A relevant activity in the requirements engineering process consists in the identification, assessment and management of potential risks, which can prevent the system-to-be from meeting stakeholder needs. However, risk analysis techniques are often time- and resource- consuming activities, which may introduce in the requirements engineering process a significant overhead. To overcome this problem, we aim at supporting risk management activity in a semi-automated way, merging the capability to exploit existing risk-related information potentially present in a given organisation, with an automated ranking of the goals with respect to the level of risk the decision-maker estimates for them. In particular, this paper proposes an approach to address the general problem of risk decision-making, which combines knowledge about risks assessment techniques and Machine Learning to enable an active intervention of human evaluators in the decision process, learning from their feedback and integrating it with the organisational knowledge. The long term objective is that of improving the capacity of an organisation to be aware and to manage risks, by introducing new techniques in the field of risk management that are able to interactively and continuously extract useful knowledge from the organisation domain and from the decision-maker expertise.

In this work, we present DECAF-a multimodal data set for decoding user physiological responses to affective multimedia content. Different from data sets such as DEAP [15] and MAHNOB-HCI [31], DECAF contains (1) brain signals acquired using the Magnetoencephalogram (MEG) sensor, which requires little physical contact with the user's scalp and consequently facilitates naturalistic affective response, and (2) explicit and implicit emotional responses of 30 participants to 40 one-minute music video segments used in [15] and 36 movie clips, thereby enabling comparisons between the EEG versus MEG modalities as well as movie versus music stimuli for affect recognition. In addition to MEG data, DECAF comprises synchronously recorded near-infra-red (NIR) facial videos, horizontal Electrooculogram (hEOG), Electrocardiogram (ECG), and trapezius-Electromyogram (tEMG) peripheral physiological responses. To demonstrate DECAF's utility, we present (i) a detailed analysis of the correlations between participants' self-assessments and their physiological responses and (ii) single-trial classification results for valence, arousal and dominance, with performance evaluation against existing data sets. DECAF also contains time-continuous emotion annotations for movie clips from seven users, which we use to demonstrate dynamic emotion prediction.

This paper presents a new multimodal database and the associated results for characterization of affect (valence, arousal and dominance) using the Magneto encephalogram (MEG) brain signals and peripheral physiological signals (horizontal EOG, ECG, trapezius EMG). We attempt single-trial classification of affect in movie and music video clips employing emotional responses extracted from eighteen participants. The main findings of this study are that: (i) the MEG signal effectively encodes affective viewer responses, (ii) clip arousal is better predicted by MEG, while peripheral physiological signals are more effective for predicting valence and (iii) prediction performance is better for movie clips as compared to music video clips.

Deciding which, among a set of requirements, are to be considered first and in which order is a strategic process in software development. This task is commonly referred to as requirements prioritization. This paper describes a requirements prioritization method called Case-Based Ranking (CBRank), which combines project's stakeholders preferences with requirements ordering approximations computed through machine learning techniques, bringing promising advantages. First, the human effort to input preference information can be reduced, while preserving the accuracy of the final ranking estimates. Second, domain knowledge encoded as partial order relations defined over the requirement attributes can be exploited, thus supporting an adaptive elicitation process. The techniques CBRank rests on and the associated prioritization process are detailed. Empirical evaluations of properties of CBRank are performed on simulated data and compared with a state-of-the-art prioritization method, providing evidence of the method ability to support the management of the tradeoff between elicitation effort and ranking accuracy and to exploit domain knowledge. A case study on a real software project complements these experimental measurements. Finally, a positioning of CBRank with respect to state-of-the-art requirements prioritization methods is proposed, together with a discussion of benefits and limits of the method.

Functional neuroimaging consists in the use of imaging technologies allowing to record the functional brain activity in real-time. Among all techniques, data produced by functional magnetic resonance is encoded as sequences of 3D images of thousands of voxels. The main investigation performed on this data, termed brain mapping, aims at producing functional maps of the brain. Brain mapping aims at the detection of the portion of voxels concerned with specific perceptual or cognitive brain activities. This challenge can be shaped as a problem of feature selection. Excessive features-to-instances ratio characterizing this data is a major issue for the computation of statistically robust maps. We propose a solution based on a Random Subspace Method that extends the reference approach (Search Light) adopted by the neuroscientific community. A comparison of the two methods is supported by the results of an empirical evaluation.

We propose a relational neural network defined as a special instance of the recurrent cascade correlation. The proposed model is designed to deal with classification tasks where classes are organized into generic graphs (e.g. taxonomies, ontologies etc.). The open challenge is to exploit the knowledge encoded in the relationships among the classes. This is particularly useful when there are many classes poorly represented by labeled examples. Exploiting the relationships we increase the bias, making the generalization more robust. The novelty of the proposed model can be seen from two different perspectives. On one hand, the temporal encoding of the standard recurrent networks is revised with a notion of non-stationary structural unfolding. On the other hand, it can be seen as a novel constructive algorithm that generates the neural network architecture exploiting the class structure. We present the results of an empirical evaluation on a hierarchical document classification task.

Recommender Systems based on Collaborative Filtering suggest to users items they might like. However due to data sparsity of the input ratings matrix, the step of finding similar users often fails. We propose to replace this step with the use of a trust metric, an algorithm able to propagate trust over the trust network and to estimate a trust weight that can be used in place of the similarity weight. An empirical evaluation on Epinions.com dataset shows that Recommender Systems that make use of trust information are the most effective in term of accuracy while preserving a good coverage. This is especially evident on users who provided few ratings.

In today’s connected world, it is possible and indeed quite common to interact with un-known people whose reliability is unknown. Trust metrics are a technique for answering questions such as “Should I trust this person?” However, most of the current research assumes that every user has a global quality score everyone agrees on and the goal of the technique is just to predict this correct value. We show on data from a real and large user community, Epinions.com, that such an assumption is not realistic because there is a significant portion of what we call controversial users, users who are trusted by many and distrusted by many: a global agreement about the trustworthiness value of these users does not exist. We argue, using computational experiments, that the existence of controversial users (a normal phenomenon in complex societies) demands local trust metrics, techniques able to predict the trustworthiness of a user in a personalized way, depending on the very personal views of the judging user as opposed to most commonly used global trust metrics, which assume a unique value of reputation for every single user. The implications of such an analysis deal with the very foundations of what we call society and culture and we conclude discussing this point by comparing the two extremes of culture that can be induced by the two different kinds of trust metrics: tyranny of the majority and echo chambers.

The general approach for automatically driving data collection using information from previously acquired data is called active learning. Traditional active learning addresses the problem of choosing the unlabeled examples for which the class labels are queried with the goal of learning a classifier. In contrast we address the problem of active feature sampling for detecting useless features. We propose a strategy to actively sample the values of new features on class-labeled examples, with the objective of feature relevance assessment. We derive an active feature sampling algorithm from an information theoretic and statistical formulation of the problem. We present experimental results on synthetic, UCI and real world datasets to demonstrate that our active sampling algorithm can provide accurate estimates of feature relevance with lower data acquisition costs than random sampling and other previously proposed sampling algorithms. 1.

The test case execution order affects the time at which the objectives of testing are met. If the objective is fault detection, an inappropriate execution order might reveal most faults late, thus delaying the bug fixing activity and eventually the delivery of the software. Prioritizing the test cases so as to optimize the achievement of the testing goal has potentially a positive impact on the testing costs, especially when the test execution time is long. Test engineers often possess relevant knowledge about the relative priority of the test cases. However, this knowledge can be hardly expressed in the form of a global ranking or scoring. In this paper, we propose a test case prioritization technique that takes advantage of user knowledge through a machine learning algorithm, case-based ranking (CBR). CBR elicits just relative priority information from the user, in the form of pairwise test case comparisons. User input is integrated with multiple prioritization indexes, in an iterative process that successively refines the test case ordering. Preliminary results on a case study indicate that CBR overcomes previous approaches and, for moderate suite size, gets very close to the optimal solution

The proliferation of text documents on the web as well as within institutions necessitates their convenient organization to enable efficient retrieval of information. Although text corpora are frequently organized into concept hierarchies or taxonomies, the classification of the documents into the hierarchy is expensive in terms human effort. We present a novel and simple hierarchical Dirichlet generative model for text corpora and derive an efficient algorithm for the estimation of model parameters and the unsupervised classification of text documents into a given hierarchy. The class conditional feature means are assumed to be inter-related due to the hierarchical Bayesian structure of the model. We show that the algorithm provides robust estimates of the classification parameters by performing smoothing or regularization. We present experimental evidence on real web data that our algorithm achieves significant gains in accuracy over simpler models.

Case-based driven approaches to requirements prioritization proved to be much more effective than first-principle methods in being tailored to a specific problem, that is they take advantage of the implicit knowledge that is available, given a problem representation. In these approaches, first-principle prioritization criteria are replaced by a pairwise preference elicitation process. Nevertheless case-based approaches, using the analytic hierarchy process (AHP) technique, become impractical when the size of the collection of requirements is greater than about twenty since the elicitation effort grows as the square of the number of requirements. We adopt a case-based framework for requirements prioritization, called case-based ranking, which exploits machine learning techniques to overcome the scalability problem. This method reduces the acquisition effort by combining human preference elicitation and automatic preference approximation. Our goal in this paper is to describe the framework in details and to present empirical evaluations which aim at showing its effectiveness in overcoming the scalability problem. The results prove that on average our approach outperforms AHP with respect to the trade-off between expert elicitation effort and the requirement prioritization accuracy.

The interoperability among distributed and autonomous systems is the ultimate challenge facing the semantic web. Heterogeneity of data representation is the main source of problems. This paper proposes an innovative solution that combines lexical approaches and language games. The benefits for distributed annotation systems on the web are twofold: firstly, it will reduce the complexity of the semantic problem by moving the focus from the full-featured ontology level to the simpler lexicon level; secondly, it will avoid the drawback of a centralized third party mediator that may become a single point of failure.The main contributions of this work are concerned with:

Managing the hierarchical organization of data is starting to play a key role in the knowledge management community due to the great amount of human resources needed to create and maintain these organized repositories of information. Machine learning community has in part addressed this problem by developing hierarchical supervised classifiers that help maintainers to categorize new resources within given hierarchies. Although such learning models succeed in exploiting relational knowledge, they are highly demanding in terms of labeled examples, because the number of categories is related to the dimension of the corresponding hierarchy. Hence, the creation of new directories or the modification of existing ones require strong investments.This paper proposes a semi-automatic process (interleaved with human suggestions) whose aim is to minimize (simplify) the work required to the administrators when creating, modifying, and maintaining directories. Within this process, bootstrapping a taxonomy with examples represents a critical factor for the effective exploitation of any supervised learning model. For this reason we propose a method for the bootstrapping process that makes a first hypothesis of categorization for a set of unlabeled documents, with respect to a given empty hierarchy of concepts. Based on a revision of Self-Organizing Maps, namely TaxSOM, the proposed model performs an unsupervised classification, exploiting the a-priori knowledge encoded in a taxonomy structure both at the terminological and topological level. The ultimate goal of TaxSOM is to create the premise for successfully training a supervised classifier.

In knowledge discovery applications, where new features are to be added, an acquisition policy can help select the features to be acquired based on their relevance and the cost of extraction. This can be posed as a feature selection problem where the feature values are not known in advance. We propose a technique to actively sample the feature values with the ultimate goal of choosing between alternative candidate features with minimum sampling cost. Our heuristic algorithm is based on extracting candidate features in a region of the instance space where the feature value is likely to alter our knowledge the most. An experimental evaluation on a standard database shows that it is possible outperform a random subsampling policy in terms of the accuracy in feature selection.

Systems which predict the items a user would like, on the basis of the preferences given by other users, are attracting growing attention as automated recommender services on the Internet. Collaborative filtering techniques are widely used to implement such recommender systems. We show how fuzzy sets and many-valued logics can be fruitfully applied in the description and design of collaborative filtering methods.

A local distance measure for the nearest neighbor classification rule is shown to achieve high compression rates and high accuracy on real data sets. In the approach proposed here, first, a set of prototypes is extracted during training and, then, a feedback learning algorithm is used to optimize the metric. Even if the prototypes are randomly selected, the proposed metric outperforms, both in compression rate and accuracy, common editing procedures like ICA, RNN, and PNN. Finally, when accuracy is the major concern, we show how compression can be traded for accuracy by exploiting voting techniques. That indicates how voting can be successfully integrated with instance-based approaches, overcoming previous negative results.