Declarative Agent Languages and Technologies IX

The context of the talk is an interest and a need to reason about issues related to cooperation in multi-agent systems, where, given the notion of a coalition (that is, a group of agents), questions arise regarding the coalitional power (what can the coalition achieve?), coalition formation (which coalitions will form?) and the result of cooperation (how will the coalition act?). Coalition Logics provide a tool to analyse some of those questions. They took off with two important developments, namely with Pauly’s formulation of Coalition Logic CL [2], and the work on Alternating-time Temporal Logic (ATL) by Alur, Henzinger and Kupferman [1].

We consider the issue of indexing plans (or rules) in the implementation of BDI languages. In particular we look at the issue of plans which are not triggered by the occurence of specific events. The selection of a plan from such a set represents one of the major bottle-necks in the execution of BDI programs. This bottle-neck is particularly obvious when attempting to use program model checkers to reason about such languages.

This paper describes the problem and examines one possible indexing scheme. It evaluates the scheme experimentally and concludes that it is only of benefit in fairly specific circumstances. It then discusses ways the indexing mechanism could be improved to provide wider benefits.

In open multi-agent systems, electronic institutions are used to form the interaction environment by defining social norms for group behaviour. However, as this paper shows, electronic institutions can be turned against agents to breach their security in a variety of ways. We focus our attention on probing attacks using electronic institutions specified in the Lightweight Coordination Calculus (LCC) language. LCC is a choreography language used to define electronic institutions in agent systems. A probing attack is an attack against the confidentiality of information systems. In this paper, we redefine the probing attack in conventional network security to be applicable in a multi-agent system domain, governed by electronic institutions. We introduce different probing attacks against LCC interaction models and suggest a secrecy analysis framework for these interactions. The proposed framework could be used to detect the possibility of certain probing attacks and to identify some forms of malicious electronic institutions.

Commitments are being used widely to specify interaction among autonomous agents in multiagent systems. While various formalizations for a commitment and its life cycle exist, there has been little work that studies commitments in relation to each other. However, in many situations, the content and state of one commitment may render another commitment useless or even worse create conflicts. This paper studies commitments in relation to each other. Following and extending an earlier formalization by Chesani et al., we identify key conflict relations among commitments. The conflict detection can be used to detect violation of commitments before the actual violation occurs during agent interaction (run-time) and this knowledge can be used to guide an agent to avoid the violation. It can also be used during creation of multiagent contracts to identify conflicts in the contracts (compile-time). We implement our method in \(\mathcal{REC}\) and present a case study to demonstrate the benefit of our method.

This paper investigates the use of high-level action languages for representing and reasoning about commitments in multi-agent domains. We introduce the language \({\cal L}^{mt}\), an extension of the language \(\cal L\), with new features motivated by the problem of representing and reasoning about commitments. The paper demonstrates how features and properties of commitments can be described in this action language. We show how \({\cal L}^{mt}\) can handle both simple commitment actions as well as complex commitment protocols. Furthermore, the semantics of \({\cal L}^{mt}\) provides a uniform solution to different problems in reasoning about commitments such as the problem of (i) verifying whether an agent fails (or succeeds) to deliver on its commitments; (ii) identifying pending commitments; and (iii) suggesting ways to satisfy pending commitments.

The Lightweight Coordination Calculus was presented in a paper to DALT 2004 as a method for specifying a class of social norms for multi-agent systems. This was intended for use in the engineering of a range of applications but at the time the original paper was written this was an aspiration and we had little experience of actual use of the method. In this paper I summarise how experience with this approach has developed in the seven years from 2004 to date.

This note is a retrospective review of our 2006 paper [1] on the properties of protocols, especially interoperability.

A bit of history is in order. By 2006, the importance of a social semantics for protocols was well-established in the multiagent systems community. Further, commitments had emerged as a preeminent abstraction for capturing the semantics. The big advantage was that specifying the meaning of protocol messages in terms of the commitments among agents enabled the agents to act flexibly.

The idea of extending the BDI architecture with cooperativity started shaping in 2003 when two independent proposals to support cooperation in a BDI setting were presented at DALT. One proposal, Coo-BDI, extended the BDI architecture by allowing agents to cooperate by exchanging and sharing plans in a quite flexible way; the other extended the BDI operational semantics for introducing speech-act based communication, including primitives for plan exchange. Besides allowing a natural and seamless integration with speech-act based communication for BDI languages, the intuitions behind Coo-BDI have proved to be promising and attractive enough to give rise to new investigations. In this retrospective review we discuss papers that were influenced by Coo-BDI and we outline other potential developments for future research.

We provide a retrospective on the research leading to and following our paper “A Distributed Architecture for Norm-Aware Agent Societies” [1], presented at DALT 2005. We do so by giving the context and motivation for that research, listing its contributions, and discussing the main developments of the research and its impact.

In this paper we revisit the motivations and the initial developments that led to our DALT 2003 paper Extending the Operational Semantics of a BDI Agent-Oriented Programming Language for Introducing Speech-Act Based Communication. We then discuss our own follow-up work which consisted in formally defining a larger set of speech-act based performatives and deploying them in Jason, a fully-fledged implementation of AgentSpeak. Subsequent research referring to the computationally grounded semantics of speech-act based agent communication that we introduced in that paper is also discussed.

In this short contribution we explain how our research has evolved from the publication of the following paper [2] with respect to its relevant aspects. This paper proposes a model of norms whose content is related to time, which are specified at design time and therefore are expressed in terms of roles played by the agents. Those norms have an activation event that is used to express the template of the events that when happen, provided that certain conditions hold, transform the norm in a social commitment. This dynamic evolution of norms is formalized by means of ECA-rules. Another relevant contribution is that the model of norms presented in this paper makes it possible to specify two types of sanctions for norms enforcement: active sanctions and passive sanctions.

Social commitments in time: Satisfied or compensated was the title of a presentation given at the 7th DALT workshop edition [34] in which we proposed a layered architecture for modeling and reasoning about social commitments. We gave emphasis to modularity and to the need of accommodating certain temporal aspects in order for a commitment modeling framework to be flexible enough to adapt to diverse commitment theories, and expressive enough to model realistic scenarios. We grounded the framework on two formalisms: the Reactive Event Calculus (\(\mathcal{REC}\)) and the Commitment Modeling Language (\(\mathcal{CML}\)). In this retrospective, we review recent developments of this line of work, and discuss our contribution in a broader context of related research.

The benefits of Service Oriented Architectures (SOA) for business, such as reduced costs and development time, are well recognised, however one of the most challenging steps in using SOA is defining the correct composition of services for a particular business process. Quickly recognised as a task where computer automation could help, various approaches have been proposed, including the use of AI techniques for planning service compositions. However, these techniques can perform poorly due to the search space explosion caused by dealing with the vast number of available services that must be composed. In this paper we present an approach to composing Web services, using software agents to enact plans of actions which achieve organisational goals, where each action specifies what should be achieved as opposed to which service to use. When enacting an action, agents use a matchmaking process to determine services that can be used to achieve the desired effects, intelligently handling any errors that may occur. The action plans are based on an organisation model in which organisational goals are refined into scenes, landmarks, and objectives, allowing the set of actions available to the plan synthesis mechanism to be tailored to the goal being targeted at that specific time, further reducing the planning search space.

In this extended abstract, I shall briefly describe the course I gave at the DALT International Spring School that took place at the University Residential Centre in Bertinoro, Italy, in April 2011. Before I do so, I feel compelled to say, although this was supposed to be a technical paper, that participating in that School was the most fabulous experience of my academic career. I attended many conferences and summer schools over the last decade, and in particular all the summer schools were incredible experiences, but none matched that of the DALT School in Bertinoro. The University Centre is located in an astounding medieval castle, where even Dante stayed for some time. The castle is at the top of a hill and the views from the castle where we were also accommodated are just breathtaking. The food was excellent and the people involved at all levels incredibly friendly; the organisation was impeccable. Of course just atmosphere does not make an academic event that memorable. Perhaps because this was the most specific summer school I ever attend in regards to the topics covered, which allowed the courses to go into much more depth than usual, but certainly not only because of that but also other factors such as the particular combination of people in that School, that was definitely the most technically productive school I ever attended. All the attendants cannot possibly thank enough the organisers for the amazing experience they created for us. Unfortunately, nothing is perfect. As memorable as the school was, I will never be able to forget the suffering it was to walk up those extremely steep hills either.