Elsevier

Knowledge-Based Systems

Volume 13, Issue 5, October 2000, Pages 261-269
Knowledge-Based Systems

EULE: A Knowledge-Based System to Support Business Processes

https://doi.org/10.1016/S0950-7051(00)00086-1Get rights and content

Abstract

Office tasks related to the processing of contracts in the insurance business are complex and highly dependent on legal and company-specific regulations. Furthermore, due to increasing competition on the market there is a strong pressure to increase the efficiency and quality of office task performance. The only way to meet these manifold requirements is to provide a computer-based guidance and interactive support for office workers. At Swiss Life, we have developed the EULE system that fulfills these requirements. EULE's functionality is in the triangle of Knowledge Representation, Business Process Modeling, and Knowledge Management — the latter because EULE encodes and stores knowledge which is crucial for the company. The system relies on a knowledge representation language which covers data and process aspects as well as the relevant legislation and company regulations.

Introduction

Some time ago, Swiss Life, as many other companies, reorganized its customer support. Office workers are no longer specialists dealing with certain kinds of office tasks only, but are also be generalists who deal with all kinds of tasks. The number of different kinds of office tasks (like changing the beneficiary, increasing the risk sum, surrender of a contract) is rather high (about 60), and many of them only occur very sporadically. An office task has an attached process description that specifies how it should be performed. These processes are typically quite complex because there are many laws and company regulations which take influence on how they are to be performed properly. As a consequence, the work of this new generation of office workers is quite demanding and urgently calls for an appropriate support in order to meet Swiss Life's high quality standards.

The following observations can be made with respect to the characteristics of the office work:

  • The experience of the office workers ranges from novices who have just joined the company to experts with many years of training.

  • About 80% of the office work consists of standard cases with medium complexity. The remaining part consists of very complicated tasks that require a lot of experience.

  • There are about 60 different kinds of office tasks. Most of them occur very seldom so that an office worker does not have the opportunity to acquire certain routine.

  • It is the idea that new employees get a training first when they join the company. However, in order to have enough participants, the training courses actually take place only two or three times a year so that most of the training is nevertheless on the job and not before.

  • The way a certain office task is properly executed changes from time to time due to new products, new company regulations, or new legal restriction.

The situation characterized above shows that there is a potential for increasing the quality of office work and decreasing the average time needed per office task by providing the office workers with an appropriate support system for executing its process steps. Such a system must show the following functionality:

  • Support of office work. The system supports all kinds of office workers, from the novice to the expert. The novice gets his training while working with the system and can properly execute the processes attached to office tasks from the very beginning due to the active support and guidance of the system. The expert profits less from the system's guidance but from getting relieved from routine tasks, like writing letters and sending memos. He can request support from the system whenever the need arises (e.g. when dealing with an exotic case).

During the specification phase for such a system (which took many cycles), it became evident to us that additional requirements will have to be met by the system to become a success:
  • Just-in-time knowledge delivery. A user is always given exactly that kind of information he needs in a specific situation so that he never needs to ask for it (also called just-in-time knowledge delivery [4]).

  • Adaptation. Users with more experience need less guidance than novices. The system offers exactly that much guidance as needed by the user.

  • Maintainability. New legislation, and new company regulations (e.g. due to new products) make it necessary to adapt the office task descriptions regularly. As office task descriptions must be up-to-date to ensure proper office work it is of paramount importance that necessary changes can be done without much effort. Ideally, this is the case when the updates are made by the insurance experts themselves without involving people from the computing department (except maybe for special tasks like defining links to database fields).

Just-in-time knowledge delivery unifies business process support with the central knowledge management issue of supplying people with the knowledge they need to do their work properly. The adaptation functionality originates from the requirement to support novices as well as experts. Maintainability ensures that keeping the system alive will be feasible with respect to time and money.

Current Workflow Management systems which also aim at automating processes, fail in these concerns due to the limitations of their modeling languages. Neither do these languages allow the representation of detailed legal aspects in a declarative way, nor do they support any associated inferences. However, both would be needed to support an office worker in making the right decisions. Instead, workflow-modeling languages concentrate on the procedural flow between the different participants of a process. This is a macro level view of business processes while the requirements mentioned above require a support on a micro level.

Therefore, a system with a functionality that meets the above requirements was developed by the Information Systems Research Group of Swiss Life. It is called EULE and is situated in the triangle of Artificial Intelligence (AI), Knowledge Management (KM), and Business Process Modeling (BPM). It is an AI system because it is realized as a knowledge-based system. It contributes to the KM efforts of a company because its knowledge base encodes knowledge that is crucial for the company, and thus preserves it and helps to make it available where needed. EULE has an impact on BPM because it provides (formal) models of processes for performing office tasks in a much more detailed view than it is usually the case with the models resulting from a BPM approach. The level of detail of the EULE models is indeed needed for the active decision support the system must provide.

Attempts to bridge BP models and whole workflows are actually hard to find in practice. In so far our approach is also a way of operationalizing (very detailed) BP models so that they can be executed. The above aspects are discussed in detail below.

Section snippets

EULE: a knowledge-based, cooperative system for supporting office work

In order to build a tool like EULE which is able to support the direct execution of arbitrary office tasks, a model-based approach is most appropriate: For each office task we formulate a model of all the involved data objects, the process steps and the relevant regulation contexts. This model is then compiled to process descriptions which run on top of the shell-like EULE system components. In the following we summarize the main features of EULE both from the user and the modeler perspectives

The impact of EULE on knowledge management and business process modeling

The EULE knowledge base captures knowledge which is important to Swiss Life. Via EULE this knowledge is made available to office workers in such a way that they always get exactly that knowledge which is relevant in the current situation. Besides for supporting office work, EULE's knowledge is useful for other people and for other purposes as well, e.g. for tutoring new employees, for inquiring about the effects of certain company regulations on office tasks, or for finding out how often a

Experience and related work

There exist only a few other systems with a functionality roughly comparable to EULE. Like EULE, they aim at supporting business processes by offering the user (access to) that information which he or she needs to successfully proceed with the current office task (see [2], [12]). Unlike EULE, those systems are not oriented towards detailed and completely described office tasks but go for weakly structured workflows instead, as they are typical for tasks which comprise a great degree of

Conclusions

Office tasks related to processing of contracts in the insurance business are complex and highly dependent on the legal and company-specific formal regulations. On the other hand, the increasing competition strongly demands for higher efficiency as well as higher quality standards. The EULE system developed at Swiss Life is designed to meet these demands by providing computer-based guidance and interactive support for office tasks dealing with private life insurances. The system relies on a

Acknowledgements

We thank our colleagues Jörg Junger, Jörg-Uwe Kietz, Claude Marksteiner, Bernd Novotny and Thomas Vetterli, for their efforts to bring EULE into existence and to make it a success.

References (17)

  • A. Levy et al.

    Combining horn rules and description logics in CARIN

    Artificial Intelligence

    (1998)
  • E.W. Stein

    Organizational memory: review of concepts and recommendations for management

    International Journal of Information Management

    (1995)
  • A. Abecker et al.

    Toward a technology for organizational memories

    Intelligent Systems

    (1998)
  • A. Abecker, A. Bernardi, M. Sintek, Enterprise information infrastructures for active, context-sensitive knowledge...
  • S. Ceri et al.

    Logic Programming and Databases

    (1990)
  • K. Cole et al.

    Just-in-time knowledge delivery

    Communications of the ACM

    (1997)
  • J.-U. Kietz, M. Staudt, Access to heterogeneous data sources for supporting business process execution, Swiss Life,...
  • A. Margelisch, U. Reimer, M. Staudt, T. Vetterli, Cooperative support for office work in the insurance business,...
There are more references available in the full text version of this article.

Cited by (42)

  • E-government implementation: Business contract legal support for Taiwanese businessmen in Mainland China

    2005, Government Information Quarterly
    Citation Excerpt :

    On the other hand, articles and clauses of law represent specific forms, or rules, and can be codified as rule knowledge. Rule knowledge needs a well-constructed domain theory, or principle, as its knowledge basis for searches for domain-specific existing rules from the rule base.45,46 The prototype system is developed on Windows NT with Personal Web Server 4.0 and Microsoft FrontPage; the relational database is implemented in Access 2000; and browser uses Internet Explorer connected to the Web with ASP.

  • Designing process guidance systems

    2019, Journal of the Association for Information Systems
  • Process analytics: Concepts and techniques for querying and analyzing process data

    2016, Process Analytics: Concepts and Techniques for Querying and Analyzing Process Data
  • Big data based clouds healthcare web service forecasting: Research challenge

    2015, SACI 2015 - 10th Jubilee IEEE International Symposium on Applied Computational Intelligence and Informatics, Proceedings
  • Towards a customer-driven value chain framework – a set-based oriented approach

    2015, IFIP Advances in Information and Communication Technology
View all citing articles on Scopus
View full text