... classes of gazetteers. For example, wecan use automatically extracted hyponymy relations(Hearst, 1992; Shinzato and Torisawa, 2004), or au-tomatically induced MN clusters (Rooth et al., 1999;Torisawa, ... running on a cluster of PCs with a moderate amount of memory (e.g., 8GB). First, we decided to store the training exampleson a file since otherwise each node would need to store all the examples ... 70%. To cover most of the named entities in the data, we need much largergazetteers. A straightforward approach is to increasethe number of Web documents used for the MN clus-tering and to use...