I am looking for a suitable method for finding which of the two nouns in a 2-word noun compound is the head noun. Can a dependency parser be used for the same?
the Connexor parser (http://www.connexor.com/nlplib/ license required!) lemmatizes compounds in a way that the nouns a compound is made of are recognizable (at least it did so for German some years ago).
Use the "IS A" test - is the compound a kind of N1 or N2. If neither it is exocentric. This test can be applied in both the syntactic and semantic domains - and there may be a mis-match of which noun is the head. See the following articles:
Bisetto, Antonietta, and Scalise, Sergio. 2005. The classification of compounds. Lingue e Linguaggio 4:319-332.
Scalise, Sergio, and Guevara, Emiliano. 2006. Exocentric compounding in a typological framework. Lingue e Linguaggio 2:185-206.
Thanks for the nice references. I had not come across these ref's. However I am still a little confused in the sense, if we consider an example say "coal mine" so IS-A test answers that it is a Mine so we assume that Mine is the Head. But if we look as examples like "car wheel" or "door knob" according to IS-A the heads should be wheel and knob respectively but I feel specially for the car wheel case that it is not a wheel but it is a part of car/ same can be applied to knob as well. Am I wrong if I am thinking that car and door should be heads here.
Regarding the car wheel - it is part of a car but not a car. Just like your arm - it's part of your body, but not your body. General with English N-N compounds the first noun is the modifier and the second noun is the head.