Abstract. Logic Programs with Annotated Disjunctions (LPADs) provide a simple and elegant framework for representing probabilistic knowledge in logic programming. In this paper I consider the problem of learning ground LPADs starting from a set of interpretations annotated with their probability. I present the system ALLPAD for solving this problem. ALLPAD modifies the previous system LLPAD in order to tackle real world learning problems more effectively. This is achieved by looking for an approximate solution rather than a perfect one. ALLPAD has been tested on the problem of classifying proteins according to their tertiary structures and the results compare favorably with most other approaches. 1

Abstract. Several probabilistic logical modelling languages are compared on the task of describing or learning the inheritance mechanism discovered by Mendel. This small exercise reveals differences with respect to how easily certain kinds of domain knowledge (which may improve the learnability of the model) can be expressed by them. 1

Abstract. A new probabilistic-logic formalism, called CHRiSM, is introduced. CHRiSM is based on a combination of CHR and PRISM. It can be used for high-level rapid prototyping of complex statistical models by means of chance rules. The underlying PRISM system can then be used for several probabilistic inference tasks, including parameter learning. We describe a source-to-source transformation from CHRiSM rules to PRISM, via CHR(PRISM). Finally we discuss the relation between CHRiSM and probabilistic logic programming, in particular, CP-logic. 1

Recursive loops in a logic program present a challenging problem to the PLP (Probabilistic Logic Programming) framework. On the one hand, they loop forever so that the PLP backward-chaining inferences would never stop. On the other hand, they may generate cyclic influences, which are disallowed in Bayesian networks. Therefore, in existing PLP approaches logic programs with recursive loops are considered to be problematic and thus are excluded. In this paper, we propose a novel solution to this problem by making use of recursive loops to build a stationary dynamic Bayesian network. We introduce a new PLP formalism, called a Bayesian knowledge base. It allows recursive loops and contains logic clauses of the form A ← A1,..., Al, true, Context, T ypes, which naturally formulates the knowledge that the Ais have direct influences on A in the context Context under the type constraints Types. We use the well-founded model of a logic program to define the direct influence relation and apply SLG-resolution to compute the space of random variables together with their parental connections. This establishes a clear declarative semantics for a Bayesian knowledge base. We view a logic program with recursive loops as a special temporal model, where backward-chaining cycles of the form A ←...A ←... are interpreted as feedbacks. This extends existing PLP approaches, which mainly aim at (non-temporal) relational models.

Abstract. PRISM is a probabilistic extension of Prolog. It is a high level language for probabilistic modeling capable of learning statistical parameters from observed data. After reviewing it from various viewpoints, we examine some technical details related to logic programming, including semantics, search and program synthesis. 1

Abstract: Most automated reasoning tasks with practical applications can be automatically reformulated into a constraint solving task. A constraint programming platform can thus act as a unique, underlying engine to be reused for multiple automated reasoning tasks in intelligent agents and systems. We identify six key requirements for such platform: expressive task modeling language, rapid solving method customization and combination, adaptive solving method, user-friendly solution explanation, efficient execution, and seamless integration within larger systems and practical applications. We then propose a novel, model-driven, component and rule-based architecture for such a platform that better satisfies as a whole this set of requirements than those of currently available platforms.

There has been a long standing division in Artificial Intelligence between logical and probabilistic reasoning approaches. While probabilistic models can deal well with inherent uncertainty in many real-world domains, they operate on a mostly propositional level. Logic systems, on the other hand, can deal with much richer representations, especially first-order ones, but treat uncertainty only in limited ways. Therefore, an integration of these types of inference is highly desirable, and many approaches have been proposed, especially from the 1990s on. These solutions come from many different subfields and vary greatly in language, features and (when available at all) inference algorithms. Therefore their relation to each other is not always clear, as well as their semantics. In this survey, we present the main aspects of the solutions proposed and group them according to language, semantics and inference algorithm. In doing so, we draw relations between them and discuss particularly important choices and tradeoffs.

We examine the relation between constructive processes and the concept of causality. We observe that causality has an inherent dynamic aspect, i.e., that, in essence, causal information concerns the evolution of a domain over time. Motivated by this observation, we construct a new representation language for causal knowledge, whose semantics is defined explicitly in terms of constructive processes. This is done in a probabilistic context, where the basic steps that make up the process are allowed to have non-deterministic effects. We then show that a theory in this language defines a unique probability distribution over the possible outcomes of such a process. This result offers an appealing explanation for the usefulness of causal information and links our explicitly dynamic approach to more static causal probabilistic modeling languages, such as Bayesian networks. We also show that this language, which we have constructed to be a natural formalization of a certain kind of causal statements, is closely related to logic programming. This result demonstrates that, under an appropriate formal semantics, a rule of a normal, a disjunctive or a certain kind of probabilistic logic program can be interpreted as a description of a causal event.

We introduce the concept of a decision-driven model, a probabilistic model that reasons directly over the uncertain information of interest to a decision maker. We motivate the use of these models from the perspective of personalized medicine. Decision-driven models have a number of benefits that are of particular value in this domain, such as being easily interpretable and naturally quantifying confidences in both evidence and predictions. We show how decision-driven models can easily be constructed using probabilistic soft logic, a recently introduced framework for statistical relational learning and inference which allows the specification of medical domain knowledge in concise first-order-logic rules with assigned confidence values. 1

Abstract. Digital Libraries contain collections of multimedia objects providing services for the management, sharing and retrieval. Involved objects have two levels of complexity: the former refers to the inner object complexity while the latter takes into account the implicit/explicit relationships among objects. Traditional machine learning classifiers do not consider the relationships among objects assuming them independent and identically distributed. Recently, link-based classification methods have been proposed, that try to classify objects exploiting their relationships (links). In this paper, we deal with objects corresponding to digital images, even if the proposed approach can be naturally applied to different kind of multimedia objects. Relationships can be expressed among the features of the same image or among features belonging to different images. The aim of this work is to verify whether a link-based classifier based on a Statistical Relational Learning (SRL) language can improve the accuracy of a classical k-nearest neighbour approach. Experiments will show that the modelling of the relationships in a real-word dataset using a SRL model reduces the classification error. 1