Full Seminar Details
Dr. Torsten Zesch
Ubiquitous Knowledge Processing lab, TU Darmstadt, Germany
This event took place on Wednesday 08 December 2010 at 11:30
Computing the semantic relatedness between words is a pervasive task in natural language processing. So far, insufficient coverage of linguistic knowledge resources has been a major impediment for using semantic relatedness measures in large-scale applications. Recently, rapidly growing collaboratively constructed resources like Wikipedia and Wiktionary have been discovered as a new kind of semantic resource.
In the talk, I will shortly introduce these new resources and show how existing semantic relatedness measures can be adapted to the new resources. I will then compare the performance of traditional resources (Wisdom of Linguists) with that of the new resources (Wisdom of Crowds), and show under which conditions collaboratively constructed semantic resources can be used as a proxy for linguistically constructed semantic resources.
Additionally, I will introduce freely available application programming interfaces to Wikipedia and Wiktionary that have been used to conduct the experiments described in my talk.