Bits & Pieces from WWW 2014
WWW 2014? Don’t worry, we have you covered with a selection of interesting talks & links gathered by the XI team on site in Seoul! If you would like to add something to the list below, just drop us an email!
Day 1 (Monday)
The DEOS 2014 Workshop keynote by Luna Dong was really worth following. Luna presented Sonya, a data fusion system that allows to merge facts from different knowledge bases and to associate a belief score to them, based on her VLDB09 publication. As a side note, wandering around Luna’s website we also found a list of papers worth reading (in her opinion) which looks quite interesting.
Just after this keynote Dmitry Ushanov presented the Yandex named entity extraction and disambiguation toolkit (anyone has a link to the paper?).
Community-based Crowdsourcing [Marco Brambilla, Stefano Ceri, Andrea Mauri and Riccardo Volonterio], SOCM 2014 Workshop. This paper directly relates to some of our own crowdsourcing work (e.g., Pick-A-Crowd, Hippocampus) where the goal is to crowdsource a task using a selected subset of people/experts belonging to a given community. Andrea Mauri presented compelling results showing how different communities were able to identify the correct picture of a CS professor among a set of candidates.
Entity Linking with a Unified Semantic Representation[Guo, Barbosa], DEOS 2014 Workshop. This paper tackles entity linking (like our own ZenCrowd) by inspecting all entities appearing in the document at once. In addition, the authors analyze the graph of entity types appearing in the document (one graph per document) to identify potential outliers.
Day 2 (Tuesday)
- At LDOW2014, Ramanathan Guha presented an update on schema.org. His talk was pretty similar to the keynote he gave at ISWC 2013. After a quick discussion on the schema.org principles (simplicity, incentives!) and the presentation of few numbers describing its extremely rapid adoption, Guha mentioned some applications of schema.org annotations like Google Now, its MS counterpart Cortana, and the rich pins feature of Pinterest. After the presentation, an interesting discussion followed up on ownership and licensing issues for such annotations.