October 28 – November 2 , 2012, Dagstuhl Seminar 12441
Foundations and Challenges of Change and Evolution in Ontologies
1 / 2 >
For support, please contact
Seminar was rescheduled
New date is 28.10.-02.11.12,
Old date was 26.-31.08.12
Dagstuhl office will inform all invited scientists
An ontology in computer science is an explicit, formal specification of the terms of a domain of application, along with the relations among these terms. An ontology provides a (structured) vocabulary which forms the basis for the representation of general knowledge. Ontologies have found extensive application in Artificial Intelligence and the Semantic Web, as well as in areas such as software engineering, bioinformatics, and database systems.
Research in ontologies in Artificial Intelligence has focussed on description logics (DL), where a description logic can be regarded as a (decidable) fragment of first order logic. Historically a DL is divided into two components, a so-called TBox, for expressing concepts and their interrelationships, and an ABox that contains assertions about specific individuals and instances. Thus, the TBox characterises a domain of application while the ABox contains information on a specific instance of a domain. A key point in description logics is that, via their limited expressiveness, one obtains "good", ideally tractable, inference algorithms. The number of description logics is large, with several prominent families of logics, and the complexity of description logics has been well studied. Research in ontology languages and related reasoning services, most notably in description logics, has also spurred work into logics that are weaker than classical systems, as well provided a substantial impetus for research into modal logic. Moreover, there has been substantial interaction with the database community.
The success of this work has led to an increasing demand for a variety of reasoning services, both classical and non-classical. Crucially, an ontology will be expected to evolve, either as domain information is corrected and refined, or in response to a change in the underlying domain. In a description logic, such change may come in two different forms: the background knowledge, traditionally stored in the TBox, may require modification, or the ground facts or data, traditionally stored in the ABox, may be modified. In the former case, the process is akin to theory revision, in that the underlying backround theory is subject to change. In the latter case, one cannot simply update instances, as is done in a relational database, since any set of instances must accord with the potentially rich structure imposed by the TBox. The result is that one must be able to deal with changing ontologies, as well as related notions from commonsense reasoning, including nonmonotonic reasoning and paraconsistent reasoning.
The issues mentioned are of common interest to the ontology, belief change, and database communities. While there has been some interaction between researchers in these communities, there has not been a comprehensive meeting to address notions of change in ontologies in a broad or comprehensive fashion.
The aim of the workshop was to bring together researchers working in the areas of logic-based ontologies, belief change, and database systems, along with researchers working in relevant areas in nonmonotonic reasoning, commonsense reasoning, and paraconsistent reasoning. Hence the workshop's goal was to facilitate discussions on the application of existing work in belief change, nonmonotonic reasoning, commonsense reasoning, and related areas on the one hand, to logic-based ontologies on the other. There has been extensive input and interest from the database community, which also has in interest in these problems. Overall the intent was to provide an interdisciplinary (with respect to computer science and mathematics) workshop for addressing both theoretical and computational issues in managing change in ontologies. In particular, the workshop has given participants a deeper understanding of the concepts, terminologies, and paradigms used in the three areas involved, and in their latest achievements and challenges. Examples of these were the distinction between data and schema level, the relation between different revision operators and justifications, the role of less expressive description logics, to name a few.
The workshop consisted of a five-day event with the following program: On the first day there were three introductory talks by a representative in each of the areas of belief change and nonmonotonic reasoning, description logics, and databases. The purpose of these introductory talks was to come to a shared understanding (and terminology) of these areas, and provide a glimpse of the state-of-the-art and current research challenges in all three areas. On day 2, three breakout groups were created and participants were assigned to them based on their expertise but also in such a way as to have representatives of the three main areas in each group. The groups were `Foundations and Techniques', `Applications', and `Perspectives and Future Directions', and their purpose was that of fostering discussions on the three fundamental components at the intersection of the above mentioned areas. Day 3 consisted of a report back from each of the groups followed by further discussion. On the fourth day there were presentations on overlapping areas and discussions of problems and issues of mutual interest for the different communities. Day 5 had a wrap-up session with a discussion on the overlap among the different areas, future challenges and next steps in this workshop series.
- Artificial Intelligence
- Data Bases / Information Retrieval
- Verification / Logic
- Artificial intelligence
- Belief change
- Description logics