In the precision_measurements.zip file each processed file has tab-delimited lines with the following:
	the tag (0 for an error, 1 for correct, and -1 indicating the first extraction before a conjunction or comma is correct.  In the experiments, we truncate the extractions on conjunctions and commas -- see the thesis -- so -1 extractions are counted as correct)
	the textual extraction
	the number of extractions (in 100 sentences)
	the sentences from which the extraction arose

Below are the randomly selected Wordnet classes we considered for precision measurements.  The "taggable" ones were included in the measurements; nouns which are not taggable were either overly general such that nearly any extraction would be correct (e.g., the class Example) or rarely or never used to name concrete instances (e.g., the class Purchases).

class	Taggable?
aspects	0
finances	0
relaxants	1
thinkers	1
examples	0
machines	1
pools	0
mushrooms	1
measures	0
keys	0
cultures	1
sections	0
wounds	1
suites	0
resorts	1
heavyweights	0
blockers	0
animations	0
announcements	0
searches	0
directions	0
purchases	0
tones	1
collectors	0
fires	0
periods	0
rings	0
devices	1
mechanisms	1
relations	0
pictures	0
comforts	0
actions	0
avenues	0
solvents	1
obligations	0
flies	1