Abstract:The International Conference on Knowledge Science, Engineering and Management (KSEM), which was held in Belfast, Northern Ireland in 2010, was the fourth success in the series of such conferences. The conference focuses on the three themes of knowledge science, engineering and management, covering a wider range of research topics in the KSEM-related areas. This event offered an invaluable opportunity to bring together researchers, engineers and practitioners to present original work, the latest advances on knowledge representation, pioneered knowledge engineering, knowledge related systems, as well as discuss and debate practical challenges in deploying knowledge-based systems and research opportunities in the research community. To highlight research activities drawn by this event and provide an insight into the latest developments in the related areas, we present this special issue which is dedicated to all the delegates and researchers in the community who made the conference a success. We selected more than 10 papers that were presented in the conference and asked the authors to extend these papers. After careful review, we finally select 6 papers to be included in this special issue. This collection of extended versions of papers consists of the KSEM2010's finest papers and covers the prominent topics of knowledge representation and reasoning, ontology engineering and applications, data mining and knowledge discovery. They represent state of the art of research in KSEM-related research areas.

Abstract:In this paper, we consider the problem of query answering over multimedia ontologies. Traditional reasoning systems may have problems to deal with large amounts of expressive ontological data (terminological as well as assertional data) that usually must be kept in main memory. We propose to overcome this problem with a new so-called filter and refine paradigm for ontology-based query answering. The contribution of this paper is twofold: (1) For both steps, algorithms are presented. (2) We evaluate our approach on real world multimedia ontologies from the BOEMIE project.

Abstract:Rare association rules correspond to rare, or infrequent, itemsets, as opposed to frequent ones that are targeted by conventional pattern miners. Rare rules reflect regularities of local, rather than global, scope that can nevertheless provide valuable insights to an expert, especially in areas such as genetics and medical diagnosis where some specific deviations/illnesses occur only in a small number of cases. The work presented here is motivated by the long-standing open question of efficiently mining strong rare rules, i.e., rules with high confidence and low support. We also propose an efficient solution for finding the set of minimal rare itemsets. This set serves as a basis for generating rare association rules.

Abstract:Modeling human concepts of object locations is essential for the development of the systems and machines that collaborate with ordinary people on spatial tasks. This paper applies a heterogeneous cardinal direction model, called HCDM, to model human concepts of object locations with both directional and topological information in a 2D space. Using its ability we illustrate where and how an object is located as seen from another even if they have different spatial extensions. For generality, we adopt a set of formal spatial concepts defined in an existing spatial ontology called GUM and associate these concepts with the patterns identified by HCDM. We also discuss the converse and composition operations on HCDM patterns for qualitative spatial reasoning and compare it with other cardinal direction models.

Abstract:We present a novel methodology for the analysis of activities engaged in an organization such as the research conducted in a University department by mapping them to a related hierarchical taxonomy such as Classification of Computer Subjects by ACM (ACM-CCS). We start by collecting data of activities of the individual components of the organization and present them as the components fuzzy membership profiles over the subjects of the taxonomy. Our method generalizes the profiles in two steps. First step finds fuzzy clusters of taxonomy subjects according to the working of the organization. Second, each cluster is mapped to higher ranks of the taxonomy in a parsimonious way. Each of the steps is formalized and solved in a novel way. We build fuzzy clusters of the taxonomy leaves according to the similarity between individual profiles by using a novel, additive spectral, fuzzy clustering method that involves a number of model-based stopping conditions, in contrast to other methods. As the found clusters are not necessarily consistent with the taxonomy, each is considered as a query set. To lift a query set to higher ranks of the taxonomy, we develop an original recursive algorithm for minimizing a penalty function that involves 'head subjects' on the higher ranks of the taxonomy together with their 'gaps' and 'offshoots'. The method is illustrated by applying it to real-world data.

Abstract:Ontologies are expected to play an important role in many application domains, as well as in software engineering in general. One problem with using ontologies within software engineering is that while UML, a widely used standard for specifying and constructing the models for a software-intensive system, has a four-layer metamodelling architecture, the standard Web Ontology Language (OWL) does not support reasoning over layered metamodels. OWL 2 provides simple metamodelling by using a punning approach, however, the interpretation function is different based on the context, which leads to non-intuitive results. The OWL FA Language has a well defined metamodelling architecture. However, there is no study and tool for supporting reasoning over OWL FA. In this paper, we discuss some reasoning tasks in OWL FA. We also introduce the OWL FA Tool kit, a simple tool kit for manipulating and reasoning with OWL FA.

Abstract:Direction relations between extended spatial objects are important commonsense knowledge. Skiadopoulos proposed a formal model for representing direction relations between compound regions (the finite union of simple regions), known as SK-model. It perhaps is currently one of most cognitive plausible models for qualitative direction information, and has attracted interests from artificial intelligence and geographic information system. Originating from Allen first using composition table to process time interval constraints; composing has become the key technique in qualitative spatial reasoning to check the consistency. Due to the massive number of basic directions in SK-model, its composition becomes extraordinary complex. This paper proposed a novel algorithm for the composition. Basing the concepts of smallest rectangular directions and its original directions, it transforms the composition of basic cardinal direction relations into the composition of interval relations corresponding to Allen's interval algebra. Comparing with existing methods, this algorithm has quite good dimensional extendibility, that is, it can be easily transferred to the tridimensional space with a few modifications.

Abstract:A novel p-margin kernel learning machine (p-MKLM) with magnetic field effect is proposed in allusion to pattern classification problem in this paper. In the Mercer induced feature space, p-MKLM can effectively resolve one-class/binary classification problems. By introducing an adjustable magnetic field density q, the basic idea of p-MKLM is to find an optimal hyperplane with magnetic field effect such that the distance between one class and the hyperplane is as small as possible due to the magnetic attractive effect, while at the same time the margin between the hyperplane and the other class is as large as possible due to magnetic repulsion, thus implementing both maximum between-class margin and minimum within-class volume so as to improve the generalization capability of the proposed method. To construct such a hyperplane with magnetic field effect, it is only needed to solve a convex quadratic programming problem which can be effciently solved with the off-the- shelf software packages for training learning machine. Experimental results obtained with benchmarking datasets show that the proposed algorithm is effective and competitive to other related methods in such cases as two-class and one-class (or novelty detection) pattern classification respectively.

Abstract:The MAGA is an effective algorithm used for global numerical optimization problems. Drawbacks, however, still existed in the neighborhood selection part of the algorithm. Based on the social cooperate mechanism of agents, an effective neighborhood construction mode is proposed. This mode imports an acquaintance net which describes the relation of agents, and uses that to construct the local environment (neighborhood) for agents. This strategy makes the new mode more reasonable than that of MAGA. The Multi-Agent Social Evolutionary Algorithm (MASEA) based on this construction mode is introduced, and some standard testing functions are tested. In the first experiments, two dimensional, 30 dimensional and 20-1000 dimensional functions are tested to prove the effectiveness of this algorithm. The experimental results show MASEA can find optimal or close-to-optimal solutions at a low computational cost, and its solution quality is quite stable. In addition, the comparative results indicate that MASEA performs much better than the CMA-ES and MAGA in both quality of solution and computational complexity. Even when the dimensions reach 10,000, the performance of MASEA is still good.