Towards Semantic Improvement of Marketing Recommendation Systems

Abstract :

The functioning of recommendation systems is traditionally based on the analysis of transactional data and customer behavior. However, raw data in many cases is not enough to make good recommendations. The article proposes the use of taxonomies to extend the perspectives of data analysis and interpretation for the needs of recommendation systems. In particular, semantic models (such as taxonomies or ontologies) may be a response to the deficiencies of recommendation systems due to a lack of data interpretation. It has also been proposed to use external, semantically related data sources to expand the analysis perspectives. Many valuable features can be obtained by semantic interpretation of transactional data and combining them with other data sets. The article presents the author's suggestions for extending the scope of data analysis that has been made available by the lingerie retail chain. The analyzes were aimed at expanding the possibilities of acquiring knowledge about customers and products. This knowledge can be used to create better recommendations and targeted marketing campaigns.