ProdLight: A Lightweight Ontology for Product Description Based on Datatype Properties

Abstract

Abstract:Web pages representing offerings of products and services are a major source of data for Semantic Web-based e-commerce. This data could be useful for numerous applications, e.g. (1) more precise product search engines and shopping bots, (2) aggregation or enrichment of multi-vendor catalogs using public product descriptions, or (3) the automated discovery of additional alternatives based on the combination of multiple items. While there are already some ontologies for products and services available, they are very large in size (20 -70,000 classes), and thus not always suitable as ontology imports. In this paper, we take a different approach: We represent the semantics of offerings on the Web using a very lightweight ontology of datatype properties in combination with popular classifications like UNSPSC and eCl@ss. We then demonstrate how this representation can be mapped easily to comprehensive ontologies for products and services like eClassOWL 1 . Our approach provides a straightforward solution for annotating offerings on the Web while avoiding the overhead of importing fully-fledged products and services ontologies in any single annotation. We can show that our proposal has technical advantages and eliminates legal problems when reusing existing standards.

Using BMEcat Catalogs as a Lever for Product Master Data on the Semantic Web

“…Separately, the gradual realization of the Semantic Web vision has motivated significant efforts aimed at representing existing e-commerce product related data and classification standards adopting open semantic technologies and data models [7,8,2]. Yet, in the context of managing product master data in particular, two previous solutions stand out [3,16] based on their similarities with respect to our problem scenario.…”