[TU Dresden]

Komplexpraktikum/Project in SS 2010

Technische Universität Dresden
Institut für Theoretische Informatik
Lehrstuhl für Automatentheorie

Initial meeting
takes place on April 21 at 13:30 in Room INF-3027. Attending the initial meeting is mandatory for participation in the praktikum/project.
People who want to participate in the praktikum/project, but have serious reasons to not attend the initial meeting, please contact Rafael Peñaloza until April 20.

Position in curriculum
- Diplomstudiengang Informatik (Diplom- und Bakkalaureatsabschluß), ab 5. Semester ;Wahlpflichtveranstaltung (-/-/4)
- Course of studies Computation Logic; project (12 credits)

for computer scientists: Pflichtvorlesung "Grundlagen der Theoretischen Informatik"

-There will be an initial meeting (see above) where different topics will be proposed to the students. Students can chose from the offered topics, one to work on.
- Students interested in doing their project, but unable to assist to the initial meeting should contact Rafael Peñaloza (:penaloza at inf.tu-dresden.de:) to discuss possible solutions.
- Each student is assigned a tutor, depending on the topic chosen. During the semester, there will be regular meetings of the student and his tutor.
- The results of the praktikum/project will be presented at the end of the semester in a talk given by the student.

Concerning the final presentations, students may choose to present their work in German or in English.

Participants Duties
The participants are expected to read the relevant literature, and to discuss it with their tutor in order to become acquainted with the topic chosen. The required implementation work should be carried out in a structured way, and has to be documented appropriately. If a topic is shared by two or more participants, acquiring team-working skills is another goal of the project. The results of the project have to be described in a project paper (~15 pages) and presented in a 30 minutes talk at the end of the semester.

It is also the duty of the participants to reserve enough time for performing the project. The sharp deadline for finishing the project is the beginning of the following semester, i.e. the allowed time for the project is one semester plus the following semester break. Failure to finish the project in time will result in no credits to be given. It is the obligation of the participant to start the project in time, and to make appointments with the supervisor for regular meetings during the semester.

When choosing a topic, please take into account the knowledge you have already acquired. For example, if you'd like to do a project concerning knowledge representation, you are expected to have successfully attended the lecture "Logic-based knowledge representation" before starting the project.

(1) Implementation of a planning algorithm for action formalisms
The project is about implementing a planning algorithm for action formalisms based on description logics with the programming language Java. After the implementation of the algorithm, the student is supposed to evaluate the implementation with significantly many testing data. The planning algorithm is provided by the following paper:
Maja Milicic. Complexity of Planning in Action Formalisms Based on Description Logics. In Proceedings of the 14th International Conference on Logic for Programming, Artificial Intelligence, and Reasoning (LPAR 2007), Lecture Notes in Artificial Intelligence. Springer-Verlag, 2007.

The planning algorithm requires the knowledge about action formalisms based on DLs. For this reason, the student should get familiar with the following paper:

F. Baader, C. Lutz, M. Milicic, U. Sattler, and F. Wolter. Integrating Description Logics and Action Formalisms: First Results. In Proceedings of the Twentieth National Conference on Artificial Intelligence (AAAI-05), Pittsburgh, PA, USA, 2005.

Tutor: Hongkai Liu

(2) Explaining GCIs in OntoComp using SuperModel
OntoComp is a Plugin for the ontology editor Protege. Its purpose is knowledge base completion. It uses a method that has been derived from FCA where a human expert is repeatedly presented general inclusion axioms (GCIs) written in a description logic (DL) language that she can add to the ontology. Sometimes it may be unclear to the expert why a certain GCI is presented. One approach to make these GCIs more understandable is to present a model of the knowledge base in which the GCI holds (or one in which it does not hold). The tool "SuperModel" allows a user to generate, display and explore such models. The goal of this project is to extend OntoComp such that it can call SuperModel when the expert is in doubt about a GCI.

Tutor: Felix Distel

(3) Implementing a Fuzzy DL Reasoner
The Description Logic (DL) community has started to show a greater interest in the ability of reasoning under uncertainty. In fuzzy DL, each axiom is annotated with a "degree of truth". It was recently shown that readily available DL reasoners can be used -- without modification -- to compute the degree of truth of the consequences of a fuzzy ontology. In this project, the student is expected to implement a tool, based on existing DL Reasoners, that computes the truth degree of a given consequence. The implementation should preferably be in JAVA.

Tutor: Rafael Peñaloza

Rafael Peñaloza