Background: Abduction and induction are forms of reasoning under incomplete information that have many applications in AI. Abduction reasons from effects to possible causes and has been used to perform tasks such as planning and diagnosis. Induction learns general rules for particular observations and is typically used for classification and knowledge acquisition. The nature and relation of abduction and induction have been explored in a series of four preceding workshops at ECAI'96, IJCAI'97, ECAI'98, AIAI'05, and are further analysed in a recent book entitled Abduction and Induction: Essays on their Relation and Integration.

Motivation: As our understanding of abduction and induction grows and our computational techniques improve, it is becoming apparent that there are significant benefits to be gained by integrating both forms of reasoning, in a cooperative way, within an incremental cycle of knowledge development. By providing the means to extend prior (background) knowledge in the light of new experience (observations), such techniques could be useful in scientific and other forms of modelling. Indeed, some promising results are now beginning to emerge from the first tentative applications of these hybrid systems.

Focus: This Workshop will provide interested researchers with an opportunity to discuss state-of-the-art research into the integration of abduction and induction and to take stock of the progress that has recently been made. Particular emphasis will be placed on identifying practical techniques for integrating abduction and induction and on investigating the practical potential of such hybrid reasoning systems in real-world applications. Thus the primary aims of the Workshop are as follows:

  • To better understand the role of abduction and induction in the processes of theory formation and revision; and to explore intelligent methodologies for integrating them in an incremental cycle of knowledge development.

  • To identify different conceptual models for integrating abduction and induction; and to investigate how this integration can be achieved in a computationally viable way.

  • To determine the benefits that could result from the combination of abduction and induction and to characterise the classes of problems that can be usefully solved with such techniques.

  • To examine possible application areas and to assess in more detail the utility of such integrated frameworks in these cases.

Possible areas of application include, but are not limited to, modelling in systems biology, social science, natural language processing, and software engineering. The Workshop will also aim to examine the links to other approaches (philosophical or cognitive) for modelling scientific and other domains.


Session 1: Paper Presentations

Hasty generalisers and hybrid abducers: external semiotic anchors and multimodal representations [slides]
Lorenzo Magnani, University of Pavia, Italy
Abduction, preduction and the fallible way of modelling nature: some epistemological consequences for the philosophy of physics [slides]
Andrés Rivadulla, University of Madrid, Spain
Abstraction, induction and abduction in scientific modelling [slides]
Demetris Portides, University of Cyprus, Cyprus
Disjunctive bottom set and its computation [slides]
Wenjin Lu, University of Wales, UK
(Joint work with R. King)
Abduction, induction, and the logic of scientific knowledge development [slides]
Antonis Kakas, University of Cyprus, Cyprus
(Joint work with P. Flach and O. Ray)

Session 2: System Presentations, Invited Talk & Panel Discussion

INTHELEX system presentation: An abduction framework for handling incompleteness in first-order learning [slides]
Stefano Ferilli, University of Bari, Italy
(Joint work with F. Esposito, N. Di Mauro, T. Basile and M. Biba)
HAIL system presentation: Using abduction for induction of normal logic programs [slides]
Oliver Ray, Imperial College London, UK
Invited talk: Abduction, induction, and the robot scientist [slides]
Ross King, University of Wales, UK
Panel Discussion: Abduction, induction, and scientific modelling

The workshop proceedings may be downloaded here.

Peter Flach
Department of Computer Science, University of Bristol, UK
Antonis Kakas
Department of Computer Science, University of Cyprus, Cyprus
Lorenzo Magnani
Department of Philosophy, University of Pavia, Italy
Oliver Ray (primary contact)
Department of Computing, Imperial College London, UK
Peter Flach (University of Bristol, UK)
Katsumi Inoue (National Institute of Informatics, Japan)
Antonis Kakas (University of Cyprus, Cyprus)
Lorenzo Magnani (University of Pavia, Italy)
Stephen Muggleton (Imperial College London, UK)
Oliver Ray (Imperial College London, UK)
Alessandra Russo (Imperial College London, UK)
Chiaki Sakama (Wakayama University, Japan)