2020/11/05 by Botha, Jan A., Shan, Zifei, Gillick, Daniel · 2 citations
#Computation and Language (cs.CL) #FOS: Computer and information sciences #H.3.3 #I.2.7 #Information Retrieval (cs.IR)
paper · doi:10.48550/arxiv.2011.02690
We propose a new formulation for multilingual entity linking, where language-specific mentions resolve to a language-agnostic Knowledge Base. We train a dual encoder in this new setting, building on prior work with improved feature representation, negative mining, and an auxiliary entity-pairing task, to obtain a single entity retrieval model that covers 100+ languages and 20 million entities. The model outperforms state-of-the-art results from a far more limited cross-lingual linking task. Rare entities and low-resource languages pose challenges at this large-scale, so we advocate for an increased focus on zero- and few-shot evaluation. To this end, we provide Mewsli-9, a large new multilingual dataset (http://goo.gle/mewsli-dataset) matched to our setting, and show how frequency-based analysis provided key insights for our model and training enhancements.