Call for Position Papers

Submission

Introducing agent-orientation to information systems (IS) is an area that is timely to address.  While the agent research area is very active, IS concerns are not yet well covered.  Agent concepts could fundamentally alter the nature of information systems of the future, and how we build them, much like structured analysis, ER modelling, and Object-Orientation has precipitated fundamental changes in IS practice.  However, at this point, what these changes might be, and how they would come to pass remain rather unclear.  We hope the workshop will provide a forum for presenting results in this area, and also acts as a catalyst to foster progress.

We encourage you to articulate your vision of the AOIS area, what are the most important aspects or issues for AOIS.  As suggested above, AOIS is not concerned with basic or generic agent concepts or technologies, but with specific orientation towards information systems and information systems development.

We outline some possible themes.  You are welcome to address these or expand on them, or identify additional themes we have missed.

1.  Is agent-orientation for IS primarily an implementation technology, or will it also be a shift in representational paradigm?

Information systems are used to store and manipulate information about the world. In what ways would agent-orientation extend the expressiveness of information systems? An AOIS can maintain higher-level representations of (natural or artificial) agents it has to deal with, in terms of their abilities, knowledge or beliefs, perceptions, commitments, expectations, trust relationships, etc. On the other hand (or at the same time!), an AOIS may itself be designed as an agent (or multiagent system) with its own intentional properties and states.

2.  Why are agent concepts relevant to IS?  Are they introduced to achieve new functionality, or to achieve non-functional qualities?

Much of AI agent research aims to achieve functionality not achievable before, i.e., various forms of "intelligence".  Mainstream IS, however, is dictated by many pragmatic concerns, such as reliability, performance, etc., and today, especially interoperability and high-level cooperation, legacy migration and evolution.  Agent-orientation would be a big boost for IS if it can help achieve these non-functional qualities, even if no new functionality is introduced.  We note that some areas of agent research are also directed at non-functional aspects, such as usability (user interface agents), believability (believable agents) and mobility.

3. What are the relevant properties of an agent from an IS perspective?

For example, in AI agent research, the sought-after properties are autonomy, situatedness, adaptivity, sociability, mobility, believability, etc. Are these the right properties for IS, and in what priority?  What are the overlaps with IS and software engineering concepts such as openness, platform-independence, location-independence, performance, robustness, evolvability, etc.?

4.  What are the implications of agent-orientation on the IS development process?

For IS, the development process and methodologies are crucial.  They must be industrial-strength production systems, not experimental systems, i.e., producible with predictable quality and schedule.  What development process will be appropriate?  Will it be substantially different from traditional IS or OO IS development?  Will agent-orientation facilitate reuse?  Can agent concepts be useful in the development process itself, and not just in the resulting product?  Many human agents are involved during the development process as well as in the usage environment.  Agent concepts have been used in requirements engineering, business process modelling, enterprise modelling, and user modelling.  Some agents can adapt, learn, evolve on their own, so the notion of development may become less clear-cut.

Could the agent concept become a new fundamental organizing concept for dealing with complex artificial phenomena, much like the concept of system in systems theory, entity in data modelling, and object in object-orientation? Would it replace or extend or augment these earlier paradigms?  Are there any conflicts among them?

5. What are the key components of an agent-oriented approach to IS?

Knowledge representation and reasoning? Communication and cooperation? Analysis and design methodologies? Should everything be treated as an agent or will there be other types of entities (such as objects and values) within the AOIS paradigm?

What are the most significant technical results from agent research (or other areas) that would make up the foundations of the AOIS area? What adaptations are needed? What are the key papers most relevant to AOIS?  Are there key papers for each of the technical areas listed in the Call for Papers? What are the technical issues and areas for AOIS?

6.  What transitional pathways can be envisaged for realizing an AOIS vision?

Would it be evolutionary or revolutionary?  Do we start from niche applications, then move incrementally into mainstream? What time line is realistic?  What level of maturity is needed, and where are we on the various fronts?  What are the technical obstacles that need to be overcome by AOIS research?  What are the non-technical obstacles?

How will AOIS co-exist with various IS developed under different paradigms? How do we agentify an existing IS?  Is object-orientation a necessary step towards AOIS?
 

Submission

Your position paper should not exceed 2 pages. It must either
  1. discuss a specific problem, or
  2. attack a specific position, or
  3. articulate a specific technology forecast.
You must indicate under which of these three categories your position paper falls. A problem discussion must begin with a section called Problem Statement and must conclude with a section called Research Questions. An attack must first describe the position to be attacked in neutral language before it presents reasons why it should be rejected. A technology forecast should consist of one or more forecast statements with additional explanation.

Position papers can be submitted until 1 May by email to  gw@inf.fu-berlin.de  in plain text or html.