Back to search

University of Macedonia Economic and Social Sciences

Ευφυή συστήματα στον σημασιολογικό ιστό

Abstract

dc:description

Traditional World Wide Web is human-centric, that is, information is available in human readable format. In older web sites information is static, usually in HTML, and is updated only when their owners devote time to edit it manually. On the other hand, modern sites have their content being dynamically pumped by databases and, although they exhibit some form of structure, the information is again targeting the human visitor of the web page.This gap aims to fill cover the semantic web, not by replacing the traditional web but by harmonically coexisting with it. Information in semantic web is in an absolutely structured format, aiming at its interpretation and editing by intelligent systems (without this to preclude its automated appearance in human-readable format).This thesis focuses on the use of intelligent systems in semantic web, giving emphasis in intelligent calendar applications. Its contribution is three-fold: Semantic representation of the information; its exploitation in specific application domains; and, finally, creating semantically annotated information from simple web pages.Initially, we have designed and proposed ontologies for semantically representing information that concerns daily human activities, in order to be exploited by intelligent calendar applications. Although we started the design of the ontologies having the already deployed intelligent calendar application SELFPLANNER in mind, the ontologies were developed fairly general so as to be usable in other similar systems as well. Furthermore, new concepts and use cases were added (e.g., meetings), that were not supported by the particular application. Finally, we have developed a library, named ONTPLANNER, to convert ontological descriptions of activities described with the proposed ontologies to Java objects, so as to directly insert them into an intelligent calendar application. This library has been integrated in the latest version of SELFPLANNER.Next, we have developed from the grounds a new intelligent system, aiming at the ontological description of educational objects, such as a university class. The new system, called COURSR, possesses descriptions of educational objects about their temporal aspects and requirements with a fine-grained level of detail. The system is able to connect with the student’s calendars, as well as with SELFPLANNER, to schedule simultaneously all of his activities, including educational, professional and personal ones. COURSR supports a new ontology about the structure and temporal aspects of educational objects. An interesting innovative feature concerns the use of natural (English) language to describe periodicity and constraints over their activities, with the underlying grammar and the parser having been developed from the grounds for this thesis. COURSR was evaluated with real users (teachers and students) with very promising results.Finally, within this thesis we have tried to extract semantic descriptions of activities from text in unknown pages. First we focused on information for educational objects, using university sites as the information source. However, the fine-grained level of information we were looking for was very complex and in most cases not available. Thus, for intelligent systems like COURSR we propose their adoption from large educational institutions and their feeding with educational object descriptions. As a simpler use case, we examined the extraction of semantically annotated information concerning activities related to attending a scientific conference, like submitting a research paper, sending the camera ready version of it, preparing the presentation and, of course, going to the conference and attending it. We have developed a system, called FUZZYEXTRACTOR, that, by reading the unknown web page of a conference, extracts all the critical dates and created ontological descriptions of the related activities in a form ready to insert to the user’s intelligent calendar application.

Degree

thesis:*
Grantor dc:publisher
University of Macedonia Economic and Social Sciences
Year dc:date
2015

Author and committee

dc:creator, dc:contributor.*
Authors dc:creator
  • Agnantis, Konstantinos
  • Αγναντής, Κωνσταντίνος

Subjects

dc:subject × 12

Rights

dc:rights
Statement dc:rights
  • BY_NC_SA
Language dc:language
gre

Identifiers

dc:identifier.*
Identifier
10.12681/eadd/41820

Chain of custody

source
Harvested from
Greek National Archive of PhD Theses
Base URL
phdtheses.ekt.gr/eadd_oai/request
Last updated
2026-11-18
Source record
OAI-PMH GetRecord
citation

Agnantis, Konstantinos; Αγναντής, Κωνσταντίνος. Ευφυή συστήματα στον σημασιολογικό ιστό. University of Macedonia Economic and Social Sciences, 2015. http://hdl.handle.net/10442/hedi/41820