|
|||||||||||||||||||
|
|
||||||||||||||||||
People at
|
PreSInt: PREFerence-based Scene Interpretation
Das Vorhaben befasst sich mit allgemeingültigen rechnerbasierten Methoden zur Deutung komplexer stationärer oder zeitveränderlicher visueller Szenen aus der Alltagswelt, z.B. Innenraumszenen im Kontext von Assistenzaufgaben oder Verkehrsszenen im Kontext von Überwachungsaufgaben. Szeneninterpretationen dieser Art erfordern einerseits umfangreiches Vorwissen über die relevanten Alltagsvorgänge, repräsentierbar mit Methoden der Wissensrepräsentation, andererseits probabilistische Modelle zur Steuerung unsicherer Entscheidungen und zur Prädiktion. In diesem Vorhaben wird eine besondere Form der Integration von probabilistischen Modellen mit formaler Wissensrepräsentation untersucht, bei der probabilistische Inferenzen mit klassischen logikbasierten Inferenzen bei der Szeneninterpretation kooperieren. Logikbasierte Inferenzen grenzen den Raum möglicher konsistenter Szeneninterpretationen ab, während probabilistische Inferenzen unter den logisch möglichen Interpretationen bevorzugte bestimmen.
BOEMIE: Bootstrapping Ontology Evolution with Multimedia Information Extraction
BOEMIE will pave the way towards automation of the process of knowledge acquisition from multimedia content, by introducing the notion of evolving multimedia ontologies which will be used for the extraction of information from multimedia content in networked sources, both public and proprietary. BOEMIE advocates a synergistic approach that combines multimedia extraction and ontology evolution in a bootstrapping process involving, on the one hand, the continuous extraction of semantic information from multimedia content in order to populate and enrich the ontologies and, on the other hand, the deployment of these ontologies to enhance the robustness of the extraction system. The ambitious scope of the BOEMIE project and the proven specialized competence of the carefully composed project consortium ensure that the project will achieve the significant advancement of the state of the art needed to successfully merge the component technologies. The main measurable objective of the project is to improve significantly the performance of existing single-modality approaches in terms of scalability and precision. Towards that goal, BOEMIE will develop a new methodology for extraction and evolution, using a rich multimedia semantic model, and realized as an open architecture. The architecture will be coupled with the appropriate set of tools, implementing the advanced methods that will be developed in BOEMIE. Furthermore, BOEMIE aims to initiate a new research activity on the automation of knowledge acquisition from multimedia content, through ontology evolution. The resulting technology has a wide range of applications in commerce, tourism, e-science, etc. During the project, the technology will be evaluated through the development of an automatic content collection and annotation service for public events in a number of major European cities. The extracted semantic information will enrich a digital map, which will provide a friendly interface to the end user.
Paper: Silvana Castano, Alfio Ferrara, Davide Lorusso, Tobias
H. Näth and R. Möller.
Mapping Validation by Probabilistic Reasoning
5th
European Semantic Web Conference (ESWC 2008),
2008.
Bibtex
entry Paper
(PDF)
Poster: Tobias H. Näth and R. Möller.
ContraBovemRufum: A System for Probabilistic Lexicographic Entailment
21st
International Workshop on Description Logics (DL2008),
2008.
Bibtexentry
Paper
(PDF)
Project Deliverable: S. Espinosa, A. Kaya, S. Melzer, R.
Möller, T. Näth, and M. Wessel.
Reasoning Engine Version 2. Technical report, Hamburg University Of
Technology, 2007.
BOEMIE Project Deliverable D4.5.
Bibtex
entry Paper
(PDF)
Master Thesis: Tobias H. Näth. Analysis of the
average-case behavior
of an inference algorithm for probabilistic description logics.
Diplomarbeit, TU Hamburg-Harburg, February 2007.
Bibtex
entry Paper
(PDF)
Project Work: Tobias H. Näth. Modellierung von
Lehrveranstaltungen
und ihre Abbildung in Learning Management Systeme. Studienarbeit, TU
Hamburg-Harburg, August 2003.
Bibtex
entry Paper
(PDF)
The ContraBovemRufum System is an implementation of the algorithms for P-SHOQ(D) described in the INFSYS RESEARCH REPORT 1843-02-06 by Rosalba Giugno and Thomas Lukasiewicz. It is able to compute consistency of a probabilistic knowledge base and tight logical and lexicographic entailment with respect to a PTbox and PAbox. Reasoning with respect to ABoxes has not been implemented yet. For details on the ContraBovemRufum System have a look at ”Analysis of the average-case behavior of an inference algorithm for probabilistic description logics”
Download ContraBovemRufum Version 0.1 here
