In this paper, we describe how we created a large, fine-grained, bigram relatedness dataset (BiRD), using a comparative annotation technique called ...
確定! 回上一頁