Negative statements considered useful

Knowledge bases (KBs) about notable entities and their properties are an important asset in applications such as search, question answering dialogue. All popular KBs capture virtually only positive statements, abstain from taking any stance on statements not stored the KB. This paper makes case for explicitly stating salient that do hold. Negative useful to overcome limitations of systems mainly geared questions; they can also contribute informative summaries entities. Due abundance invalid effort compile them needs address ranking by saliency. We present a statisticalinference method compiling negative based expectations related peer groups. Experimental results, with variety datasets, show effectively discover extrinsic studies underline usefulness entity summarization. Datasets code released resources further research.