[Corpora-List] New Decade Phrase Detectives Competition
Poesio, Massimo
poesio at essex.ac.uk
Sat Jan 9 23:55:25 UTC 2010
To celebrate its first year year of being online, Phrase Detectives:
www.phrasedetectives.org
a game-with-a-purpose designed to gather data about anaphora, announces a $500 New Decade competition aimed at creating the world’s largest collection of anaphorically annotated data.
Modern statistical methods for natural language interpretation require hundreds of thousands of examples of language interpretation. But creating such large amounts of data takes a very long time if done by a handful of people. Web collaboration is a potential solution to this dilemma. In particular, `games with a purpose’ like ESP have been used to label great amounts of data as a byproduct of the activities of people playing such games on the Web. Phrase Detectives, a game with a purpose for anaphoric annotation developed at the University of Essex, has already collected more than 700,000 examples of anaphoric annotation.
To celebrate the first year of being online, the developers of Phrase Detectives are launching a competition to complete the annotation of the first one million words of text, the largest collection of anaphoric data for English in the world. To encourage participation, Phrase Detectives are offering big cash prizes for the top players in January. If you are the highest scorer on January 31st you will win the top prize of $500. We put together a collection that includes around 600,000 words of fiction (from Alice in Wonderland to Sherlock Holmes) and around 600,000 words of text from Wikipedia. The data will be made publically available through LDC and the Anaphoric Bank, www.anaphoricbank.org.
For more information, visit www.phrasedetectives.org or contact
Massimo Poesio – poesio at essex.ac.uk
Jon Chamberlain – jchamb at essex.ac.uk
Udo Kruschwitz – udo at essex.ac.uk
_______________________________________________
Corpora mailing list
Corpora at uib.no
http://mailman.uib.no/listinfo/corpora
More information about the Corpora
mailing list