Extracting Knowledge from Collaboratively Annotated Ad Video Content

摘要

Creative advertising support tools have relied so far on static knowledge represented in creativity templates and decision making systems that indirectly impose restrictions on the brainstorming process. PromONTotion is a system under development that aims at creating a support tool for advertisers for the creative process of designing a novel ad campaign that is based on user-driven, generic, automatically mined and thus dynamic semantic knowledge. Semantic terms and concepts are collaboratively provided via crowdsourcing. The present work describes data mining techniques that are applied to the collected annotations and some interesting initial results regarding ad content, ad genre, ad style and ad impact/popularity information.