Extraction of hypernymy information from text

This chapter presents the results of three studies in extracting hypernymy information from a text. In the first, a method based on a single extraction pattern applied to the web is compared with a set of patterns applied to a big corpus. In the second study, it is examined how relation extraction can be performed reliably from a text without having access to a word sense tagger. And in a third experiment, it is checked what the effect of elaborate syntactic information has on the extraction process. Both using more data and the removal of ambiguities from the training data is found to be beneficial for the extraction process. But it is surprising to find a positive effect of additional syntactic information.