Anwendung der Konzeptanalyse und ontologische Modellierung in der Wirtschaftsinformatik: Ontology-based Product Catalogues: An Example Implementation

Philipp Nowakowski, Heiner StuckenschmidtOpenalex Unknown 2010 · DOI · OpenAlex

Abstract

Electronic Product Catalogues are the basis for offering and selling products in online market places. To be efficient, these catalogues have to provide a semantically precise description of product features to allow for effective matchmaking of products and customer requests. At the same time, the description has to follow a common terminology that allows the integration with the catalogues of other providers on the same market place. In this paper, we argue for the use of ontologies as a basis for defining product catalogues. More precisely, we suggest using eClassOWL, an ontological model of the eClass Standard as a basis for encoding product information as formal class descriptions in the web ontology language OWL. We support our argument by presenting an ontology based implementation of a real product catalogue that uses eClass terminology and supports flexible matchmaking based on approximate logical reasoning.

Introduction

In B2B e-commerce, finding specific industrial components is a complex task that often exceeds the capabilities of standard keyword-based search engines. Traditional Electronic Product Catalogues (EPCs) face two primary challenges. First, they often rely on "all-or-nothing" logic; if a product does not perfectly match every user constraint, it is excluded, even if it is a viable alternative. Second, the lack of standardized terminology across different suppliers makes cross-catalogue integration difficult. This paper proposes a semantic matchmaking approach using the Web Ontology Language (OWL) and the eCl@ss standard to enable flexible, approximate reasoning.

System Design and Methodology

The core of the proposed system is the transformation of industrial data into a structured ontological format. By utilizing eClassOWL—a mapping of the eCl@ss 5.1 standard into OWL—the system establishes a global hierarchy of product classes and properties. This provides the "semantic glue" necessary for interoperability.

The technical architecture, implemented using the Grails framework and the OWLAPI, centers on components that manage ontologies and perform query rewriting to find partial matches.

UML class diagram of the ontology-based matchmaking system architecture.

Matchmaking via Relaxation

To overcome the binary nature of traditional logic, the system employs stepwise constraint relaxation. Instead of a single pass-fail check, the algorithm iteratively "ignores" certain properties to identify the next best matches. This process is guided by user-defined categories:

  • Blacklist: Mandatory constraints that cannot be relaxed (e.g., physical dimensions).
  • Whitelist: Low-priority properties that can be relaxed immediately.
  • Greylist: Properties that the algorithm relaxes sequentially to find approximate matches.
The prototype interface showing property categorization for stepwise relaxation.

Furthermore, the system utilizes taxonomic expansion. If no matches exist within a specific product category, the reasoner automatically explores "sibling" categories within the ontology hierarchy to find similar alternatives.

Implementation and Results

The prototype was validated using a real-world product catalogue from Pepperl+Fuchs, a manufacturer of industrial sensors. The implementation successfully demonstrated qualitative improvements in search results. For instance, when an exact sensor was unavailable, the system could suggest a functional alternative with a different connector type, explicitly highlighting which constraints were relaxed.

Beyond flexible matching, the use of an OWL reasoner (such as Pellet2) allowed for automated consistency checks. The system could identify incorrectly classified products within the catalogue, a task that is significantly more difficult to automate in traditional relational database systems.

Discussion and Conclusion

While the prototype demonstrates the power of semantic matchmaking, several challenges remain. The "mapping problem"—the effort required to align proprietary legacy data with global standards like eCl@ss—remains a significant bottleneck. Additionally, while symbolic logic excels at categorical reasoning, future iterations must better integrate numerical reasoning for concrete values like voltage or temperature ranges.

In conclusion, moving from keyword searches to ontology-based reasoning allows for a more nuanced and effective B2B e-commerce experience. By treating product descriptions as formal class descriptions, systems can provide "good enough" matches that would otherwise be lost in traditional search paradigms.

Summary updated

Abstract

Electronic Product Catalogues are the basis for offering and selling products in online market places. To be efficient, these catalogues have to provide a semantically precise description of product features to allow for effective matchmaking of products and customer requests. At the same time, the description has to follow a common terminology that allows the integration with the catalogues of other providers on the same market place. In this paper, we argue for the use of ontologies as a basis for defining product catalogues. More precisely, we suggest using eClassOWL, an ontological model of the eClass Standard as a basis for encoding product information as formal class descriptions in the web ontology language OWL. We support our argument by presenting an ontology based implementation of a real product catalogue that uses eClass terminology and supports flexible matchmaking based on approximate logical reasoning.
BibTeX
@inproceedings{nowakowski2010anwendung,
  author = {Philipp Nowakowski and Heiner Stuckenschmidt},
  title = {{Anwendung der Konzeptanalyse und ontologische Modellierung in der Wirtschaftsinformatik: Ontology-based Product Catalogues: An Example Implementation}},
  booktitle = {Openalex Unknown},
  year = {2010},
  doi = {10.17875/gup2010-1479},
}
Figure