Catálogo de publicaciones - libros

Formulating complex queries is hard, especially when users cannot understand all the data structures of multiple complex knowledge bases. We see a gap between simplistic but user friendly tools and formal query languages. Building on an example comparison search, we propose an approach in which reusable search components take an intermediary role between the user interface and formal query languages.

Existing semantic search tools have been primarily designed to enhance the performance of traditional search technologies but with little support for ordinary end users who are not necessarily familiar with domain specific semantic data, ontologies, or SQL-like query languages. This paper presents SemSearch, a search engine, which pays special attention to this issue by providing several means to hide the complexity of semantic search from end users and thus make it easy to use and effective.

In this paper we introduce a novel kind scenario where users use Rich Personal Semantic Web Clients to cooperatively create knowledge within “Semantic Web Communities”. Such communities are formed around P2P channels which work by exchanging patches of RDF information among clients. Once sufficient information has been collected locally at each client, rich and fast browsing of such "Semantic Web" becomes possible without generating external traffic or computational load. A prototype of such client, DBin, is presented and issues such as user interfaces and social aggregation model are discussed. We will focus in particular on the "Brainlet" paradigm, which enables community leaders to create and deliver domain specific user interfaces. The Brainlet creation process does not require programming skills, so that Semantic Web communities can be started up by domain experts rather than programmers.

Many communities need to organize and structure data to improve their utilization and sharing. Much research has been focused on this problem. Many solutions are based on a Terminological and Ontological Resource (TOR) which represents the domain knowledge for a given application. However TORs are often designed without taking into account heterogeneous data from specific resources. For example, in the biomedical domain, these sources may be medical reports, bibliographical resources or biological data extracted from GOA, Gene Ontology or KEGG. This paper presents an integrated visual environment for knowledge engineering. It integrates heterogeneous data from domain databases. Relevant concepts and relations are thus extracted from data resources, using several analysis and treatment processes. The resulting ontology embryo is visualized through a user friendly adaptive interface displaying a knowledge map. The experiments and evaluations dealt with in this paper concern biological data.

In this article we report on a user study aimed at evaluating and improving a thesaurus browser. The browser is intended to be used by documentalists of a large public audio-visual archive for finding appropriate indexing terms for TV programs. The subjects involved in the study were documentalists of the Dutch National Audiovisual Archives and of broadcasting corporations. The study provides insight into the value of various thesaurus browsing and searching techniques.

The Semantic Web technology should enable publishing of numerous resources of scientific and other, highly formalized data on the Web. The application of mining these huge, networked Web repositories seems interesting and challenging. In this paper we present and discuss an inductive reasoning procedure for mining frequent patterns from the knowledge bases represented in OWL DLP. OWL DLP, also known as Description Logic Programs, lies at the intersection of the expressivity of OWL DL and Logic Programming. Our method is based on a special trie data structure inspired by similar, efficient structures used in classical and relational data mining settings. Conjunctive queries to OWL DLP knowledge bases are the language of frequent patterns.

Case-based reasoning (CBR) uses various knowledge containers for problem solving: cases, domain, similarity, and adaptation knowledge. These various knowledge containers are characterised from the engineering and learning points of view. We focus on adaptation and similarity knowledge containers that are of first importance, difficult to acquire and to model at the design stage. These difficulties motivate the use of a learning process for refining these knowledge containers. We argue that in an adaptation guided retrieval approach, similarity and adaptation knowledge containers must be mixed. We rely on a formalisation of adaptation for highlighting several knowledge units to be learnt, i.e. dependencies and influences between problem and solution descriptors. Finally, we propose a learning scenario called “active approach” where the user plays a central role for achieving the learning steps.

Background knowledge is a natural resource for knowledge-intensive methods: Its exploitation can often improve the quality of their results significantly. In this paper we present a methodological view on knowledge-intensive subgroup discovery: We introduce different classes and specific types of useful background knowledge, discuss their benefit and costs, and describe their application in the subgroup discovery setting.

This paper concerns the iterative implementation of a knowledge model in a data mining context. Our approach relies on coupling a Bayesian network design with an association rule discovery technique. First, discovered association rule relevancy isenhanced by exploiting the expert knowledge encoded within a Bayesian network, i.e., avoiding to provide trivial rules w.r.t. known dependencies. Moreover, the Bayesian network can be updated thanks to an expert-driven annotation process on computed association rules. Our approach is experimentally validated on the Asia benchmark dataset.

We propose a multilayered semantic social network model that offers different views of common interests underlying a community of people. The applicability of the proposed model to a collaborative filtering system is empirically studied. Starting from a number of ontology-based user profiles and taking into account their common preferences, we automatically cluster the domain concept space. With the obtained semantic clusters, similarities among individuals are identified at multiple semantic preference layers, and emergent, layered social networks are defined, suitable to be used in collaborative environments and content recommenders.