Research Projects

Application-Oriented Research Projects

BAND

The Project BAND will develop advanced adaptive technologies for serving information on the Internet. Today most World Wide Web sites are either static or dynamic. But most of them fail to adapt to users needs. Answering to questions is possible with database queries or full text retrieval. For creating next generation web sites a new adaptive approach seems to be necessary. New tools and abstractions are needed which integrate dynamic servers and rich knowledge representation for presentation generation and information retrieval.

Beispiele

Literature:

Computing Probabilistic Least Common Subsumers in Description Logics

Th. Mantay, R. Möller, A. Kaplunova

To appear in: Proceedings KI-99, 23. Deutsche Jahrestagung für Künstliche Intelligenz, Springer-Verlag, 1999.

Computing least common subsumers in description logics is an important reasoning service useful for a number of applications. As shown in the literature, it can, for instance, be used for similarity-based information retrieval where information retrieval is performed on the basis of the similarities of user-specified examples. In this article, we first show that, for crisp DLs, in certain cases the set of retrieved information items can be too large. Then we propose a probabilistic least common subsumer operation based on a probabilistic extension of the description logic language ALN. We show that by this operator the amount of retrieved data can be reduced avoiding information flood.

Bericht über das Projekt BAND: Benutzeradaptiver Netzinformationsdienst

R. Möller, K. Hidde, R. Joswig, Th. Mantay, B. Neumann

LKI-Memo LKI-M-99/01, Labor für Künstliche Intelligenz, Fachbereich Informatik, Universität Hamburg, May 1999,
in German.
Postscript File, PDF File

In dem Projekt BAND wurden Grundlagen zur Entwicklung benutzeradaptiver Netz-Informationsdienste am Beispiel eines "TV-Assistenten" entwickelt. Zugriffs und Präsentationstechniken für Netzinformationen werden durch wissensbasierte Techniken dynamisch an die Bedürfnisse eines Benutzers angepaßt. Die erreichten Ergebnisse wurden anhand einer prototypischen Implementation des TV-Assistenten evaluiert.

BAND - Benuteradaptiver Netz-Informationsdienst

Proceedings KI-98, 22. Deutsche Jahrestagung für Künstliche Intelligenz, Selected German AI Research Projects, Springer-Verlag, 1998, p. 319.

Content-Based Information Retrieval by Computation of Least Common Subsumers in a Probabilistic Description Logic

Thomas Mantay, Ralf Möller

In: Proc. International Workshop on Intelligent Information Integration,
ECAI'98, Aug. 23-28, Brighton UK, 1998,
Postscript File, PDF File

Semantics-Based Information Retrieval

Ralf Möller Volker Haarslev, Bernd Neumann.

In: Proc. IT&KNOWS-98: International Conference on Information Technology and Knowledge Systems,
31. August- 4. September, Vienna, Budapest, 1998.
Postscript File, PDF File

In this paper we investigate the use of conceptual descriptions based on description logics for content-based information retrieval and present several innovative contributions. We provide a query-by-examples retrieval framework which avoids the drawback of a sophisticated query language. We extend an existing DL to deal with spatial and temporal concepts. We provide a content-based similarity measure based on the least common subsumer which extracts conceptual similarities of examples.

MONET

INDIA

Literature:

Resistive Networks Revisited: Exploitation of Network Structures and Qualitative Reasoning about Deviations is the Key

Lothar Hotz, Heiko Milde, Ralf Möller, Bernd Neumann

Proc. DX-97, The Eight International Workshop on Principles of Diagnosis. Mont-Saint-Michel, France. September 15-18, 1997.
Postscript File, PDF File

For diagnostic purposes, analog circuits may be qualitatively modeled as resistive networks. We demonstrate that approaches to this task show certain weaknesses if they are based on sign-based qualitative values. In order to overcome these deficiencies, we first introduce qualitative deviation values with a semantics that enables us to model different classes of faults arising in analog circuits. The qualitative values adequately describe different effects that faults may have. Then we present a sound and complete inference algorithm for computing these effects using qualitative operators and local propagation techniques.

Keywords: qualitative reasoning, reasoning about deviations, model-based diagnosis.

Using Behavior Deviations and an Interval-Based Calculus for Modeling Electronic Circuits

Lothar Hotz, Heiko Milde, Ralf Möller, Bernd Neumann

Proc. Workshop der Fachgruppe Modellbasiertes und Qualitatives Schließen, also published as LKI-Memo No. 97/1, 1997.
Postscript File

This paper introduces an approach for modeling and simulating a technical device for diagnostic purposes. A resistive network is used as a component-oriented device model which is automatically transformed into a simulation model that explicitly represents the global topological structure of the circuit. The simulation model is based on qualitative descriptions of behavior deviations and supports an interval-based qualitative calculus for reasoning about deviations.

Keywords: qualitative reasoning, reasoning about deviations, model-based diagnosis.

PETS

Literature:

Avoiding Combinatorial Explosion in Automatic Test Generation: Reasoning about Measurements is the Key

Harald Lange, Ralf Möller, Bernd Neumann

Proc. KI'96, Dresden, September 1996.
Postscript File

The main thesis of this paper is that reasoning about measurements can be used as a basic mechanism for generating test plans for analogical circuits. Motivated by an application scenario, reasoning about measurements incorporates domain knowledge about testing conditions, local behavior of circuit components (fault modes covered by measurements) and the topological structure of the circuit to be tested. With the test generation architecture introduced in this paper, a combinatorial explosion which is problematic in model-based test generation approaches can be avoided.

Keywords: qualitative reasoning, model-based test generation, diagnosing electric circuits.

PETS Manual

Alissa Kaplunova, Ralf Möller

Postscript File, PDF File

The graphical editor provided by the PETS system supports input and modification of object-oriented circuit diagrams for relay assembly groups. The object-oriented representation gen-erated by the graphical editor can be used for generating test plans. The PETS interface supports both components, the editor for circuit drawings and the browser for examining test plans generated by PETS.

This part of the PETS documentation presents classes and functions developed for the interactive interface (project part AP2). The classes for describing concrete circuit elements (like resistors, relay contacts, etc.) are designed to provide the basis for the graphical interface as well as for test plan generation processes. We also describe classes for modeling simple abstractions from the basic data structures (e.g. concerning connection topology) which are also required for consistency checking.

FIS

The system developed in this project has served as the basis for the development of the Hamburg Traffic Information System GEOFOX.

Literature (in German):

Suchalgorithmen und Interaktionstechniken für Fahrplan-Informationssysteme

Ralf Möller, Lothar Hotz

Labor für Künstliche Intelligenz, Bericht Nr. LKI-M-91/3, 1991.
Postscript File

This paper describes search algorithms to nd optimal connections in a local traffic system. AI search techniques are combined with optimization methods developed in other disciplines (e.g. optimization theory) to realize a prototype of a time-table information system. We present concepts of the user interface which visualizes the net of the traffic system in different views and at different scales.

Keywords: search techniques, applications of A*, traffic information systems.