Conclusions

Abstract

Marketplaces for B2B electronic commerce have great economic potential. They provide openness and flexibility in establishing commercial relationships for their clients. In order to provide this service they have to overcome serious obstacles, especially the integration of various styles to describe the content and the structure of the exchanged information. Product catalogues correspond with large and complex domain ontologies, and in the case of horizontal standards, with upper-layer ontologies. Large scale modelling and mapping efforts are required to handle these descriptions. Content managers have to structure, classify, reclassify, and personalize large volumes of data to make product descriptions automatically accessible via B2B marketplaces. B2B marketplaces have to deal with a plethore of different product catalog and business document formats that require extensive aligning efforts to provide integrated services to their clients. Here, many concepts and techniques developed in Intelligent Information Integration ⁶³ (III, cf. [Wache & Fensel, 2000], [Fensel et al., 200la]) and related areas can significantly help to overcome these problems. However, most of them require adaptation to the specific needs of electronic commerce. In the book we discussed many examples of how this can be achieved:

• GoldenBullet applies the information retrieval metaphor to the problem of automatically classifying product descriptions according to code systems. The product description is taken as a query and the coding system is taken as a collection of documents. Finding the right code is solved by retrieving the right “document”.

• We described a multi-layered mapping architecture that helps to align different standards by selecting and combining simple mapping patterns.