Year: 2019
Type: conference-paper
Abstract: Business processes realize a business goal by coordinating the tasks undertaken by multiple interacting parties. Even if it is possible to monitor the execution of such complex distributed process, current approaches do not allow participants to report to the right account taker the causes of the success or failure of their duties. However, as in exception management in programming languages, having such information could enable the account taker to properly handle errors. We claim that an explicit representation of accountability and responsibility assumptions provides the right abstractions to engineer multi-agent systems, that execute such business processes, both at the level of design and at the level of programming. Basing our programming approach on multi-agent organizations, we present two accountability patterns for developing accountable agents. To illustrate this approach we use the JaCaMo multi-agent programming platform.
Language: English
Subfield: Artificial Intelligence
Field: Computer Science
Domain: Physical Sciences
Open Access status: green
OpenAlex ID: W2941860672
