Skip to Main Content
We present a corpus-based method for estimating the im- portance of sentences. Our main contribution is two-fold. First, we introduce the idea of using the increasing amount of manually labeled category information (that is becoming available through collaborative knowledge creation efforts) to identify "typical information" for categories of entities. Second, we provide multiple types of empirical evidence for the usefulness of this notion of typical-information-for-a- category for estimating the importance of sentences.