DESUMA

To support customers to find the right product for a given use-case is a crucial aspect for both offline as well as online retailers. In the latter case, online product advisors are used to guide customers, based on a set of carefully crafted questions, to the right product set. However, these advisors require a lot of work in terms of data quality, question phrasing and interface design which make them too costly for most small to medium enterprises (SMEs). The Eurostars funded research project DESUMA (E! 10189) aims to provide a marketplace that contains advisor templates which are easy to customize and integrate, thus enabling aforementioned SMEs to facilitate advisors within their systems to improve customer experience and conversion.

In order to create a good product advisor knowledge about the product domain as well their application in real life scenarios is necessary. Based on that knowledge, questions that narrow down all available products to a manageable number of items are needed. The main goal of the Interactive Systems Group is to provide intelligent support for the creation of such questions. State of the art text-mining and analysis algorithms are leveraged to extract information about the given product domain. Extracted information is used to give the advisor template creator insights about the product domain. For example, statistics about feature mentions and sentiments are visualized to guide the design of questions. In summary, intelligent tools and visualizations are developed that support the creation of advisors templates which in turn will be sold to the prospected customers of the marketplace.

Research partners are the Smart Information Systems GmbH, which creates the advisor marketplace application based on their leading advisor technology and the NFC21 GmbH which creates advisors using developed tools and provides insights from real world application.

Related research topics

Contact

J├╝rgen Ziegler

Full Professor

Further contributors

Werner Gaulke

Researcher

Timo Stegemann

Researcher