Future Scope – Communication, Coordination, and Collaboration with Intelligent Agents
By Michael Redford, SAP AG – 09/12/2003
Discovering new ways to think about and approach SAP's own product development, as well as stimulating discussion and the active exchange of innovative ideas. These are the goals of a new SAP-internal lecture series called "Future Scope." This event series is a mix of presentations, demonstrations, and workshops designed to give SAP employees the opportunity to experience the compelling technology and business innovations currently being developed by prestigious global research institutes – and discover what these new ideas could mean for SAP, its product portfolio, and software tools of tomorrow.
In the third lecture of the "Future Scope" series, Professor Hermann Krallmann, head of the Institute of System Analysis in Berlin, was guest speaker in SAP University's Audimax auditorium.
Figure 1: Prof. Krallmann
Imagine the following. A company has a market request, but is unable to process
it without an additional company's help, and so someone has to pick up the phone
to work out the details. Ideally, it shouldn't take a phone call, but the internal
business software of the two companies would work out the details automatically.
This is the idea behind "Intelligent Agents," presented by Professor
Hermann Krallmann on Monday, September 8, in the Audimax in St-Leon Rot.
Prof. Krallmann, head of the Institute of System Analysis in Berlin, has been
conducting research in the area of multiagent systems applied to manufacturing
for some time. Intelligent Agents is the subject of his current research, and
this Future Scope lecture. The goal is to decompose business processes into
smaller, atomic units – business process particles – and
then apply a knowledge-based approach to model business processes. Intelligent
Agents are the product of this model, and they have the potential to autonomously
and dynamically reconfigure business processes, leading to on-demand collaboration.
Prof. Krallmann has been a full professor at the Technical University of Berlin since 1979 and has fifteen years of research experience in the area of business technologies. At the Institute of System Analysis in Berlin, Prof. Krallmann, 55 staff members, and 1000 students collaborate with partners both in academia (for example, MIT, Stanford, the Institute of Technology in New Delhi) and in industry (for example, Sun, Ford, Intel). The current research is being carried out in cooperation with SAP; Prof. Krallmann already worked with SAP on several projects over the last few years, including helping to realize an E-learning project with SAP for Karstadt-Quelle, the German retail giant.
Figure 2: Prof. Krallmann on stage
The Intelligent Agents project consists of following steps:
- Create a framework: Analyze and decompose processes; determine common phases and common activities; classify these activities in object-oriented categories, stored in a process repository.
- Define the semantic systems behind the Intelligent Agents' actions: Knowledge representation; ontologies; rules or predicates; resolution calculus.
- Create a knowledge library: Map roles to customer requests, determine
dependencies and requirements, and set the triggers for specific processes.
This information is then stored in a knowledge library, shared by all parties.
After two months, Prof. Krallmann's research group was able to show the first
Intelligent Agents prototype at the lecture. The scenarios were typical
business processes: A customer wants to purchase a PC and desk and pick them
up immediately, or a customer wants to lease a PC and desk and pick them up
immediately. Members of the research team discussed the technical details of
how the Intelligent Agents used public knowledge to initiate the processes and
facts and rules to define the internal processes required to complete the test
scenarios. A workshop was held immediately after the event, to allow
employees to take a more detailed look at the technology.
In the subsequent question and answer session, members of the audience
asked penetrating questions about the originality of the model and pointed out
that the model largely mirrors the concept of inheritance, which is central
to object-oriented programming. In addition, the audience also questioned the
feasibility of the ontologies and decision calculus behind the semantic portion
of the model. These questions aside, the talk pointed out a central stumbling
block: because an infrastructure is missing, business processes are mostly peer-to-peer,
with little or no automatic coordination among partners. Just as XML and UDDI
have created generic formats for data exchange, the Intelligent Agents project,
because it attempts to define business process particles and a knowledge library,
will certainly further the goal of automatic business collaboration.
See also Future Scope – Tangible Information and Future Scope – A Networked World
References
