A large-scale system for annotating and querying quotations in news feeds

Abstract

In this paper, we describe a system that automatically extracts quotations from news feeds, and allows efficient retrieval of the semantically annotated quotes. APIs for real-time querying of over 10 million quotes extracted from recent news feeds are publicly available. In addition, each day we add around 60 thousand new quotes extracted from around 50 thousand news articles or blogs. We apply computational linguistic techniques such as coreference resolution, entity recognition and disambiguation to improve both precision and recall of the quote detection. We support faceted search on both speakers and entities mentioned in the quotes.