Download the SLIDES presented at ECIR 2012.
Links to Demos:
Entity Ranking in Wikipedia
Time Aware Entity Retrieval over News articles
Instead of merely exploiting the syntactic structure of the Web and its documents, it is now possible to leverage semantic information about Web resources. Today, Web Search goes beyond finding documents, and evolves into an interface for finding Web-mediated solutions for user goals involving any type of entity. Therefore, current research challenges include the extraction of information buried within Web pages that can be aggregated and presented to the end user. One example is the query "New York restaurants'' which is not aiming at a ranked list of Websites any of which could provide information about a restaurant in New York; rather, the user would be best satisfied by a list of entities with additional information such as the average price and, possibly, a map displaying the results.
To go beyond current Web search, the next step is to rank, rather than documents, information units of varying type, complexity, and structure, that is, entities. Being able to retrieve entities rather than documents would also allow current Search Engines to answer more complex user queries such as "Countries where I can pay in Euro" or "Italian Nobel Prize winners". In the commercial setting, new prototypes going in this direction are being developed.
In this full-day tutorial (6 hours) we present an overview of techniques and algorithms for different entity search tasks: Expert Finding, Entity Ranking in Wikipedia, and ranking structured data extracted from web pages or published as Linked Data.
The objective is to show how core IR and Semantic Web techniques are making a difference in tasks such as expert finding, entity ranking, semantic search, etc. Additionally, we report results from evaluation studies performed at standard evaluation initiatives such as TREC and INEX. We use the evolution of the Entity Search field through time as a guideline for organizing the tutorial content.
By attending the tutorial, attendants will:
* Acquire knowledge about the different search tasks involved in the Entity Search area
* Acquire an understanding of current Expert Finding, Entity Search, and Semantic Search techniques
* Understand how to properly evaluate Entity Search systems
* Understand how novel search systems based on semantic data operate
Full-day (6 hours plus breaks) on April 1st 2012.
Dr. Gianluca Demartini (eXascale Infolab, University of Fribourg, Switzerland)
Dr. Peter Mika (Yahoo! Research, Barcelona, Spain)
Dr. Thanh Tran (Institut AIFB, Universität Karlsruhe, Germany)
Prof. Arjen P. de Vries (Centre for Mathematics and Computer Science, CWI, The Netherlands)