Automatic Semantic Annotation of Multimodal Product Descriptions

Aim of this project is the development of the technological foundation for the automatic processing of standard product descriptions in the context of eCommerce systems. Product descriptions are to be automatically assigned to product categories using textual and visual information, and specific products are to be identified according to a differentiation using EAN Codes. As things are today, the techniques used in search engines and database inquiries only allow superficial text comparison, which then leads to insecure results and hence needs manual reworking. The project is supported by the EU via the RIS Hamburg.

In cooperation with:

• Gimahhot GmbH, Hamburg
• Innovationsstiftung Hamburg

Involved staff:

• Wiebke Frauen
• Matthias Haringer
• Lothar Hotz
• Thorsten Krebs
• Prof. Dr. Bernd Neumann
• Johannes Schultze
• Michael Wessel