For the Treehouse meeting on 10/11/06, we will be discussing Rion Snow, Dan Jurafsky, and Andrew Y. Ng. 2006, which received the best paper award at COLING-ACL 2006, by working through the critical reading questions.

This paper is likely to be quite challenging for the newer students. If you encounter a technical term you don't recognize, please post it here. Likewise, if you can define one of the technical terms posted here, please contribute your definition.

-- EmilyBender - 07 Oct 2006

Meronymy seems to describe two different phenomena: (1) the denotation by one lexeme of a part of a whole that is denoted by another lexeme (a door is part of a building), and (2) the polysemy of a lexeme in which its denotations include primarily a part and secondarily the part's whole (he lives three doors down the street). I gather that phenomenon 1 is the one that Snow et al. talk about.

-- JonathanPool - 11 Oct 2006

I wonder about another issue that might be added to the lists of critical-reading questions. Writings of all these types are communications from authors to readers. So, it seems appropriate to ask what presuppositions (or explicit prerequisites) the writing imposes on the qualifications of readers, and whether these are consistent with the knowledge of the members of the apparently intended audience.

-- JonathanPool - 11 Oct 2006

From a cursory reading, I can't determine whether the authors assert that the existence of a pre-existing semantic taxonomy, e.g. a WordNet, is a requirement for the application of their method. Only about 33 languages in the world have WordNets, so a method of inducing semantic taxonomies for languages without WordNets would be much more valuable than one requiring a WordNet.

-- JonathanPool - 11 Oct 2006

Another question that remains murky for me is whether this article describes a method for discovering facts about a language or a method for discovering facts about the way in which people understand the universe. The article refers to both WordNet and Cyc as "semantic taxonomies", but WordNets are language-specific taxonomies of sets of lexemes, while the "Cyc knowledge base (KB) is a formalized representation of a vast quantity of fundamental human knowledge". To me, this is a basic distinction, and the article seems not to try to specify its position with regard to it.

-- JonathanPool - 11 Oct 2006

Topic revision: r2 - 2006-10-11 - 17:27:37 - JonathanPool
 

This site is powered by the TWiki collaboration platformCopyright & by the contributing authors. All material on this collaboration platform is the property of the contributing authors.
Ideas, requests, problems regarding TWiki? Send feedback
Privacy Statement Terms & Conditions