Call for Papers

Abstraction is a process that is exploited in human reasoning and understanding. Although the word itself comes from the meaning of "to draw away", there is no precise definition that is able to cover all the meanings that it gains depending on its utilisation. Various meanings of abstraction are interpreted in different disciplines such as Philosophy, Cognitive Science, Biomimetics, Mathematics and AI, with the shared consensus of the aim to distil the essential.

From the early days of AI research, including in the work of Alan Turing, such abstraction learning has been seen as a crucial heuristic for problem-solving. First, the problem is solved in a relaxed or reduced space, and then the abstract solution is used to guide the search for a solution in the original space. Since the success in solving a problem relies on how "good" the abstraction is, theoretical approaches for defining abstractions with desired properties have been and continue to be investigated while adhering to certain principles of simplification and/or generalization. Abstraction is also being used as a representation technique. Having different layers of representation that enable reasoning at a high level and refining to more low-level details only when necessary, e.g., in Robotics, allows one to determine the focus points of the problem. While usually the representation decisions are left to the experts, there are also methods, e.g. in Model Checking, to automatically find abstractions that allow one to check desired properties of the system at the abstract level. More recently, abstraction is becoming an essential technique for AI systems to present a “model of self”, overviewing their complex structures via showing the key elements making it easier for humans to understand their decision-making.

This workshop aims to bring together researchers from different sub-areas of KR and related communities who work on different aspects of abstraction in their respective areas, with the goal of exchanging theories and methods.
The following lists topics (but is not limited to these):

Submission

We invite short (6 pages) and long papers (13 pages) of unpublished work, or extended abstracts (2 pages) of already published works or works in progress. Reviewing will be single-blind, but anonymous submissions are possible.

Authors of all accepted original contributions can opt to publish their work on CEUR proceedings. Accepted non-original contributions will be given visibility on the workshop website, including a link to the original publication, if already published.
Following the workshop, there will be an open call for inclusion in a special issue of the German Journal of Artificial Intelligence (KI).

Contact

In case of questions, please write one of the organizers listed at the organization tab.