Applying electronic contracting to the aerospace aftercare domain

Abstract

The contract project was a European Commission project whose aim was to develop frameworks, components and tools to model, build, verify and monitor distributed electronic business systems based on electronic contracts. In this context, an electronic contract provides a specification of the expected behaviours of individual services, with the assumption that these services are often enacted by autonomous agents. Using the theoretical tools created by the project, in this paper we describe the complete life cycle of instantiating an electronic contracting system using the contract framework within the aerospace aftercare domain. Thus, we use a natural language description of parts of the types of contracts used in this domain to generate individual norms amenable to a computational representation, and how these norms are used to generate a concrete contract monitor. Moreover, we describe a concrete implementation of contract agents in the AgentSpeak(L) language and how these agents interact within a concrete instantiation of contract.

Introduction

As more and more business is conducted electronically over the internet, the need to guarantee transactions through formal electronic contracts is becoming increasingly important. In traditional business transactions, people need to meet and agree to terms of contracts before any exchange of goods and services takes place, but such a model is too slow, and inappropriate, in an interconnected world in which computer systems can replace human decision makers in closing deals and enacting decisions without constant supervision. In particular, systems of autonomous agents may be able to conduct business in an environment governed by formally documented norms, whereby different parties agree to the terms of business interactions specified in formal electronic contracts. These contracts must be represented in a way that allows autonomous agents to reason about them so that the agents can steer their behaviour towards compliance (Meneguzzi and Luck, 2009) and, more importantly, so that others can process such contracts and determine whether they have indeed been complied with.

To address this need, we have sought to create a framework, and guidelines, for the development of electronic contracting systems for real world applications. Like their physical counterparts, electronic contracts contain a set of clauses describing the expected behaviour of the signatories. In this view, clauses in such electronic contracts are comprised of individual norms (Oren et al., 2008) that describe expected agent behaviour in terms of the deontic modalities of obligations, prohibitions and permissions. For example, in the aerospace aftercare domain, aircraft engine manufacturers are contracted not only to sell new engines, but also to provide for their long term maintenance, subject to various terms by the airlines, and depending on terms from the suppliers of the component parts of these engines.

While we have previously outlined our basic computational framework for electronic contracting (Modgil et al., 2009), in this paper we review that framework and provide an instantiation of it for such an aerospace aftercare use case (Jakob et al., 2008), demonstrating the end-to-end application of the techniques developed, and how the generic framework can be used in a concrete real-world application of electronic contracting. The resulting application provides a representation of the required electronic contracts used in the multi-agent system that simulates the aftercare domain, as well as a monitoring process that is capable of generating explanations for contract violations (Modgil et al., 2009).

We start the paper by describing the domain in Section 2, including two specific scenarios in which electronic contracting is to be demonstrated. Next, we provide an overview of the contracting architecture in Section 3, explaining the components of an electronic contracting system following our framework. Once we have laid out the motivating aspects of our work, we proceed by describing the representation of the electronic contracts required for the aerospace domain in Section 4, showing how such contracts are represented originally in natural language, but eventually leading to the machine processable format needed in the system. After a contract has been created and signed by the relevant parties, we need to enact the contract, and monitor this enactment, as described in Section 5. The monitoring process detects when agents fail to comply with the norms they are committed to, but does not help to trace the original cause for a chain of violations, which may be critical in avoiding future failures. Hence, we describe a violation explanation mechanism in Section 6. We finish by showing the results of our practical implementation and empirical tests in Section 7.