Events have been categorized, modeled and recorded by researchers and practitioners for many centuries. Life and events are always being the philosopher's topic of debate. For us, event is any happening worth of remembering. This thesis makes an in-depth philological query of the nature of events and their intricate relationship to other events in the tapestry of complex social structures. We tried to understand our life events from grainy to vast in nature and size. Causation and effects are investigated and a simplified model is proposed for a user centric personal event management system which is fundamentally different from any existing system. Facts as a priori and stories as a posteriori has been separated by formal definition. Novel visualization and interaction is proposed to meet every individual's needs. The concept of lifelines has been introduced for the organizational requirements of a single person's life events that made it possible to distinguish from being the part of an event and being the witness of an event. This visualization model made it easier to manage causal relationships between events. Rich and intuitive interaction has been developed and proposed through the user-centric design process.

Mathematical models are powerful tools for simulating plausible epidemic spread scenarios and for evaluating the impact of control policies. They represent the scientific basis on which public health policy makers should take their decisions on the intervention strategies that should be performed at local, national and international scale. In this context, Individual-Based simulation Models (IBM) have become one of the much relevant approaches. The crucial point of this thesis project is to override some of the limit of the current generation of IBM. Specifically, highly detailed models of the sociodemography and mobility of the Italian and European population have been developed; a model of individuals and households demographics, which leads the network of contacts among individual to evolve over time, has been introduced; an analysis of the role of different assumptions on the ``random''contacts among the individuals of a population on the spread of epidemics has been performed. Results such as the development, for the first time in literature, of an IBM working on a continental scale and of an IBM suitable for the investigation of endemic diseases represent a crucial improvement for the community of epidemic modelers. Moreover, the achieved results in terms of evaluation of the effectiveness of (individually-targeted) public health control measures have had a practical application. In fact, they have been used by the Italian Ministry of Health for assessing the efficacy of the Italian pandemic preparedness plan and for planning the mitigation strategies for the 2009 A(H1N1) influenza pandemic.

Recent years have witnessed a growing interest in analyzing the huge amount of human behavioral data generated by new technologies such as mobile phones, social media and credit cards. These technologies leave a trail of "digital breadcrumbs" that allow us to have new quantitative insights that may reveal patterns of individual and group behaviors. Moreover, it allows us to better understand human behavior at a fine-grained resolution and for periods of time that were previously inconceivable. Researchers can now observe human behavior, ask research questions and run experiments in ways that were simply impossible in the recent past due to qualitative methods that, despite their undeniable benefits, proved to be time and resource consuming and therefore difficult to apply to large scale studies. Studying social interaction and social networks extracted from these data sources, allow us to understand not only individual behaviors and their characteristics, but also to observe the relationships between individuals, the structure, the content and their dynamics over long periods of time. Given the capacity of mobile phones to capture real observations of communications between people, we took advantage of the data collected from these devices to further explore and investigate human behavior. Specifically, in this dissertation, we (i) present the Mobile Territorial Lab (MTL) project and illustrate the advantages of using a living lab approach to collect a longitudinal set of data from a target group of parents; (ii) investigate how the personality dispositions of an individual influence how (s)he manages her/his social network; (iii) investigate whether and how the behavior of an individual as sensed through her/his mobile phone behavior is related to the future adoption and use of the leading mobile money service M-Pesa.

Smartphones play an important role in our daily lives. Once used only for communication purposes are now also used for several day-to-day activities ranging from social media and entertainment to privacy sensitive operations such as data storage, fitness tracking, mobile banking and sending/receiving business e-mails. This is achieved thanks to the several smartphone applications (apps) that are available. One of the most popular smartphone operating systems is Android. As of now, there are more than 3 million apps for Android. The Android platform facilitates reuse of apps' functionalities by allowing an app to request a task from another app installed on the same device through inter-process communication mechanism. This possibility is probably one of the reasons for the popularity of Android where an app can reuse a feature available in other apps. However, this integration also poses security risks to the privacy of the end-users if it is not implemented properly. Permission re-delegation vulnerability is a kind of privilege escalation that happens when unprivileged apps exploit this integration feature to make privileged apps perform a privileged action on their behalf. Static analysis techniques as well as run-time protections have been proposed to detect permission re-delegation vulnerabilities. However, as acknowledged by their authors, most of these approaches are affected by many false positives and, hence, fall short of precision because, they do not discriminate between intentional task requests and actual permission re-delegation vulnerabilities. In this thesis, we propose automatic techniques to classify potential permission re-delegation vulnerabilities detected by static analysis in real world Android apps as intentional task requests or actual vulnerabilities and to automatically generate test cases that show how the vulnerabilities can be exploited. This could be helpful for developers to easily analyze their apps and fix vulnerabilities before releasing their apps. The proposed approaches have been experimentally validated with thousands of real world apps and have been seen to perform better than state-of-the-art tools and techniques in terms of precision.

A core problem in data mining is to retrieve data in an easy and human friendly way. Automatically translating natural language questions into SQL queries would allow for the design of effective and useful database systems from a user viewpoint. In this thesis, we approach such problem by carrying out a mapping between natural language (NL) and SQL syntactic structures. The mapping is automatically derived by applying machine learning algorithms. In particular, we generate a dataset of pairs of NL questions and SQL queries represented by means of their syntactic trees automatically derived by their respective syntactic parsers. Then, we train a classifier for detecting correct and incorrect pairs of questions and queries using kernel methods along with Support Vector Machines. Experimental results on two different datasets show that our approach is viable to select the correct SQL query for a given natural language questions in two target domains. Given that preliminary results were encouraging we implemented an SQL query generator that creates the set of candidate SQL queries which we rerank with a SVM-ranker based on tree kernels. In particular we exploit linguistic dependencies in the natural language question and the database metadata to build a set of plausible SELECT, WHERE and FROM clauses enriched with meaningful joins. Then, we combine all the clauses to get the set of all possible SQL queries, producing candidate queries to answer the question. This approach can be recursively applied to deal with complex questions, requiring nested sub-queries. We sort the candidates in terms of scores of correctness using a weighting scheme applied to the query generation rules. Then, we use a SVM ranker trained with structural kernels to reorder the list of question and query pairs, where both members are again represented as syntactic trees. The f-measure of our model on standard benchmarks is in line with the best models (85% on the first question), which use external and expensive hand-crafted resources such as the semantic interpretation. Moreover, we can provide a set of candidate answers with a Recall of the answer of about 92% and 96% on the first 2 and 5 candidates, respectively.}

Software vulnerabilities are a well-known problem in current software projects. The situation becomes even more complicated, due to the ever-increasing complexity of the interconnections between both commercial and free open-source software (FOSS) projects. In this dissertation, we are aiming to facilitate the security assessment process in an industrial context. We start from the level of the own code of an individual software project, for which we propose a differential benchmarking approach for automatic assessment of static analysis security testing tools. We have demonstrated this approach, using 70 revisions of four major versions of Apache Tomcat with 62 distinct vulnerability fixes as a ground-truth set to test 7 tools. Since modern software projects often import functionality via software dependencies, that can also introduce vulnerabilities into the dependent project, we propose a methodology for counting actually vulnerable dependencies. We have evaluated the methodology on the set of 200 most used industry-relevant FOSS libraries, that resulted in 10905 distinct library instances when considering all the library versions. Finally, we have investigated the situation on the level of the FOSS ecosystem. Here we have studied decision-making strategies of developers for selecting and updating dependencies, as well as the influence of security concerns on the developers' decisions from quantitative and qualitative perspectives. For the qualitative study we have run 15 semi-structured interviews with software developers from 15 companies located in 7 countries.

The world is extremely diverse and its diversity is obvious in the cultural differences and the large number of spoken languages being used all over the world. In this sense, we need to collect and organize a huge amount of knowledge obtained from multiple resources differing from one another in many aspects. A possible approach for doing that is to think of designing effective tools for construction and maintenance of linguistic resources and localized domain ontologies based on well-defined knowledge representation methodologies capable of dealing with diversity and the continuous evolvement of human knowledge. In this thesis, we present a collaborative platform which allows for knowledge organization in a language-independent manner and provides the appropriate mapping from a language independent concept to one specific lexicalization per language. This representation ensures a smooth multilingual enrichment process for linguistic resources and a robust construction of ontologies using language-independent concepts. The collaborative platform is designed following a workflow-based development methodology that models linguistic resources as a set of collaborative objects and assigns a customizable workflow to build and maintain each collaborative object in a community driven manner, with extensive support of modern web 2.0 social and collaborative features.

Textual Entailment (TE) aims at capturing major semantic inference needs across applications in Natural Language Processing. Since 2005, in the TE recognition (RTE) task, systems are asked to automatically judge whether the meaning of a portion of text, the Text, entails the meaning of another text, the Hypothesis. Although several approaches have been experimented, and improvements in TE technologies have been shown in RTE evaluation campaigns, a renewed interest is rising in the research community towards a deeper and better understanding of the core phenomena involved in textual inference. In line with this direction, we are convinced that crucial progress may derive from a focus on decomposing the complexity of the TE task into basic phenomena and on their combination. Analysing TE in the light of the notions provided in logic to deï¬ ne an argument, and to evaluate its validity, the aim of our work is to understand how the common intuition of decomposing TE would allow a better comprehension of the problem from both a linguistic and a computational viewpoint. We propose a framework for component-based TE, where each component is in itself a complete TE system, able to address a TE task on a speciï¬ c phenomenon in isolation. Five dimensions of the problem are investigated: i) the deï¬ nition of a component-based TE architecture; ii) the implementation of TE-components able to address speciï¬ c inference types; iii) the linguistic analysis of the phenomena relevant to component-based TE; iv) the automatic acquisition of knowledge to support component-based entailment judgements; v) the development of evaluation methodologies to assess component-based TE systems capabilities to address single phenomena in a pair.

Over the last decade, Online Social Networks (OSNs) have attracted hundreds of millions of users worldwide, establishing themselves as one of most successful communication tools to date. Yet, the business model adopted by current centralized approaches makes them inherently prone to privacy issues and hostile to openness, as service providers rely on the commercial exploitation of their userbases' private data as their means of survival. We believe, as others do, that decentralization could represent a solution to this fundamental problem. In this work, we propose a novel P2P approach to decentralized OSNs in which peers are organized as a social overlay (SO): an overlay network that effectively mirrors an underlying social network by constraining communication to pairs of peers whose owners are friends. SOs are special in two ways. First, by embodying friendship in their links, SOs can help us either solve or mitigate fundamental trust-related issues that arise in P2P systems. Second, SOs exhibit an inherent compatibility towards OSNs, a result of the former being shaped after human communication, and the latter being human communication tools. These give raise to an inherent potential for synergy, which we propose to reap by means of a simple approach that provides a key functionality of modern OSNs: profile-based communication. In our approach, nodes cache the profile pages of their friends locally, and updates get proactively disseminated only by trusted nodes, over a user's ego network: the subgraphs of social networks composed by a user, her friends, and the connections among them. The contributions of this thesis then emerge as we tackle this seemingly simple problem of update dissemination over ego networks and, along the way, uncover issues that lead us to progressively deeper problems and understanding, and, ultimately, to effective solutions. In the first part of this thesis, we explore the use of push gossip protocols as the means to achieve efficient dissemination of updates over ego networks. We show that mainstream gossip protocols cannot be applied in this context, due to the largely non-uniform structure of ego networks. By taking these structural properties into account, we develop a novel gossip protocol that is able to adapt to, and leverage this non-uniformity, providing efficient and timely dissemination of updates. The study of these dissemination protocols under peer churn leads us to uncover the second problem we tackle in this thesis -- namely, the network-induced communication delays that emerge from the interaction of the social graph with the underlying peer dynamics. By means of a small-scale simulation study, we find that not only these delays can be rather extreme, but that they matter more than the underlying dissemination protocol on the long run. While this realization is in itself a contribution, we also find that evaluating the problem in more depth, as well as identifying opportunities for improvement, cannot be done by simulations alone. This is due to three factors: i) the size of the target networks under study, ii) the large parameter space inherent to availability modelling, and iii) the large number of repetitions required for obtaining good quality estimators. Put together, these translate into prohibitive costs. We therefore propose a novel hybrid analytical/simulation framework that enables the estimation of dissemination delays at a practical cost. In the third part of this thesis, we show how to further develop this framework by deriving analytical, closed-form expressions that describe delays as a function of a graph and availability parameters, when the underlying availability model is based on a certain class of simpler distributions. Finally, by putting together the lessons we learnt along the way -- our dissemination protocol and the knowledge we acquired about the workings of communication delays -- we devise the final contribution of this thesis: a hybrid, cloud-assisted P2P architecture that enables efficient dissemination in social overlays under churn. This solution, as we show, provides performance that rivals that of centralized solutions, while incurring modest economical costs.

The increasing number of repeated malware penetrations into official mobile app markets poses a high security threat to the confidentiality and privacy of end users' personal and sensitive information. Protecting end user devices from falling victims to adversarial apps presents a technical and research challenge for security researchers/engineers in academia and industry. Despite the security practices and analysis checks deployed at app markets, malware sneak through the defenses and infect user devices. The evolution of malware has seen it become sophisticated and dynamically changing software usually disguised as legitimate apps. Use of highly advanced evasive techniques, such as encrypted code, obfuscation and dynamic code updates, etc., are common practices found in novel malware. With evasive usage of dynamic code updates, a malware pretending as benign app bypasses analysis checks and reveals its malicious functionality only when installed on a user's device. This dissertation provides a thorough study on the use and the usage manner of dynamic code updates in Android apps. Moreover, we propose a hybrid analysis approach, StaDART, that interleaves static and dynamic analysis to cover the inherent shortcomings of static analysis techniques to analyze apps in the presence of dynamic code updates. Our evaluation results on real world apps demonstrate the effectiveness of StaDART. However, typically dynamic analysis, and hybrid analysis too for that matter, brings the problem of stimulating the app's behavior which is a non-trivial challenge for automated analysis tools. To this end, we propose a backward slicing based targeted inter component code paths execution technique, TeICC. TeICC leverages a backward slicing mechanism to extract code paths starting from a target point in the app. It makes use of a system dependency graph to extract code paths that involve inter component communication. The extracted code paths are then instrumented and executed inside the app context to capture sensitive dynamic behavior, resolve dynamic code updates and obfuscation. Our evaluation of TeICC shows that it can be effectively used for targeted execution of inter component code paths in obfuscated Android apps. Also, still not ruling out the possibility of adversaries reaching the user devices, we propose an on-phone API hooking based app introspection mechanism, AppIntrospector, that can be used to analyze, detect and prevent runtime exploitation of app vulnerabilities that involve dynamic code updates.