Design Challenges for Entity Linking Xiao Ling Sameer Singh Daniel S. Weld University of Washington, Seattle WA fxiaoling,sameer,
[email protected] Abstract provide semantic annotations to human readers but also a machine-consumable representation of Recent research on entity linking (EL) has in- the most basic semantic knowledge in the text. troduced a plethora of promising techniques, Many other NLP applications can benefit from ranging from deep neural networks to joint in- ference. But despite numerous papers there such links, such as distantly-supervised relation is surprisingly little understanding of the state extraction (Craven and Kumlien, 1999; Riedel et al., of the art in EL. We attack this confusion by 2010; Hoffmann et al., 2011; Koch et al., 2014) that analyzing differences between several versions uses EL to create training data, and some coreference of the EL problem and presenting a simple systems that use EL for disambiguation (Hajishirzi yet effective, modular, unsupervised system, et al., 2013; Zheng et al., 2013; Durrett and called VINCULUM, for entity linking. We con- Klein, 2014). Unfortunately, in spite of numerous duct an extensive evaluation on nine data sets, papers on the topic and several published data comparing VINCULUM with two state-of-the- art systems, and elucidate key aspects of the sets, there is surprisingly little understanding about system that include mention extraction, candi- state-of-the-art performance. date generation, entity type prediction, entity We argue that there are three reasons