Automated Knowledge Base Construction (AKBC) 2013: The 3rd Workshop on Knowledge Extraction at CIKM 2013 in San Francisco, October 27-28, 2013.
If you were an author or reviewer, please complete our OpenReview survey!
The advances in information extraction, machine learning, and natural language processing have led to the creation of large knowledge bases (KBs) from Web sources. Notable endeavors in this direction include Wikipedia-based approaches (such as YAGO, DBpedia, and Freebase), systems that extract from the entire Web (such as NELL and PROSPERA) or from specic domains (such as Rexa), and open information extraction approaches (TextRunner, PRISMATIC). This trend has led to new applications that make use of semantics. Most prominently, all major search engine providers (Yahoo!, Microsoft Bing, and Google) nowadays experiment with semantic tools. The Semantic Web, too, benefits from the new approaches.
With this year’s workshop, we would like to resume the positive experiences from two previous workshops: AKBC-2010 and AKBC-WEKEX-2012. The AKBC-2013 workshop will serve as a forum for researchers working in the area of automated knowledge harvesting from text. By having invited talks by leading researchers from industry, academia, and the government, and by focusing particularly on vision papers, we aim to provide a vivid forum of discussion about the field of automated knowledge base construction.
|9:15||9:45||Fabian Suchanek (MPI)||Welcome Message (slides)|
|9:45||10:30||Chris Manning (Stanford)||Texts are Knowledge (slides)|
|11:00||11:45||Bonnie Dorr (DARPA)||HLT Programs at DARPA: Challenges, Solutions, and Applications|
|11:45||12:30||Evgeniy Gabrilovich (Google)||Knowledge Vault: A Web-Scale Approach to Probabilistic Knowledge Fusion
Recent years have witnessed a proliferation of large-scale knowledge bases, including Wikipedia, Freebase, YAGO, Microsoft's Satori, and Google's Knowledge Graph. To increase the scale even further, we need to use automatic methods for knowledge base construction. Previous approaches have primarily focused on text-based extraction, which can be very noisy. In this talk, we will introduce Knowledge Vault, a Web-scale probabilistic knowledge base that combines text-based extractions with prior knowledge, derived from existing knowledge repositories. We employ supervised machine learning methods for fusing these distinct information sources. The Knowledge Vault is substantially bigger than any previously published structured knowledge repository, and features a probabilistic inference system that computes calibrated probabilities of fact correctness. In the second part of the talk, we will discuss the frontiers of research in knowledge discovery on the Web.
|1:30||1:40||Sameer Singh||Joint Inference of Entities, Relations and Coreference|
|1:40||1:50||Xiao Ling||Extracting Meronyms for a Biology Knowledge Base Using Distant Supervision|
|1:50||2:00||Michael Wick||Assessing Confidence of Knowledge Base Content with an Experimental Study in Entity Resolution|
|2:00||2:10||Jay Pujara||Ontology-Aware Partitioning for Knowledge Graph Identification|
|2:10||2:20||Jonathan Gordon||Reporting Bias and Knowledge Extraction|
|2:20||2:30||Bhavana Dalvi||Classifying Entities into an Incomplete Ontology|
|2:30||2:40||Peter Clark||A Study of the AKBC Requirements for Passing an Elementary Science Test|
|2:40||2:50||Luis Galárraga||Mining Rules to Align Knowledge Bases|
|2:50||3:00||Thomas Huet||Mining History with Le Monde|
|9:00||9:45||Andrew McCallum (UMass)||TBA|
|9:45||10:00||Dan Weld (UWashington)||Distant Supervision for Relation Extraction: Progress & Problems|
|11:00||11:45||Tom Mitchell (CMU)||NELL: Three new directions|
|11:45||12:30||James Mayfield (JHU)||TBA|
|1:30||2:15||Alon Halevy (Google)||Biperpedia: An ontology for search applications|
|2:15||3:00||Haixun Wang (Google)||TBA|
All accepted papers should be presented by at least one author. Every paper should be presented a poster. The poster dimensions are as specified here.
In addition, the papers listed under "Talk & posters" should be presented by a short talk. We allocate 10 minutes per talk, which includes speaker change, the talk itself, and one or two quick questions. Please see above for the schedule of the talks. The slides for the talk should be emailed to firstname.lastname@example.org by 9am Pacific on October 27, 2013. We shall use a single comuter (Mac) to save time during talk transition. Please name your file XX_YYY.ZZZ, where XX is the two-digit decimal number of your talk slot (01, 02, ...), YYY is your name, and ZZZ is any of the supported file formats (PPT, PPTX, PDF, SVG, Keynote).