First provide a summary of the paper, and then address the following criteria: Quality, clarity, originality and significance. This paper presents a method for learning the structure of stochastic And-Or grammars. The paper suggests that this generalizes previous work on structure learning, but it's actually a special case, which makes the problem tractable. This is a reasonable point on its own, and the paper makes a nice contribution, so I wouldn't try to argue that the problem is more general than the structure learning problem faced in NLP. The basic algorithm is sensible and successful when compared against other methods for inducing grammars.