This article reports the development and experimental evaluation of an Internet-enabled multi-agent prototype system, called AgentStra, for developing marketing strategies, competitive strategies and associated IT/IS/e-commerce strategies. Firstly, the multi-agent architecture of the AgentStra system is presented with relevant strategy agents described. Secondly, the logical flow and screen examples of the system execution are illustrated with guidelines on coupling AgentStra with human judgement proposed. Thirdly, the pilot evaluation of the system’s effectiveness and efficiency is documented with preliminary findings discussed. Finally, the conclusions are drawn with further research work envisaged.
In the past decades, research has been undertaken by many researchers to develop computerised decision support and intelligent systems to assist the process of strategic planning. Typical work in related fields includes: decision support systems (Belardo et al., 1994 and Moormann and Lochte-Holtgreven, 1993), expert systems (Carlsson et al., 1996, Houben et al., 1999, McDonald, 1989a and Moutinho et al., 1993), fuzzy logic (Levy & Yoon, 1995), artificial neural networks (Chien et al., 1999 and Dikmen and Birgonul, 2004), case-based reasoning (Changchien & Lin, 2005), intelligent agents (Orwig et al., 1997 and Pinson et al., 1997), and hybrid intelligent systems (Duan and Burrell, 1995, Li, 2000, Li, 2005, Li and Davies, 2001, Li et al., 2002, Li et al., 1999 and Li and Sugden, 2004).
While significant progress has been made to investigate the use of various decision support and intelligent techniques in the area of strategic planning, much less effort has been devoted to explore the applications of intelligent software agents in strategic decision-making. Pioneering work in creating intelligent agents for strategic planning may be found in Orwig et al., 1997 and Pinson et al., 1997. Pinson et al. (1997) reports a distributed multi-agent decision support system that incorporates the users as human agents in the solution formation process and enables strategic knowledge and domain knowledge to be distributed in different agents which communicate through various blackboards and message passing. A prototype has been created on a SUN SPARC II workstation using SMECI development environment based on an object-oriented formalism and on LISP (Pinson et al., 1997). Orwig et al. (1997) presents a multi-agent model of strategic planning using group support systems and artificial intelligence (AI). The study is mainly concerned with general strategic planning in the group support systems setting and exploring the application of AI-assisted categorization that helps reducing the cognitive loads placed on the facilitator and group participants.
The intention of this study is to explore how the process of marketing strategy development, competitive strategy-making and related IT/IS/e-commerce strategy formulation can be improved by an Internet-based multi-agent intelligent system. Thus, the aim of this paper is to present and discuss the system development work and some evaluation findings. The remainder of the paper is organised as follows. It begins with a description of the multi-agent architecture and related strategy agents of the system for recommending marketing strategies, competitive strategies and associated IT/IS/e-commerce strategies. The next section delivers a discussion of the logical flow and screen examples of the multi-agent system with guidelines on combining the multi-agent system with human judgement. There follows an account of the pilot evaluation process and a discussion of the findings. The final section presents the conclusions of the paper and outlines further work in this field.
This study has been sought to explore how the process of marketing strategy formulation, competitive strategy-making and associated IT/IS/e-commerce strategy development can be improved by an Internet/intranet-enabled multi-agent intelligent system. To achieve this aim, an Internet/intranet-based multi-agent prototype system, called AgentStra, has been created and described. The efficiency and effectiveness of the multi-agent system, coupled with human judgement, have been evaluated, with comparison to the use of paper-based models. Preliminary findings indicate that the AgentStra system is efficient and effective in terms of improving strategy-making speed, improving confidence about strategy development, coupling strategic analysis with human judgment and creativity, helping strategic thinking, improving the quality of strategic decision-making, etc.
Further work will be undertaken to: improve the user interface of the AgentStra system; include more strategic variables or factors that affect the strategy formulation; create more strategic analysis agents that represent more planning models; evaluate the new version of the AgentStra system with more participants including industrial users.