Benefit from Automated Product Recommendations at kdmz.

Abstract

There are a number of recommender systems with different characteristics (Nageswara Rao and Talwar 2008). For a company it is not easy to select the right system. For the introduction of a recommender system is decisive, what kind of data exists and how suitable this data is for calculating recommendations. The question must be answered which algorithm calculates the best recommendations with these data. Afterwards the recommender system has to be developed and integrated into the existing systems (e. g. e-shop). This paper describes the case where a recommender system was introduced successfully into an existing system. For the description of this case the eXperience Methodology was used (Schubert and Wölfle 2007). According to this methodology the company is described first, and then some views onto the final solution, the project and development and to conclude the experiences which were made with the solution.