Some might argue that semantic search is in its infancy. Indeed, when we look at the level of understanding that is involved in modern-day search engines, both on the web and elsewhere, we come to the obvious conclusion that there’s a lot of room for improvement. Most of the time, information objects are retrieved and ordered solely using their textual representation, using little or no knowledge of what it these texts actually mean. My scientific work aims to incorporate this knowledge, e.g., by identifying concepts in queries, tweets, or any other text, and shows that it can be successfully applied to improve information access and retrieval. My work hinges on two things: statistical, bottom-up information access methods and semantic, top-down knowledge. It is where these two come together to form a kind of semantic search where interesting things happen.
Research projects I have been involved with over the years include VL-e, CCCT, and DAESO. CCCT is a collaboration between institutes and groups of several faculties of the University of Amsterdam and the Hogeschool van Amsterdam, which work together on a project basis in the field of content and technology. The aim is to combine the strengths of these groups in research and education in a unique multi-disciplinary setting, and preferably in collaboration with the (creative) industries in and around Amsterdam. DAESO investigated the detection of semantic overlap between Dutch sentences and the exploitation of this knowledge in a range of NLP applications. I also worked on DutchSemCor and LiMoSINe, two projects that center around semantic search, semantic annotations, and semantic information access. Currently, I’m a research scientist at Yahoo Labs where I mainly work on using entities to improve information access in general and the overall user experience in particular.