CLIMA on Thursday

Detailed program
Thursday August 1st, 2002
See also the unified by-slot program


All sessions take place in auditorium 7.

Session 1: Agents: Arguments and Updates

09:00-09:30  Michael Schroeder, City U, London, UK
Ralf Schweimeier, City U, London, UK
Arguments and misunderstandings: Fuzzy unification for negotiating agents
In this paper, we develop the notion of fuzzy unification and incorporate it into a novel fuzzy argumentation framework for extended logic programming. We make the following contributions: The argumentation framework is defined by a declarative bottom-up fixpoint semantics and an equivalent goal-directed top-down proof-procedure for extended logic programming. Our framework allows one to represent positive and explicitly negative knowledge, as well as uncertainty. Both concepts are used in agent communication languages such as KQML and FIPA ACL. One source of uncertainty in open systems stems from mismatches in parameter and predicate names and missing parameters. To this end, we conservatively extend classical unification and develop fuzzy unification based on normalised edit distance over trees.

09:30-10:00  João Alexandre Leite, New U of Lisbon, Portugal
José Júlio Alferes, New U of Lisbon, Portugal
Luís Moniz Pereira, New U of Lisbon, Portugal
Halina Przymusinska, Cal Poly Pomona, USA
Teodor C. Przymusinski, U California-Riverside, USA
A language for multi-dimensional updates
Dynamic Logic Programming (DLP) was introduced to deal with knowledge about changing worlds, by assigning semantics to sequences of generalized logic programs, each of which represents a state of the world. These states permit the representation, not only of time, but also of specificity, strength of updating instance, hierarchical position of the knowledge source, etc. Subsequently, the Language of Updates LUPS was introduced to allow for the association, with each state, of a set of transition rules. It thereby provides for an interleaving sequence of states and transition rules within an integrated declarative framework. DLP (and LUPS), because defined only for a linear sequence of states, cannot deal simultaneously with more than a single dimension (e.g. time, hierarchies,...). To overcome this limitation, Multi-dimensional Dynamic Logic Programming (MDLP) was therefore introduced, so as to make it possible to organize states into arbitrary acyclic digraphs (DAGs). In this paper we now extend LUPS, setting forth a Language for Multi-dimensional Updates (MLUPS). MLUPS admits the specification of flexible evolutions of such DAG organized logic programs, by allowing not just the specification of the logic programs representing each state, but to the evolution of the DAG topology itself as well.

10:00-10:30  Antonis C. Kakas, U Cyprus
Pavlos Moraitis, U Cyprus
Argumentative agent deliberation, roles and context
This paper presents an argumentation based framework to support an agent's deliberation process for drawing conclusions under a given policy. The argumentative policy of the agent is able to take into account the roles agents can have within a context pertaining to an environment of interaction.

Session 2: Logics for Agents

11:00-11:30  Katsuhiko Toyama, Nagoya U, Japan
Takahiro Kojima, Nagoya U, Japan
Yasuyoshi Inagaki, Nagoya U, Japan
Translating multi-agent autoepistemic logic into logic program
Multi-agent autoepistemic Logic (MAEL) is a natural framework to formalize beliefs and reasoning including inheritance, persisitence, and causality. To develop a proof procedure of it, we introduce a method that translates a MAEL theory into a logic program with integrity constraints. We also investigate relations between MAEL and other nonmonotonic reasonings.

11:30-12:00  Pierangelo Dell'Acqua, Linköping U, Sweden
Ulf Nilsson, Linköping U, Sweden
Luís Moniz Pereira, New U of Lisbon, Portugal
A logic based asynchronous multi-agent system
We present a logic programming based asynchronous multi-agent system in which agents can: communicate with one another; update themselves and each other; eliminate contradictory update rules; abduce hypotheses to explain observations, and use them to generate actions. The knowledge base of the agents is comprised of generalized logic programs, integrity constraints, active rules, and of abducibles. We characterize the interaction among agents via an asynchronous transition rule system, and provide a stable models based semantics. An example is developed to illustrate how our approach functions.

12:00-12:30  James Harland, RMIT U, Australia
Michael Winikoff, RMIT U, Australia
Language design issues for agents based on linear logic (extended abstract)
Agent systems based on the Belief, Desire and Intention model of Rao and Georgeff have been used for a number of successful applications. However, it is often difficult to learn how to apply such systems, due to the complexity of both the semantics of the system and the computational model. In addition, there is a gap between the semantics and the concepts that are presented to the programmer. One way to bridge this gap is to re-cast the foundations of such systems into a logic programming framework. In particular, the integration of backward- and forward-chaining techniques for linear logic provides a natural starting point for this investigation. In this paper we discuss the language design issues for such a system, and particularly the way in which the potential choices for rule evaluation in a forward-chaining manner is crucial to the behaviour of the system.

Session 3: BDI Agent Systems

14:00-14:30  Rafael H. Bordini, Federal U of Rio Grande do Sul, Brazil
Álvaro F. Moreira, U Caxias do Sul, Brazil
Proving the asymmetry thesis principles for a BDI agent-oriented programming language
In this paper, we consider each of the nine principles of BDI logics as defined by Rao and Georgeff based on Bratman's asymmetry thesis, and we verify which ones are satisfied by Rao's AgentSpeak(L), a computable logic language inspired by the BDI architecture for cognitive agents. This is in line with Rao's original motivation for defining AgentSpeak(L): to bridge the gap between theory and practice of BDI agent systems. In order to set the grounds for the proof, we first introduce a particular way in which to define the informational, motivational, and deliberative modalities of BDI logics for AgentSpeak(L) agents, according to its structural operational semantics (that we introduced in a recent paper). This provides a framework that can be used to investigate further properties of AgentSpeak(L) agents, contributing towards giving firm theoretical grounds for BDI agent programming.

14:30-15:00  Tadashi Araragi, NTT Communication Science Laboratories, Japan
Shiro Takata, ATR Media Information Science Laboratories, Japan
Naoyuki Nide, Nara Women's U, Japan
A verification method for a commitment strategy of the BDI architecture
We present a method to solve a verification problem that arises in implementing a commitment strategy for the BDI architecture. This problem introduces a new aspect of verification such that a state transition depends on a verification done at each state. We formalize this problem and give a decision procedure for the verification.

15:00-15:30  Naoyuki Nide, Nara Women's U, Japan
Shiro Takata, ATR Media Information Science Laboratories, Japan
Tadashi Araragi, NTT Communication Science Laboratories, Japan
Deduction systems for BDI logics with mental state consistency
BDI Logics, introduced by Rao et al., have been used as the theoretical basis of specification and implementation of rational agents. The aim of our research is to make full use of the expressive power of BDI Logics as executable specification languages of rational agents. To this end, we previously presented deduction systems for CTL-based propositional BDI Logics using sequent calculus. Since these systems have a decision algorithm that is extended from Wang's algorithm, they are suitable for applications such as automatic proving. However, they do not incorporate mental state consistency features, which are important for dealing with rational agents. In this paper, we extend our deduction systems by introducing mental state consistency features and explain their soundness and completeness. This approach allows us to check and prove the specifications and properties described by BDI Logics for rational agents.

Session 4: Agents: Speculative Computation and Introspection

16:00-16:30  Hisashi Hayashi, Toshiba, Japan
Kenta Cho, Toshiba, Japan
Akihiko Ohsuga, Toshiba, Japan
Speculative computation and action execution in multi-agent systems
In some multi-agent systems, when an agent cannot retrieve information from another agent, the agent makes an assumption and tentatively performs the computation. When the agent comes across a mistake in the preliminary assumption, the computation is modified. This kind of speculative computation is effective when the assumption is correct. However, once the agent executes an action, it is impossible to modify the computation in these systems. This paper shows how to integrate speculative computation and action execution through logic programming.

16:30-17:00  Koji Iwanuma, Yamanashi U, Japan
Katsumi Inoue, Kobe U, Japan
Conditional answer computation in SOL as speculative computation in multi-agent environments
In this paper, we study speculative computation in a master-slave multi-agent system where reply messages sent from slave agents to a master are always tentative and may change from time to time. In this system, default values used in speculative computation are only partially determined in advance. Inoue et al. [2001] formalized speculative computation in such an environment with tentative replies, using the framework of a first-order consequence-finding procedure SOL with the well-known answer literal method. We shall further refine SOL calculus, using conditional answer computation and skip-preference in SOL. The conditional answer format has an great advantage of explicitly representing the dependency relation to tentative replies and defaults which are used to derive a conclusion. The dependency representation is significantly important to avoid unnecessary recomputation of tentative conclusions. The skip-preference has the great ability of preventing irrational/redundant derivations. Finally, we show an incremental answer computation method within the SOL tableau calculus.

17:00-17:30  Thomas Bolander, Techn. U of Denmark
Maximal introspection of agents
This paper concerns syntactical representations of introspective belief and knowledge in multi-agent systems. It is well-known that reasoning frameworks for introspective beliefs easily become inconsistent as a consequence of the presence of paradoxical self-reference. In the paper we explore the maximal sets of introspective beliefs that an agent can consistently obtain and retain. Hereby some previous results by Perlis [1985] and Rivière & Levesque [1988] are generalized.

Panel Discussion

Chair: Paolo Torroni

17:30-18:30  Logics and Multi-agents: towards a new symbolic model of cognition