Representing and processing fuzzy information - an XML-based approach

Fuzzy approaches to improve decision making within business applications are widely spread. However, these solutions are often restricted to special areas, and are less than fully integrated into company-wide business application systems. We propose to use a common way to represent fuzzy information to soften this lack of integration. In this paper, we introduce a formal syntax for important fuzzy data types used to store fuzzy information. We make this syntax operatable by de®ning appropriate document type de®nitions (DTD), and we show how fuzzy information, whose description is based on these DTDs, can be exchanged between application systems by means of the extensible markup language (XML). With this, we give an approach to a better integration of business applications using fuzzy approaches into company-wide business application systems. This allows for a better collaboration of application systems and development tools, which use fuzzy approaches or which are intended to develop fuzzy approaches, in general. q 2002 Elsevier Science B.V. All rights reserved. 1. XML as cross-domain technique to fuse fuzzy-approaches Even though a lot of business applications, which use fuzzy approaches, were proposed and successfully applied, cf. e.g. [1], there are still obstacles that make it dif®cult to use (computer aided) fuzzy approaches for business applications and as parts of business application systems. Major problems arises with integration of fuzzy applications [2]. Integration problems concern combining fuzzy applications with each other, and integrating fuzzy applications into conventional business application systems. The problem of technically integrating fuzzy approaches into existing business application systems was already addressed in Ref. [3]. There a ¯exible Java-based fuzzy kernel was presented, where business applications could be built on. Taking this approach as a basis, we enhanced the idea of the Java-based fuzzy kernel by subdividing the kernel into smaller parts, so-called software components [4]. By doing so, we focussed on the problem of developing business applications using fuzzy techniques, as the use of software components simpli®es the task of programming the application itself, since, in most cases, no program has to be written. The mentioned techniques work ®ne as far as fuzzy approaches, which need to be combined with each other or integrated into a company-wide business application system, are developed with the proposed fuzzy components. However, collaboration between fuzzy applications that were developed using different development environments still leads to integration problems. In this paper, we propose a common interchange format for fuzzy information to …