, based on three pillars: social sensing, social mining and the idea of trust networks and privacy-aware social mining. We discuss the ingredients of a science and a technology necessary to build the PNS upon the three mentioned pillars, beyond the limitations of their respective state-of-art. Social sensing is aimed at developing better methods for harvesting the big data from the techno-social ecosystem and make them available for mining, learning and analysis at a properly high abstraction level. Social mining is the problem of discovering patterns and models of human behaviour from the sensed data across the various social dimensions by data mining, machine learning and social network analysis. Trusted networks and privacy-aware social mining is aimed at creating a new deal around the questions of privacy and data ownership empowering individual persons with full awareness and control on own personal data, so that users may allow access and use of their data for their own good and the common good. The PNS will provide a goal-oriented knowledge discovery framework, made of technology and people, able to configure itself to the aim of answering questions about the pulse of global society. Given an analytical request, the PNS activates a process composed by a variety of interconnected tasks exploiting the social sensing and mining methods within the transparent ecosystem provided by the trusted network. The PNS we foresee is the key tool for individual and collective awareness for the knowledge society. We need such a tool for everyone to become fully aware of how powerful is the knowledge of our society we can achieve by leveraging our wisdom as a crowd, and how important is that everybody participates both as a consumer and as a producer of the social knowledge, for it to become a trustable, accessible, safe and useful public good. Graphical abstract

An analytical framework to nowcast well-being using mobile phone data

An intriguing open question is whether measurements derived from Big Data recording human activities can yield high-fidelity proxies of socio-economic development and well-being. Can we monitor and predict the socio-economic development of a territory just by observing the behavior of its inhabitants through the lens of Big Data? In this paper, we design a data-driven analytical framework that uses mobility measures and social measures extracted from mobile phone data to estimate indicators for socio-economic development and well-being. We discover that the diversity of mobility, defined in terms of entropy of the individual users’ trajectories, exhibits (i) significant correlation with two different socio-economic indicators and (ii) the highest importance in predictive models built to predict the socio-economic indicators. Our analytical framework opens an interesting perspective to study human behavior through the lens of Big Data by means of new statistical indicators that quantify and possibly “nowcast” the well-being and the socio-economic development of a territory.

Data mining for intelligent Web caching

Presents a vertical application of data warehousing and data mining technology: intelligent Web caching. We introduce several ways to construct intelligent Web caching algorithms that employ predictive models of Web requests; the general idea is to extend the LRU (least recently used) policy of Web and proxy servers by making it sensible to Web access models extracted from Web log data using data mining techniques. Two approaches have been studied, in particular one based on association rules and another based on decision trees. The experimental results of the new algorithms show substantial improvements over existing LRU-based caching techniques in terms of the hit rate, i.e. the fraction of Web documents directly retrieved in the cache. We designed and developed a prototypical system, which supports data warehousing of Web log data, extraction of data mining models and simulation of the Web caching algorithms, around an architecture that integrates the various phases in the knowledge discovery process. The system supports a systematic evaluation and benchmarking of the proposed algorithms with respect to existing caching strategies.

The purpose of motion: Learning activities from Individual Mobility Networks

The large availability of mobility data allows us to investigate complex phenomena about human movement. However this adundance of data comes with few information about the purpose of movement. In this work we address the issue of activity recognition by introducing Activity-Based Cascading (ABC) classification. Such approach departs completely from probabilistic approaches for two main reasons. First, it exploits a set of structural features extracted from the Individual Mobility Network (IMN), a model able to capture the salient aspects of individual mobility. Second, it uses a cascading classification as a way to tackle the highly skewed frequency of activity classes. We show that our approach outperforms existing state-of-the-art probabilistic methods. Since it reaches high precision, ABC classification represents a very reliable semantic amplifier for Big Data.

Mining Clinical, Immunological, and Genetic Data of Solid Organ Transplantation

Clinical databases store large amounts of information about patients and their medical conditions. Data mining techniques can extract relationships and patterns implicit in this wealth of data, and thus be helpful in understanding the progression of diseases and the efficacy of the associated therapies. In this perspective, in Pisa (Italy) we have started an important data collection and analysis project, where a very large number of epidemiological, clinical, immunological and genetic variables collected before the transplantation of a solid organ, and during the follow-up assessment of the patients, are stored in a datawarehouse for future mining. This on-going data collection involves all liver, kidney, pancreas and kidney-pancreas transplantations of the last five years of one of the largest (as to number of transplantations) centers in Europe. The project ambitious goal is to gain deeper insights in all the phenomena related to solid organ transplantation, with the aim of improving the donor-recipient matching policy used nowadays. In this chapter we report in details two different data mining activities developed within this project. The first analysis involves mining genetic data of patients affected by terminal hepatic cirrhosis with viral origin (HCV and HBV) and patients with terminal hepatic cirrhosis with non-viral origin (autoimmune): the goal is to assess the influence of the HLA antigens on the course of the disease. In particular, we have evaluated if some genetic configurations of the class I and class II HLA are significantly associated with the triggering causes of the hepatic cirrhosis. The second analysis involves clinical data of a set of patients in the follow-up of a liver transplantation. The aim of the data analysis is that of assessing the effectiveness of the extracorporeal photopheresis (ECP) as a therapy to prevent rejection in solid organ transplantation. For both analyses we describe in details, the medical context and goal, the nature and structure of the data. We also discuss which kind of data mining technique is the most suitable for our purposes, and we describe the details of the knowledge discovery process followed and extracted knowledge.

\( \mathcal{L}\mathcal{D}\mathcal{L} - \mathcal{M}_{ine} \) : Integrating Data Mining with Intelligent Query Answering

Current applications of data mining techniques highlight the need for flexible knowledge discovery systems, capable of supporting the user in specifying and re.ning mining objectives, combining multiple strategies, and de.ning the quality of the extracted knowledge. A key issue is the de.nition of Knowledge Discovery Support Environment, i.e., a query system capable of obtaining, maintaining, representing and using high level knowledge in a uni.ed framework. This comprises representation and manipulation of domain knowledge, extraction and manipulation of new knowledge and their combination.Current applications of data mining techniques highlight the need for flexible knowledge discovery systems, capable of supporting the user in specifying and refining mining objectives, combining multiple strategies, and defining the quality of the extracted knowledge. A key issue is the definition of Knowledge Discovery Support Environment, i.e., a query system capable of obtaining, maintaining, representing and using high level knowledge in a unified framework. This comprises representation and manipulation of domain knowledge, extraction and manipulation of new knowledge and their combination. In such a context, in [2,3] we envisaged an integrated architecture of data mining, further developed and experimented in [5] and resulting in the LDL−Mine environment. The basic philosophy of the environment is to integrate both inductive and deductive capabilities in a unified framework. A LDL−Mine program is composed of three main parts: source knowledge, modeled by facts; background knowledge, modeled by deductive clauses; and induced knowledge, modeled by inductive clauses. Inductive clauses provide a suitable interface to data mining algorithms: they define predicates that represent mining patterns, but can be used as deductive predicates and facts. This allows to amalgamate induction and deduction, and to model both interactive and iterative features of a data mining process. Figure 1 shows the main features of the system. LDL−Mine is built on top of the LDL++ system [8]. Indeed, the system exploits most of the functionalities of the LDL++ system, such as Application programming interface, deductive engine, and access to external databases. In addition, LDL−Mine implements an inductive engine that allows, by means of inductive clauses, interaction between mining algorithms and deductive components. In its current stage, the inductive engine implements three main data mining schemes, namely association rules mining [1], Bayesian Classification [7], and (both supervised and unsupervised) discretization of continuous attributes [6]. Each scheme corresponds to a specific inductive clause. In the following we show by examples how the notion of inductive clause is formalized within the LDL−Mine system, and some specific inductive clauses currently implemented.