Hearst states that she tried to apply this strategy to the part-of relation, but failed. We comment later on We present a method for extracting parts of objects the differences in our approach that we believe were from wholes (. "speedometer" from "car"). Given most important to our comparative success. a very large corpus our method finds part words with Looking more widely still, there is an ever55% accuracy for the top 50 words as ranked by the growing literature on the use of statistical/corpussystem. .
