Product configuration systems are used to configure modular products according to customer requirements. When they are used in the sales process, they are often called sales configuration systems. With shorter product life-cycles, increased use of third-party products, and the need to reduce lead times, manual handling of the product configuration task is becoming a bottle-neck. The cost of configuration errors alone represents several percent of company revenues. For big companies, this means hundreds of millions of dollars. The possibility to automate the product configuration task was recognized in the 1980's, and is now a rapidly growing industry (cf. for example the commercial success of Trilogy).
Previous attempts to automate this task failed, because of difficulties in maintaining the systems when the configuration knowledge was implemented in procedural programs. Product configuration systems need large amounts of knowledge, about the mapping from requirements to components, and about technical restrictions. This knowledge is subject to continuous change, and hard to validate. For this reason, methods and technology for knowledge engineering are needed in the development and maintenance of such systems, as recognized by all tool vendors in the field.
Researchers in knowledge engineering have long known that tools with general inference engines, e.g. expert system shells, are not powerful enough for efficient knowledge engineering. What is needed is tools that use a problem solving method that is specialized for the kind of task at hand, e.g. technical diagnosis or product configuration, and where the application specific knowledge is represented in a vocabulary suited for this task. Such tools specialized for product configuration tasks are called configurators.
Back to Product Configuration at