Differences in social responses of individuals can often be related to differences in functioning of neurological mechanisms. This paper presents a cognitive agent model capable of showing different types of social response patterns based on such mechanisms, adopted from theories on mirror neuron systems, emotion regulation, empathy, and autism spectrum disorders. The presented agent model provides a basis for human-like social response patterns of virtual agents in the context of simulation-based training (e.g., for training of therapists), gaming, or for agent-based generation of virtual stories.

Recent years have seen a lot of work towards extending the established AGM belief revision theory with respect to iterating revision, preserving conditional beliefs, and handling sets of propositions as new information. In particular, novel postulates like independence and evidence retainment have been brought forth as new standards for revising epistemic states by (sets of) propositional information. In this paper, we propose a constructive approach for revising epistemic states by sets of (propositional and conditional) beliefs that combines ideas from nonmonotonic reasoning with conditional belief revision. We also propose a novel principle called enforcement that covers both independence and evidence retainment, and we show our revision operator to comply with major postulates from the literature. Moreover, we point out the relevance of our approach for default reasoning.

The computational complexity of relevant corerelatedquestions for coalitional games is addressed from the coalition structure viewpoint, i.e., withoutassuming that the grand-coalition necessarily forms. In the analysis, games are assumed to be in "compact" form, i.e., their worth functions are implicitly given as polynomial-time computable functions over succinct game encodings provided as input. Within this setting, a complete picture of the complexity issues arising with the core, as well as with the related stability concepts of least core and cost of stability, is depicted. In particular, the special cases of superadditive games and of games whose sets of feasible coalitions are restricted over tree-like interaction graphs are also studied.

The use of ontologies for accessing data is one of the most exciting new applications of description logic in databases and other information systems. A realistic way of realising sufficiently scalable ontology- based data access in practice is by reduction to querying relational databases. In this paper, we describe the ‘combined approach,’ which incorporates the information given by the ontology into the data and employs query rewriting to eliminate spurious answers. We illustrate this approach for ontologies given in the DL-Lite family of description logics and briefly discuss the results obtained for the EL family.

A major challenge of query language design is the combination of expressivity with effective static analyses such as query containment. In the setting of XML, documents are seen as finite trees, whose structure may additionally be constrained by type constraints such as those described by an XML schema. We consider the problem of query containment in the presence of type constraints for a class of regular path queries extended with counting and interleaving operators. The counting operator restricts the number of occurrences of children nodes satisfying a given logical property. The interleaving operator provides a succinct notation for describing the absence of order between nodes satisfying a logical property. We provide a logic-based framework supporting these operators, which can be used to solve common query reasoning problems such as satisfiability and containment of queries in exponential time.

This paper introduces the Seqbin meta-constraint with a polytime algorithm achieving generalized arc-consistency. Seqbin can be used for encoding counting constraints such as Change, Smooth, or InncreasingNValue. For all of them the time and space complexity is linear in the sum of domain sizes, which improves or equals the best known results of the literature.

In this paper, we propose a novel approach for learning regression rules by transforming the regression problem into a classification problem. Unlike previous approaches to regression by classification, in our approach the discretization of the class variable is tightly integrated into the rule learning algorithm. The key idea is to dynamically define a region around the target value predicted by the rule, and considering all examples within that region as positive and all examples outside that region as negative. In this way, conventional rule learning heuristics may be used for inducing regression rules. Our results show that our heuristic algorithm outperforms approaches that use a static discretization of the target variable, and performs en par with other comparable rule-based approaches, albeit without reaching the performance of statistical approaches.

This paper presents an algorithm for finding approximately optimal policies in very large Markov decision processes by constructing a hierarchical model and then solving it approximately. It exploits factored representations to achieve compactness and efficiency and to discover connectivity properties of the domain. We provide a bound on the quality of the solutions and give asymptotic analysis of the runtimes; in addition we demonstrate performance on a collection of very large domains. Results show that the quality of resulting policies is very good and the total running times, for both creating and solving the hierarchy, are significantly less than for an optimal factored MDP solver.

We consider an extension of the propositional modal logic S4 which allows <> to act not only on isolated formulas, but also on sets of formulas. The interpretation of <>A is then given by the tangled closure of the valuations of formulas in A, which over finite transitive, reflexive models indicates the existence of a cluster satisfying A. This extension has been shown to be more expressive than the basic modal language: for example, it is equivalent to the bisimulation-invariant fragment of FOL over finite S4 models, whereas the basic modal language is weaker. However, previous analyses of this logic have been entirely semantic, and no proof system was available. In this paper we present a sound proof system for the polyadic S4 and prove that it is complete. The axiomatization is fairly standard, adding only the fixpoint axioms of the tangled closure to the usual S4 axioms. The proof proceeds by explicitly constructing a finite model from a consistent set of formulas.

Most of the text mining tasks, such as clustering, is dominated by statistical approaches that treat text as a bag of words. Semantics in the text is largely ignored in the mining process, and the mining results are often not easily interpretable. One particular challenge faced by such approaches is short text understanding, as short text lacks enough content from which a statistical conclusion can be drawn. For example, traditional topic analysis methods consider topic segments with tens of hundreds of words. Latent topic modeling, such as latent Dirichlet allocation, also requires sufficient words to infer document topic distribution. We enhance machine learning algorithms by first giving the machine a probabilistic knowledgebase that contains as big, rich, and consistent concepts (of worldly facts) as those in our mental world. Then a Bayesian inference mechanism is developed to conceptualize words and short text. We conducted comprehensive tests of our method on conceptualizing set of text terms, as well as clustering Twitter messages (tweets), which are typically approximately ten words long. Compared to latent semantic topic modeling and other four kinds of methods that using WordNet, Freebase and Wikipedia (category links and explicit semantic analysis), we show significant improvements in terms of tweets clustering accuracy.