Search

The goal of DELOS Task 4.8 Task-centered Information Management is to provide the user with a Task-centered Information Management system (TIM), which automates user's most frequent activities, by exploiting the collection of personal documents. In previous work we have explored the issue of managing personal data by enriching them with semantics according to a Personal Ontology, i.e. a user-tailored description of her domain of interest. Moreover, we have proposed a task specification language and a top-down approach to task inference, where the user specifies main aspects of the tasks using forms of declarative scripting. Recently, we have addressed new challenging issues related to TIM user's task inference. More precisely, the first main contribution of this paper is the investigation of task inference theoretical issues. In particular, we show how the use of the Personal Ontology helps for computing simple task inference. The second contribution is an architecture for the system that implements simple task inference. In the current phase we are implementing a prototype for TIM whose architecture is the one presented in this paper. [ABSTRACT FROM AUTHOR]

To achieve improved availability and performance, often, local copies of remote data from autonomous sources are maintained. Web search engines are the primary examples of such services. Increasingly, these services are utilizing the Semantic Web as it is often envisioned as a machine-interpretable web. In order to keep the local repositories current, it is essential to synchronize their content with that of their original sources. Change detection is the first step to accomplish this. It is essential to have efficient change detection mechanisms as the size of the local repositories is often very large. In this paper, we present an approach that exploits the semantic relationships among the concepts in guiding the change detection process. Given changes to some seed instances, a reasoning engine fires a set of pre-defined rules to characterize the profile of the changed target instances. In addition to change detection, our proposed semantics-based approach of utilizing semantic associations can be utilized in other applications such as guiding information discovery for agents, consistency maintenance among distributed information sources, among others. [ABSTRACT FROM AUTHOR]

XML is rapidly emerging as a dominant standard for representing and exchanging information. The ability to transform and present data in XML is crucial and XSLT is a relatively recent programming language, specially designed to support this activity. Despite its utility, however, XSLT is widely considered a difficult language to learn. In this paper, we present a novel system called XSLTGen, an automatic XSLT Generator. This system automatically generates an XSLT stylesheet, given a source XML document and a desired output HTML or XML document. It allows users to become familiar with and learn XSLT stylesheets, based solely on their knowledge of XML or HTML. Our method for automatically generating XSLT transformations is based on the use of semantic mappings between the input and output documents. We show how such mappings can be first discovered and then employed to create XSLT stylesheets. The results of our experiments show that XSLTGen works well with a number of different varieties of XML and HTML documents. [ABSTRACT FROM AUTHOR]

Ontologies have been created for many different subjects and by independent groups around the world. The nonexistence of a commonly accepted and used general purpose upper-ontology makes it difficult to integrate these ontologies through merge and alignment operations. The majority of the algorithms proposed so far rely on syntactic analysis, disregarding the structural properties of the source ontologies. In our previous work, we proposed an alignment method that considers the structural properties of an upper-ontology constructed using a thesaurus and Formal Concept Analysis technique (FCA). We also analyzed the FCA's lattice structure and proposed a measure of similarity based on Tversky's model, which allowed us to identify closely related concepts in different source ontologies. In this paper, we apply the alignment method to ontologies developed for a completely different domain, and enhance the solution by providing a navigational aid for the lattice. It is well known that one of the main drawbacks of the application of FCA is that the resulting lattice soon becomes cluttered when the number of objects and attributes increases. The proposed solution is based on hyperbolic visualization and on structural elements of the lattice. Keywords: ontology alignment, Formal Concept Analysis, lattice visualization, similarity measures [ABSTRACT FROM AUTHOR]

Inclusion dependencies (INDs) between databases are assertions of subset-relationships between sets of attributes (dimensions) in two relations. Such dependencies are useful for a number of purposes related to information integration, such as database similarity discovery and foreign key discovery. An exhaustive approach at discovering INDs between two relations suffers from the dimensionality curse, since the number of potential mappings between the attributes of two relations is exponential in the number of attributes. For this reason, levelwise (Apriori-like) approaches at discovery do not scale beyond relations with 8 to 10 attributes. Approaches modeling the similarity space as graphs or hypergraphs are promising, but also do not scale very well. This paper discusses approaches to scale discovery algorithms for INDs and some other similarity patterns in databases. The major obstacle to scalability is the exponentially growing size of the data structure representing potential INDs. Therefore, the focus of our solution is on heuristic techniques that reduce the number of IND candidates considered by the algorithm. Despite the use of heuristics, the accuracy of the results is good for real-world data. Experiments are presented assessing the quality of the discovery results versus the runtime savings. We conclude that the heuristic approach is useful and improves scalability significantly. It is particularly applicable for relations that have attributes with few distinct values. [ABSTRACT FROM AUTHOR]

Roles are meant to capture dynamic and temporal aspects of real-world objects. The role concept has been used with many semantic meanings: dynamic class, aspect, perspective, interface or mode. This paper identifies common semantics of different role models found in the literature. Moreover, it presents a set of conceptual modelling patterns for the role concept that include both the static and dynamic aspects of roles. In particular, we propose the Role as Entity Types conceptual modelling pattern to deal with the full role semantics. A conceptual modelling pattern is aimed at representing a specific structure of knowledge that appears in different domains. The use of these patterns eases the definition of roles in conceptual schemas. In addition, we describe the design of schemas defined by using the patterns in order to implement them in any object-oriented language. [ABSTRACT FROM AUTHOR]

In the past, most conceptual schemas of information systems have been developed essentially from scratch. Currently, however, several research projects are considering an emerging approach that tries to reuse as much as possible the knowledge included in existing ontologies. Using this approach, conceptual schemas would be developed as refinements of (more general) ontologies. However, when the refined ontology is large, a new problem that arises using this approach is the need of pruning the concepts in that ontology that are superfluous in the final conceptual schema. This paper proposes a new method for pruning ontologies in this approach. We also show how to adapt the method to prune ontologies in other contexts. Our method is general and it can be adapted to most conceptual modeling languages. We give the complete details of its adaptation to the UML. On the other hand, the method is fully automatic. The method has been implemented. We illustrate the method by means of its application to a case study that refines the Cyc ontology. [ABSTRACT FROM AUTHOR]

In open networked systems a varying number of nodes interact each other just on the basis of their own independent ontologies and of knowledge discovery requests submitted to the network. Ontology matching techniques are essential to enable knowledge discovery and sharing in order to determine mappings between semantically related concepts of different ontologies. In this paper, we describe the H-Match algorithm and related techniques for performing matching of independent ontologies in open networked systems. A key feature of H-Match is that it can be dynamically configured for adaptation to the semantic complexity of the ontologies to be compared, where the number and type of ontology features that can be exploited during the matching process is not known in advance as it is embedded in the current knowledge request. Furthermore, this number can vary, also for the same ontologies, each time a new matching execution comes into play triggered by a knowledge request. We describe how H-Match enforces this capabilities through a combination of syntactic and semantic techniques as well as through a set of four matching models, namely surface, shallow, deep, and intensive. Then, we describe the application of H-Match and its implementation for knowledge discovery in the framework of the Helios peer-based system. Finally, we present experimental results of using H-Match on different test cases, along with a discussion on precision and recall. [ABSTRACT FROM AUTHOR]

The traditional perception of isolated data systems is changing to a new one where the interest of a real and efficient interoperation among those data systems is recognized. However, many problems must be solved yet before a real interoperation becomes true. In order to overcome the existing problems, there is a considerable number of proposals that can be found in the specialized literature that promote the idea of semantic interoperability. A new framework to achieve semantic interoperability among agent based data systems at a communication level is proposed in this paper. This framework permits agents belonging to different data systems 1) to send each other suitable messages without requiring the establishment of a common communication pattern in advance; 2) to understand, completely or partially, those messages that are interchanged among themselves; and 3) to invoke web services offered by the data systems at a high level without needing to go into technical details. An ontology that captures the semantics of different communication acts is the key element for supporting the functionalities provided by the framework. Furthermore, the framework has been extended to support semantic descriptions of web services, which favor their automatic discovery. The usefulness of the presented framework is evaluated using two case study of interoperation among heterogeneous data systems; on one side through the agents of those systems, and on the other side, through the combination of an agent and a web service. [ABSTRACT FROM AUTHOR]

We introduce a novel view on how to deal with the problems of semantic interoperability in distributed systems. This view is based on the concept of emergent semantics, which sees both the representation of semantics and the discovery of the proper interpretation of symbols as the result of a self-organizing process performed by distributed agents exchanging symbols and having utilities dependent on the proper interpretation of the symbols. This is a complex systems perspective on the problem of dealing with semantics. We highlight some of the distinctive features of our vision and point out preliminary examples of its application. [ABSTRACT FROM AUTHOR]

Real data is often affected by errors and inconsistencies. Many of them depend on the fact that schemas cannot represent a sufficiently wide range of constraints. Data cleaning is the process of identifying and possibly correcting data quality problems that affect the data. Cleaning data requires to gather knowledge on the domain to which the data refer. Anyway, existing data cleaning techniques still access this knowledge as a fragmented collection of heterogenous rules and ad hoc data transformations. Furthermore, data cleaning methodologies for an important class of data based on the semistructured XML data model have not yet been proposed. In this paper we introduce the OXC framework, that offers a methodology for XML data cleaning based on a uniform representation of domain knowledge through an ontology We describe how to define XML related data quality metrics based on our domain knowledge representation, and give a definition of various metrics related to the completeness data quality dimension. [ABSTRACT FROM AUTHOR]