Catálogo de publicaciones - libros

Psychology and cognitive science show that human concepts possess particular structures (conceptual structures). However, in the process of ontology modeling information concerning the structure of human concepts is lost. In ontologies concepts are typically represented as undifferentiated collections of necessary (or necessary and sufficient) conditions. The lack of representation of conceptual structure may cause ontologies to be inadequate and may limit their usability for human users. We present an attempt to bring ontology modeling closer to theories of conceptual structures, in particular to psychological essentialism. A metaontology is developed to support the representation of conceptual structure, in particular the distinction between essential and merely necessary conditions.

In this paper we propose an intelligent web search method with customized results. This approach adopts a cosine method to calculate the similarity between document profile and customer profile. The document profile is derived from the similarity score of documents. The customers’ search history is captured to generate customer profile. Then the customized search results are recommended to the end users based upon the similarity between document profile and customer profile.

In this paper we propose ontology-based improvement of the Croatian Web directory search mechanism, which is currently not capable of executing queries that take into account the structure and semantics of user’s query. The proposed approach is verified by introducing an ontology in the domain (directory category) of “tourism”. We address three problems related to the search mechanism: low recall, high recall and low precision and vocabulary mismatch. The results show significant improvements in terms of subjective relevance and quality of results.

Several Web-based systems need complex reasoning mechanisms and problem solving capabilities in order to perform their tasks. These systems should protect their users against the complexity of their inference engine. Such a protection should be offered both to the final users and to the domain experts that instantiate the Web systems on the different particular domains. We claim that to this purpose an intermediate representation layer, with the role of filling the gap between the implementation-oriented view of the domain (needed by the reasoning module) and the human-oriented view of the same domain (suitable both for the final users and for the domain experts), is required. This intermediate layer also provides a representation which is independent from the specific reasoning approach adopted. In the paper we discuss these issues by referring to a specific example, i.e. the STAR system (a Web-based system which exploits a configuration engine to support the user in organizing a personalized tour) together with STAR-IT, a tool that supports the instantiation of STAR on different domains.

The tour guide services are a LBS application that provides the information based on the location of users. The recent researches of the tour guide services focus on the context-sensitive computing, but the tour guide services based on HTML contents can provide only the static information. Therefore, the goals of our research are to semantically search the information. For the purpose, we describe the ontologies of the tour guide services, and express the tour information as the RDF contents based on the ontologies. In this paper, we present the architecture of the tour information Services based on the Semantic Web technologies and describe the process of the implementation.

Web service is the next ware of Internet computing. Discovery web services is becoming a new challenge due to the increasing number of available services on the World Wide Web. In this paper, I investigate the semantics discovery of Web services based on domain ontology. Multiple service discovery strategies are defined and explained. The semantics service discovery methods provide a new way to locate and utilize published Web services. The approach is flexible and extensible to accomplish complex web service requests.

There is currently a growing new focus in data mining – Ubiquitous Data Mining (UDM). UDM is the process of mining data streams in a ubiquitous environment, on resource constrained devices [KPP02]. UDM is widely applied in facilitating real-time decision making in mobile and highly dynamic environments/applications, such as road safety and mobile stock portfolio monitoring. A significant challenge in these contexts is the interpretation and analysis of results produced through unsupervised techniques (which are invaluable since little is known about the streamed data). We propose a novel fuzzy approach that leverages the significant benefits of UDM clustering and supplements the interpretation and use of these results through using expert/background knowledge.

Clustering is a difficult problem especially when we consider the task in the context of a data stream of categorical attributes. In this paper, we propose σ -SCLOPE, a novel algorithm based on SCLOPE’s intuitive observation about cluster histograms. Unlike SCLOPE however, our algorithm consumes less memory per window and has a better clustering runtime for the same data stream in a given window. This positions σ -SCLOPE as a more attractive option over SCLOPE if a minor lost of clustering accuracy is insignificant in the application.

Even though there are many databases for gene/protein interactions, most such data still exist only in the biomedical literature. They are spread in biomedical literature written in natural languages and they require much effort such as data mining for constructing well-structured data forms. As genomic research advances, knowledge discovery from a large collection of scientific papers is becoming more important for efficient biological and biomedical researches. In this paper, we present a relation kernel based interaction extraction method to resolve this problem. We extract gene/protein interactions of Yeast () from text documents with relation kernel. Kernel for relation extraction is constructed with predefined interaction corpus and set of interaction patterns. Proposed relation kernel for interaction extraction only exploits shallow parsed documents. Experimental results show that the proposed kernel method achieves a recall rate of 78.3% and precision rate of 79.9% for gene/protein interaction extraction without full parsing efforts.

The ubiquity of textual information nowadays reflects its great significance in knowledge discovery. However, effective usage of these textual materials is always hampered by data incompleteness in real-life applications. In this paper, we apply a closest fit approach to attack textual missing values. To evaluate the closeness of texts in this application, we present an order perspective of text similarity and propose a hybrid order-semisensitive measure, M-similarity, to capture the proximity of texts. This measure combines single item matching, maximum sequence matching and potential matching and get a proper balance between usage of sequence information and efficiency. We incorporate M-similarity into two closest fit methods to missing values in textual attributes and evaluate them on data sets of Traditional Chinese Medicine (TCM). Experimental results illustrate the effectiveness of these methods with M-similarity.