A Deep Learning Approach to Geographical Candidate Selection through Toponym Matching

摘要: Recognizing toponyms and resolving them to their real-world referents is required provide advanced semantic access textual data. This process often hindered by the high degree of variation in toponyms. Candidate selection task identifying potential entities that can be referred a previously recognized toponym. While it has traditionally received little attention, candidate significant impact on downstream tasks (i.e. entity resolution), especially noisy or non-standard text. In this paper, we introduce deep learning method for through toponym matching, using state-of-the-art neural network architectures. We perform an intrinsic matching evaluation based several datasets, which cover various challenging scenarios (cross-lingual regional variations, as well OCR errors) assess its performance context geographical English Spanish.