A multi-agent programming language for simulating work practice

A human work system is an example of a complex system of collaboration and coordination between intelligent agents, namely human beings. Our thesis is that if we can develop an agent-oriented language with which we can describe the collaboration between human agents, we will have created a multi-agent programming language that deals with agent interactions, such as the coordination, collaboration, and mobility of intelligent agents, including software agents. Brahms is a multi-agent programming language for modeling and simulating human collaboration in a work system. The Brahms language has its roots in other agent-based languages, such as AGENT-0 (Torrance 1991), and PLACA (Thomas 1993). The Brahms language is based on the formal logic of computational multi-agent systems, as described by Wooldridge (Wooldridge 1992). However, as Wooldridge described, his theory was not intended as a model of human social systems. Brahms incorporates a theory of human social systems. Our theory focuses on meso human social systems — “as a mid-level theory that links a micro-level mechanisms to macro-level phenomena, in our case the physical and social to the cognitive” (Carley and Prietula 1994)— meaning that we try to describe a specific type of human system, namely that of a human activity system (Checkland and Scholes 1990). As such we needed to extend Wooldridge’ formal logic with provisions for modeling human-actors (social agents); including their activities, collaboration, their environment, and the fact that they are situated in the real world, acting and observing, reacting to and interacting with other agents, objects, and artifacts. In this paper, we will describe the Brahms language features using as an example a simulation of two human agents collaborating in an activity of correcting errors on work orders. The model shows the implicit coordination of the work activities of these two agents through the faxing of orders and a telephone conversation to resolve the error on the order.

[1]  W. Clancey The conceptual nature of knowledge, situations, and activity , 1997 .

[2]  Patricia Sachs,et al.  Transforming work: collaboration, learning, and design , 1995, CACM.

[3]  Etienne Wenger,et al.  Communities of Practice: Learning, Meaning, and Identity , 1998 .

[4]  Drew McDermott,et al.  Introduction to artificial intelligence , 1986, Addison-Wesley series in computer science.

[5]  Peter Checkland,et al.  Soft Systems Methodology in Action , 1990 .

[6]  Maarten Sierhuis,et al.  Brahms: simulating practice for work systems design , 1998, Int. J. Hum. Comput. Stud..

[7]  Guus Schreiber,et al.  KADS : a principled approach to knowledge-based system development , 1993 .

[8]  Michael J. Prietula,et al.  ACTS theory: extending the model of bounded rationality , 1994 .

[9]  R. A. Brooks,et al.  Intelligence without Representation , 1991, Artif. Intell..

[10]  William J. Clancey,et al.  Knowledge-based tutoring: the GUIDON program , 1987 .

[11]  Michael Wooldridge,et al.  The logical modelling of computational multi-agent systems , 1992 .

[12]  Andrzej Nowak,et al.  Social dilemmas exist in space , 1984 .

[13]  A. Newell Unified Theories of Cognition , 1990 .

[14]  Allen Newell,et al.  SOAR: An Architecture for General Intelligence , 1987, Artif. Intell..

[15]  Sarah Rebecca Thomas,et al.  PLACA, an agent oriented programming language , 1993 .

[16]  Allen Newell,et al.  Human Problem Solving. , 1973 .

[17]  Nicholas R. Jennings,et al.  Intelligent agents: theory and practice , 1995, The Knowledge Engineering Review.