Workshop on New Trends in Belief Change
The workshop will take place on May 10, 2016.
ORGANIZATION
Stefan Woltran
Jean-Guy Mailly
Adrian Haret
AIMS AND SCOPE
Belief change is a central field in the area of Artificial Intelligence, at the intersection of Computer Science, Logics and Philosophy. Belief change considers formal models of how the beliefs of an agent (or a group of agents) evolve in different situations. Over the latest decades, a huge body of results on belief change has emerged. In particular, belief change operations such as revision, contraction, update and merging have been characterized through axiomatic approaches (rationality postulates) and constructive approaches (representation theorems). In recent years novel challenges within belief change have arisen, such as the study of the impact of fragments of classical logic on belief change operators, or the problem of adapting belief change operators to other settings.
The aim of this workshop is to bring together researchers from different fields in order to discuss recent results and ongoing work on these new challenges.
Topics of interest include:
- Belief Change in Fragments of Classical logic
- Belief Change in Nonclassical logics
- Belief Change and Computational Social Choice
- Belief Change and Stream Reasoning
- Belief Change and Ontologies
PROGRAM AND SCHEDULE
Tuesday, May 10
ABSTRACTS
AGM Meets Abstract Argumentation: Expansion and Revision for Dung Frameworks
Ringo Baumann, Universität Leipzig, Germany
This talk is about AGM-style expansion and revision operators for Dung's abstract argumentation frameworks (AFs). Our approach is based on a reformulation of the original AGM postulates for revision in terms of monotonic consequence relations for AFs. The latter are defined via a new family of logics, called Dung logics, which satisfy the important property that ordinary equivalence in these logics coincides with strong equivalence for the respective argumentation semantics. Based on these logics we define expansion as usual via intersection of models. We show the existence of such operators. This is far from trivial and requires to study realizability in the context of Dung logics. We then study revision operators. We show why standard approaches based on a distance measure on models do not work for AFs and present an operator satisfying all postulates for a specific Dung logic.
Belief Contraction Within Fragments of Propositional Logic
Nadia Creignou, Aix-Marseille Université, France
Recently, belief change within the framework of fragments of propositional logic has gained attention. In the context of revision it has been proposed to refine existing operators so that they operate within propositional fragments, and that the result of revision remains in the fragment under consideration. In this paper we generalise this notion of refinement to belief change operators. Whereas the notion of refinement allowed us to define concrete rational operators adapted to propositional fragments in the context of revision and update, it has to be specified for contraction. We propose a specific notion of refinement for contraction operators, called reasonable refinement. This allows us to provide refined contraction operators that satisfy the basic postulates for contraction. We study the logical properties of reasonable refinement of two well-known model-based contraction operators. Our approach is not limited to the Horn fragment but applicable to many fragments of propositional logic, like Horn, Krom and affine fragments.
Generalised Belief Revision
James Delgrande, Simon Fraser University, Canada
An agent will need to change its beliefs in response to new information. In AI, the standard approach to belief change, the AGM approach, assumes that the agent's underlying logic contains classical propositional logic. As a result, the AGM approach is not immediately applicable to approaches that lack the full expressive power of propositional logic, including approaches based on definite or Horn clauses, answer set programming, some description logics and others. In this talk I reconsider the AGM approach, and show that it may be formulated with respect to basically any logic. In this, one needs assume little more than that we have a language with sentences that are true as models, or possible worlds. The classical AGM postulates are expressed in this framework and representation results are established characterising the postulates in terms of preorders on possible worlds. Several examples will be given to illustrate the approach, including Horn clause revision and belief revision in an extremely basic logic called literal revision. These results shed light on the theoretical underpinnings of belief change, since they show that the dominant approach to revision is applicable even in very weak logics, and consequently is applicable in a much broader class of systems than previously believed. As well, given the correspondence between belief revision and nonmonotonic consequence operators identified by Gärdenfors and Makinson, this work may also shed light on this latter area.
Belief Change, Belief Merging and Judgment Aggregation: The Significance of
Social Choice Theory
Daniel Eckert, University of Graz, Austria
Starting from the trivial observation that any rule for aggregating information establishes a dependency between the change of the collective acceptance and the change of the individual support for a given proposition, we survey how different approaches to the aggregation of information like (distance-based) belief merging and (boolean-valued) judgment aggregation capture and exploit this dependency, the focus on which is strongly shaped by social choice theory.
Some Thoughts on Stream Reasoning, Revision and Update
Thomas Eiter, TU Wien, Austria
In the recent years, stream reasoning has gained increasing attention. This talk aims at elucidating the notion and to point out some issues that may link the topic, where change of information is intrinsic, to theory change, both semantically and computationally.
Logical Consequence and Belief Change Arising From Semi-Abstract Argumentation
Chris Fermüller, TU Wien, Austria
Dung's seminal paper introducing abstract argumentation frames (1995) lead to an impressive amount of research devoted to a host of different semantics, corresponding algorithms, instantiation mechanisms, classifications of argument forms, etc, but also to various applications in AI and beyond. Nevertheless, we claim that there is still room for alternative foundations inspired by Dung's basic concept. In this spirit we propose to extract (new and old) consequence relations from given semi-abstract argumentation frames. Semi-abstract argumentation arises from abstract argumentation by instantiating the claims of arguments with (potentially logically complex) propositions, while ignoring the structure of their support. In this manner one may define logical consequence as arising from first principles like "every argument attacking A also attacks the conjunction of A and B" and "every argument attacking a disjunctive claim also attacks each of its disjuncts". This amounts to an alternative semantic framework, closer to, e.g., Brandom's incompatibility semantics than to the traditional Tarskian notion of truth in a model. We will show that classical logic emerges only if principles are imposed that, arguably, are too strong with respect to intended interpretations. Nonclassical consequence relations arise from principles that correspond to some, but not all logical introduction rules of Gentzen's sequent calculus LK. This observation allows one to relate corresponding "argumentative consequence relations" to nondeterministic matrix semantics as studied by Avron, Lev, and Lahav. In the final part of this contribution we will indicate potential consequences of the outlined research endeavor for belief change. It seems natural to consider belief change as arising from the addition or retraction of arguments. We will indicate how our "logic of arguments" relates to approaches in belief revision and update.
Beyond IC Postulates: Classification Criteria for Merging Operators
Adrian Haret, TU Wien, Austria
Merging is one of the central operations in the field of belief change, not least due to its relationship to concepts like judgment aggregation and voting. Representation theorems characterize the family of merging operators satisfying some natural desiderata (IC postulates), by representing the problem of merging beliefs as an aggregation problem on rankings of models, or possible worlds. From this latter point of view, several concrete merging operators have been put forth. However, little is known about how these operators can be further distinguished. In the field of social choice, on the other hand, numerous properties have been proposed in order to classify voting rules. In this work, we adapt these properties to the context of merging and investigate how they relate to the standard postulates. Our results thus lead to a more fine-grained classification of merging operators and shed light on the question of which particular merging operator is best suited in a concrete application domain.
Extension-Based Belief Change in Abstract Argumentation
Thomas Linsbichler, Jean Guy Mailly, TU Wien, Austria
Dynamic aspects of argumentation frameworks (AFs) have received increasing attention in recent years. However, the first approaches to tackle this question did not benefit from the well-known AGM theory for belief change in logical settings. In this talk, we discuss our adaptation of the AGM theory to abstract argumentation. More specifically, we have adapted the KM belief revision (model-based counterpart of AGM revision) to define an extension-based revision approach for AFs. Since our revision approach led to a set of AFs as the result in general, we have used recent insights on the expressiveness of AF-semantics for a refinement of this approach, guaranteeing the result to be a single AF. Then, we have been interested in the merging of AFs, i.e. we have proposed an approach to obtain a global representation of a group of agents, each of them using her own AF.
Managing Change in Graph-Structured Data Using Description Logics
Magdalena Ortiz, TU Wien, Austria
We will discuss the setting of graph-structured data that evolves as a result of operations carried out by users or applications. We will discuss different reasoning problems, which range from ensuring the satisfaction of a given set of integrity constraints after a given sequence of updates, to deciding the (non-)existence of a sequence of actions that would take the data to an (un)desirable state, starting either from a specific data instance or from an incomplete description of it. We consider an action language in which actions are finite sequences of conditional insertions and deletions of nodes and labels, and use Description Logics for describing integrity constraints and (partial) states of the data.
Kinetic Consistency, Relevance, and Iterated Revision
Pavlos Peppas, University of Patras, Greece
This talk is divided into two parts. In the first part we examine a critical aspect of belief revision that has been left unattended in the classical AGM framework. We call it the principle of kinetic consistency. In contrast to the AGM postulates, this principle introduces relationships between revision policies assigned to different belief sets. We formalise kinetic consistency axiomatically and constructively, and we present a representation result explicitly connecting the two. Moreover, we show that our postulates are consistent with the AGM postulates for revision, as well as Parikh's postulate for relevant change. The second part of the talk discusses a fundamental relationship that we claim that exists between kinetic consistence, relevance, and iterated revision.
Towards Updating Wikipedia via DBpedia Mappings and SPARQL
Vadim Savenkov, WU Wien, Austria
DBpedia is one of the best known Linked Open Data sources, extracted from the Wikipedia articles with the help of fairly simple declarative mappings. In fact, the development and maintenance of DBpedia mappings themselves is an open community effort, very much like Wikipedia itself. Despite its immense importance for the Semantic Web community, the impact of DBpedia on the original Wikipedia data remains quite limited. In particular, little attention has been paid to potential benefits of the semantic infrastructure for maintaining the wiki content, for instance in order to ensure that the effects of a wiki edit are consistent across different pages. In this talk, we consider DBpedia mapping mechanism as an ontology based data management setting (OBDM), and discuss the specifics and possibilities of OBDM implemented on top of a curated data source like Wikipedia.
New Trends in Belief Change. Challenges and Applications
Stefan Woltran, TU Wien, Austria
Understanding the behavior of belief change operators for fragments of classical logic has received increasing interest over the last years. Results in this direction are mainly concerned with adapting representation theorems. However, fragment-driven belief change also leads to novel research questions. One example is the concept of belief distribution, which can be understood as the reverse task of merging and ask whether arbitrary knowledge can be represented by a set of knowledge bases of simpler structure, such that the task of merging allows for a reconstruction of the original knowledge. In this talk, we report on some results for distribution and, in addition, propose further novel reasoning problems which arise in the context of fragment-driven belief change.
PARTICIPANTS (Preliminary List)
Michael Abseher, TU Wien
Ringo Baumann, Universität Leipzig
Harald Beck, TU Wien
Nadia Creignou, Aix-Marseille University
James Delgrande, Simon Fraser University
Martin Diller, TU Wien
Daniel Eckert, Univ. Graz
Thomas Eiter, TU Wien
Chris Fermueller, TU Wien
Adrian Haret, TU Wien
Markus Hecher, TU Wien
Thomas Linsbichler, TU Wien
Jean-Guy Mailly, TU Wien
Jan Maly, Universität Wien
Magdalena Ortiz, TU Wien
Pavlos Peppas, University of Patras
Andreas Pfandler, TU Wien
Andreas Pieris, TU Wien
Vadim Savenkov, WU Wien
Mantas Šimkus, TU Wien
Sebastian Skritek, TU Wien
Christof Spanring, TU Wien
Stefan Woltran, TU Wien
VENUE
Seminarraum 351
Elektrot. Institutsg., Gußhausstraße 27-29
Neues EI, Stiege II, 1. Stock
Raumnummer: CA0138
GETTING THERE
The room where the workshop will take place is in the Neues Elektrotechnisches Institutsgebäude (Neues EI), and is accessible via the U1 metro line (in between the Karlsplatz and Taubstummengasse metro stations), or the Paulanergasse trams station (Trams 1 and 62).
TU Wien is distributed among several buildings around the area of Karlsplatz, and is serviced by a reliable and fast transportation system:
- Karlsplatz metro station (along the U1/U2/U4 metro lines), for TU Wien main building
- Resselgasse tram station (along Trams 1 and 62), for TU Wien main building
- Taubstummengasse metro station (along the U1 metro line), for Favoritenstrasse 9-11, where DBAI is located
- Paulanergasse tram station (along Trams 1 and 62), for Favoritenstrasse 9-11, where DBAI is located
TRAVEL INFORMATION
There are several options of hotels around the workshop venue.
Hotel Astoria
Clima Cityhotel
Carlton Opera Hotel
Hotel Erzherzog Rainer
Hotel Johann Strauss
PHOTOS
SPONSORS
The workshop is supported by the VCLA and the Wolfgang Pauli Institute.
