An efficient clustering algorithm for class-based language models

This paper defines a general form for class-based probabilistic language models and proposes an efficient algorithm for clustering based on this. Our evaluation experiments revealed that our method decreased computation time drastically, while retaining accuracy.