Profiling applications are omnipresent in today's online life: product recommendations in online shops, personalised search platforms etc. Often, however, profiling systems fail to model the complexity of user interests and preferences in all their depth. Especially in efficiency-dependent contexts, profiling systems rely on simple data structrures, such as keyword lists, to ensure system performance. In my talk, I will present an approach for ontology-integrated profiling which is motivated by an application in digital advertising. In collaboration with an industrial actor, we developed a customised semantic model of the advertising-specific profiling process. In particular, we focussed on pushing the boundaries of ontology integration. Instead of using the ontology-based knowledge base as a passive storage facility, we aimed to transform it into an active contributor to the profiling result.
05.11.2015 - 10:15