Stefano Bragaglia

An hybrid architecture integrating forward rules with fuzzy ontological reasoning

In recent years there has been a growing interest in the combination of rules and ontologies Notably, many works have focused on the theoretical aspects of such integration, sometimes leading to concrete solutions However, solutions proposed so far typically reason upon crisp concepts, while concrete domains require also fuzzy expressiveness. In this work we combine mature technologies, namely the Drools business rule management system, the Pellet OWL Reasoner and the FuzzyDL system, to provide a unified framework for supporting fuzzy reasoning After extending the Drools framework (language and engine) to support uncertainty reasoning upon rules, we have integrated it with custom operators that (i) exploit Pellet to perform ontological reasoning, and (ii) exploit FuzzyDL to support fuzzy ontological reasoning. As a case study, we consider a decision-support system for the tourism domain, where ontologies are used to formally describe package tours, and rules are exploited to evaluate the consistency of such packages.

Reactive event calculus for monitoring global computing applications

In 1986 Kowalski and Sergot proposed a logic-based formalism named Event Calculus (EC), for specifying in a declarative manner how the happening of events affects some representation (the state ) of the world. Since its introduction, EC has been recognized for being an excellent framework to reason about time and events. Recently, with the advent of complex software systems decomposed into sets of autonomous, heterogeneous distributed entities, EC has drawn attention as a viable solution for monitoring them, where monitoring means to represent their state and how events dynamically affect such state. In this work we present the fundamentals of a reactive and logic-based version of EC, named REC, for monitoring declarative properties, while maintaining a solid formal background. We present some results about its formal as well as practical aspects, and discuss how REC has been applied to a variety of application domains, namely BPM, SOC, CGs and MAS. We also highlight some key issues required by the monitoring task, and finally discuss how REC overcomes such issues.

Approximate inference for logic programs with annotated disjunctions

Logic Programs with Annotated Disjunctions (LPADs) are a promising language for Probabilistic Inductive Logic Programming. In order to develop efficient learning systems for LPADs, it is fundamental to have high-performing inference algorithms. The existing approaches take too long or fail for large problems. In this paper we adapt to LPAD the approaches for approximate inference that have been developed for ProbLog, namely k-best and Monte Carlo. k-Best finds a lower bound of the probability of a query by identifying the k most probable explanations while Monte Carlo estimates the probability by smartly sampling the space of programs. The two techniques have been implemented in the cplint suite and have been tested on real and artificial datasets representing graphs. The results show that both algorithms are able to solve larger problems often in less time than the exact algorithm.

A rule-based calculus and processing of complex events

Rules are definitely among the main kinds of knowledge representation in Artificial Intelligence. In recent years, there has been much discussion about production rules and logic programming to understand whether the two paradigms could be joined or, alternatively, which was the better. Conversely, the idea to program a production system with logic without actually relying on logic programming was proposed. In this paper we present a software component that implements a typical logic formalism, the Event Calculus within a production rules system. This component allows to perform deductive reasoning tasks (temporal projection or prediction, such as monitoring) and thanks to some technical choices, it proves to be quite efficient. In addition, thanks to its strong modular nature, it can adapt to the domains requirements and complement other forms of reasoning at the same time. We also present some preliminary results on tests that we have conducted to show that our system based on a Java rules engine is almost as efficient as an equivalent logic program running on the fastest C++ Prolog interpreter. Furthermore we show how our framework can be used to effectively observe the evolving state of our use case --- a Service Oriented Architecture server --- in a way that qualifies as Complex Event Processing.

Fuzzy conformance checking of observed behaviour with expectations

In some different research fields a research issue has been to establish if the external, observed behaviour of an entity is conformant to some rules/specifications/ expectations. Research areas like Multi Agent Systems, Business Process, and Legal/Normative systems, have proposed different characterizations of the same problem, named as the conformance problem. Most of the available systems, however, provide only simple yes/no answers to the conformance issue. In this paper we introduce the idea of a gradual conformance, expressed in fuzzy terms. To this end, we present a system based on a fuzzy extension of Drools, and exploit it to perform conformance tests. In particular, we consider two aspects: the first related to fuzzy ontological aspects, and the second about fuzzy time-related aspects. Moreover, we discuss how to conjugate the fuzzy contributions from these aspects to get a single, fuzzy score representing a conformance degree.

A rule-based implementation of fuzzy tableau reasoning

The integration of distinct reasoning styles such as the ones exploited by description logics and rule-based systems is still an open challenge because of the differences among them. Such integration may be achieved by following two complementary approaches: loose integration vs. tight integration. Loosely integrated hybrid systems couple existing tools, so they have to handle mutual interactions and keep their models aligned. Tightly-coupled hybrid systems, instead, are based on a unified model supporting both reasoning styles. In this paper we present a basic implementation of a fuzzy tableau algorithm for description logics by means of rules. It is a step towards tight integration because it requires only one rule engine while preserving the semantics of both reasoning styles. In particular, the adoption of a fuzzy tableau in a fuzzy rule engine allowed us to extend the expressiveness of the latter while handling description logics reasoning coherently.

Event condition expectation (ECE-) rules for monitoring observable systems

The standardization and broad adoption of Service Oriented Architectures, Web Services, and Cloud Computing is raising the complexity of ICT systems. Hence, assuring correct system behavior with regard to established design and business constraints is of the utmost importance. Run-time monitoring, where the outcomes of an observed system are continuously checked against what is expected of it, is one possible approach to providing the required oversight. In this paper, we discuss this notion of rule expectations, their violation and/or fulfillment, and use these concepts to define the concept of an Event-Condition-Expectation (ECE-) rule, a variation of the traditional Event-Condition-Action rule pattern. To demonstrate these concepts, we present extensions to the syntax used by the production rule engine, Drools, and describe their use in a medical case study. The clinical decision support system being developed monitors rule evaluations and expectations, detects constraint violations and is able to take recovery/ compensation actions as appropriate.

Nonmonotonic Learning in Large Biological Networks

This paper introduces a new open-source implementation of a nonmonotonic learning method called XHAIL and shows how it can be used for abductive and inductive inference on metabolic networks that are many times larger than could be handled by the preceding prototype. We summarise several implementation improvements that increase its efficiency and we introduce an extended form of language bias that further increases its usability. We investigate the systems scalability in a case study involving real data previously collected by a Robot Scientist and show how it led to the discovery of an error in a whole-organism model of yeast metabolism.

Visual Decision Support for Policy Making: Advancing Policy Analysis with Visualization

Today’s politicians are confronted with new information technologies to tackle complex decision-making problems. In order to make sustainable decisions, a profound analysis of societal problems and possible solutions (policy options) needs to be performed. In this policy-analysis process, different stakeholders are involved. Besides internal direct advisors of the policy makers (policy analysts), external experts from different scientific disciplines can support evidence-based decision making. Despite the alleged importance of scientific advice in the policy-making process, it is observed that scientific results are often not used. In this work, a concept is described that supports the collaboration between scientists and politicians. We propose a science–policy interface that is realized by including information visualization in the policy-analysis process. Therefore, we identify synergy effects between both fields and introduce a methodology for addressing the current challenges of science–policy interfaces with visualization. Finally, we describe three exemplary case studies carried out in European research projects that instantiate the concept of this approach.

A Distributed System Using MS Kinect and Event Calculus for Adaptive Physiotherapist Rehabilitation

In many countries of the world, the life expectancy increases but the population ages so rapidly that it is expected that soon it will be difficult to ensure a good life quality to the elder people when health issues arise. In this paper, we consider this problem from the point of view of the physiotherapy rehabilitation which nowadays is perceived as costly and inconvenient for the elder patients. In order to lessen these problems, we propose a distributed architecture to allow the physiotherapists to remotely assist their patients while they comfortably do exercises from home. As in other proposals, the Human Pose Recognition is delegated to a computer equipped with MS Kinect and neural networks. Our approach, however, differs from others because it includes a logical framework based on Event Calculus augmented with Expectations which provides a higher-level description of the exercises and a mean to measure how well they were done.