EX3: EXplain and EXploit Knowledge EXtracted to Improve ASP
EX3: EXplain and EXploit Knowledge EXtracted to Improve ASP
Disciplines
Computer Sciences (100%)
Keywords
-
Answer Set Programming,
Knowledge Representation and Reasoning,
Declarative Problem Solving,
Program Optimization,
Explainability,
Artificial Intelligence
Answer Set Programming (ASP) is a method that uses expert knowledge to calculate optimal decisions. Instead of explaining exactly how to solve a problem, experts simply describe it in a clear and structured way. ASP then uses this description to automatically find suitable solutions, taking many influencing factors into account. This can be used, for example, for finding hospital shift schedules or for product configurations. One major advantage of ASP is that experts dont need to be programming specialists. They can describe problems at a high level, and a specialized software takes care of the complex search for answers. Often, this works quickly and effectively. But sometimes, depending on how the problem is described, the search can become slow or get stuck. Fixing such issues often requires deep technical knowledge and a lot of time. Our project aims to make ASP even more helpful by developing a system on top of it that not only finds solutions but also supports domain experts in better understanding the problem itself. To achieve this, we will build methods that automatically detect important patterns or structures within the described problem. These patterns can help speed up the solving process and identify useful assumptions that make problem-solving more efficient. At the same time, the system will provide clear, visual explanations of what it has learned. Experts can use these insights to check whether their problem description is complete or to improve their understanding of the problem. They will also be able to interact with the system to guide it and refine the input as needed. The result of our project will be a new system called EX3, named after its three main goals, i.e., to Extract, Exploit, and Explain knowledge.
- Universität Klagenfurt - 100%
- Bart Bogaerts, Katholieke Universiteit Leuven - Belgium
- Torsten Schaub, Universität Potsdam - Germany
- Mario Alviano, Università della Calabria - Italy
- Marco Maratea, Università di Calabria - Italy
- Mark Law, Imperial College London