A Virtuous Cycle of Semantic Enhancement with DBpedia Spotlight
The talk was delivered by Chris Bizer and Pablo Mendes at the 2012 Semantic Tech & Business Conference (#SemTechBiz) which was held at the Seminaris CampusHotel Berlin in Berlin, Germany on 6-7 February 2012.
Wikipedia is one of the most important repositories of human knowledge, containing millions of interlinked articles. The DBpedia project extracts and combines Wikipedia information into a large multilingual knowledge base that enables semantic processing in a wide range of applications. We have built DBpedia Spotlight, a tool that recognizes ambiguous terms in text and automatically assigns unambiguous definitions to those terms by connecting them to DBpedia. Such interconnection enriches information by providing explicit semantic relationships, enabling semantic indexing, faceted exploration, among other data processing enhancements. In this talk we will describe how DBpedia Spotlight can be applied to establish a virtuous cycle of semantic enhancement. On the one hand, it can enhance knowledge interconnectivity in document collections. On the other hand, it learns how to better annotate from user feedback. Such a positive feedback loop can be applied on Wikipedia itself, or in enterprises to alleviate the cold start problem and knowledge management costs.