Â¥Ö÷

Summer school Dresden 2006 ICCL Summer School 2006  -  Course Program Reasoning in Description Logics. Franz Baader    (Technische Universität Dresden, Germany) Description Logics (DLs) are a successful family of logic-based knowledge representation formalisms, which can be used to represent the conceptual knowledge of an application domain in a structured and formally well-understood way. They are employed in various application domains, such as natural language processing, configuration, and databases, but their most notable success so far is the adoption of the DL-based language OWL as standard ontology language for the semantic web. This course concentrates on designing and analyzing reasoning procedures for DLs. It is not concerned with how to use DLs in applications or how to obtain efficient implementations of the described algorithms. After a short introduction into DLs, it describes both tableau- and automata-based algorithms for expressive DLs. Then, it analyzes the computational complexity of reasoning in various DLs, and finally considers reasoning in inexpressive DLs, which allow for polynomial-time algorithms. Finger Exercises in Formal Concept Analysis. Bernhard Ganter    (Technische Universität Dresden, Germany) An introduction is given to the theory and practice of Formal Concept Analysis. We will demonstrate that its most elementary definitions already can be of concrete use in practice, and will invite the participants to basic exercises with concrete data. Building on that, we discuss the interplay of methods, algorithms, theory and implementations. Knowledge, Reasoning, and the Semantic Web. Pascal Hitzler    (AIFB Universität Karlsruhe, Germany) With amazing speed, the world wide web has become a widespread means of communication and information sharing. Today, it is an integral part of our society, and will continue to grow. However, most of the information available cannot be processed easily by machines, but has to be read and interpreted by humans. In order to overcome this limitation, a world-wide research effort is currently being undertaken to make the contents of the world wide web accessible, interpretable, and usable by machines. The resulting extension of the World Wide Web is commonly referred to as the Semantic Web, and the underlying technological infrastructure which is currently being developed is referred to as Semantic Technologies. In this process, a key idea is that web content should be provided with conceptual background - often referred to as ontologies - which allows machines to put information into context, making it interpretable. Most recently - having established RDF and RDFSchema as basic syntax - the OWL Web Ontology Language, which is a decidable fragment of first-order logic, has been recommended by the world wide web consortium (W3C) for the ontology vocabulary. Conceptual knowledge is provided by means of statements in some logical framework, and the discussion concerning suitable logics is still ongoing. Description Logics play a major role, as they provide the foundation for OWL, but other approaches are also being considered. Currently, the development of an expressive rule-based logic layer on top of OWL for the inference of ontological knowledge is being investigated. But also fragments of OWL, including Horn and propositional languages, are being used, as different application scenarios necessitate different tradeoffs between expressiveness, conceptual and computational complexity, and scalability. In this course, we will talk about the semantic web, its vision, the current state of the art, and challenges for the future. We will introduce standards and proposals for representing and reasoning with knowledge on the web, including freely available tools for content and knowledge management and automated deduction. The course will conclude with a critical discussion of future research challenges. Contents: The Semantic Web Vision. Knowledge Representation on the Web via Ontologies as Metadata. Web Ontology Language OWL. Rule Languages for the Web. Reasoning and the Semantic Web. Challenges for future research. Text clustering and Semantic Web mining. Andreas Hotho    (Universität Kassel, Germany) An important technique in Semantic Web Mining is to combine similar text documents to "`clusters"'. However, common text clustering techniques offer rather poor capabilities to their users why a particular result has been achieved. They have the disadvantage that they do not relate semantically nearby terms and that they cannot explain how resulting clusters are related to each other. We discuss a way of integrating a large thesaurus and the computation of lattices of the resulting clusters to overcome these two problems. As its major result, our approach achieves an explanation using an appropriate level of granularity at the concept level as well as an appropriate size and complexity of the explaining lattice of resulting clusters. Inductive Logic Programming. Stefan Kramer    (Technische Universität M¨¹nchen, Germany) The course will give a survey of modern techniques for inductive logic programming (ILP) and multi-relational data mining (MRDM). ILP is concerned with learning in logic (mostly clausal theories), whereas MRDM focuses on the analysis of relational databases. After a short introduction, various representations of data, from propositional to fully relational will be presented. Given these preliminaries, the course will first focus on representation changes, transforming relational data into propositional form (propositionalization). One of the algorithms presented, Warmr, is an "upgrade" of the well-known data mining algorithm APriori, finding frequently succeeding queries instead of frequent itemsets. Subsequently, I will present two classical rule and decision tree induction algorithms from ILP, FOIL and Tilde, that can be viewed as upgrades of popular propositional machine learning schemes. Next, instance-based and distance-based methods (e.g., k-Nearest-Neighor) for relational data will be discussed. In the final lecture, the most fundamental ideas of statistical relational learning (SRL) will be sketched. A literature survey of clustering algorithms: applications to biomedicine and software engineering. Cluster Methodology Bill Andreopoulos    (York University, Toronto, Canada) We will discuss clustering algorithms for categorical and numerical data. We leverage these data types to problems in software clustering, protein interaction network clustering, biomedical data set clustering. We discuss methods for hierarchical clustering, partioning clustering, model-based clustering, grid-based clustering, density-based clustering, spectral clustering and conceptual clustering. We discuss general problems of clustering such as formally defining the goals, time-accuracy tradeoffs and the curse of dimensionality. We will also discuss supervised classification methods (support vector machines), principal component analysis, multi-dimension reduction and singular value decomposition. Machine Learning and Formal Concept Analysis. Sergei Kuznetsov    (VINITI, Moscow, Russia) A model of learning from positive and negative examples is naturally described in terms of Formal Concept Analysis (FCA). In these terms, a result of learning consists of two sets of intents (closed subsets of attributes): the first contains intents such that the corresponding extents consists only of positive examples. The second contains intents such that the corresponding extents consists only of negative examples. On the one hand, we show how FCA allows one to realize learning with various data representations, from standard object-attributes to labeled graphs. On the other hand, we use FCA to describe some standard models of Machine Learning such as version spaces and decision trees. http://www.computational-logic.org/content/events/iccl-ss-2006/lectures.php?id=24    Êղؠ  ·ÖÏí   ¶¥(0)