Collective List-only Entity Linking: A Graph-based Approach

List-only entity linking is the task of mapping ambiguous mentions in texts to target 8 entities in a group of entity lists. Different from traditional entity linking task, which leverages rich 9 semantic relatedness in knowledge bases to improve linking accuracy, list-only entity linking can 10 merely take advantage of co-occurrences information in entity lists. State-of-the-art work utilizes 11 co-occurrences information to enrich entity descriptions, which are further used to calculate local 12 compatibility between mentions and entities to determine results. Nonetheless, entity coherence is also 13 deemed to play an important part in entity linking, which is yet currently neglected. In this work, in 14 addition to local compatibility, we take into account global coherence among entities. Specifically, we 15 propose to harness co-occurrences in entity lists for mining both explicit and implicit entity relations. 16 The relations are then integrated into an entity graph, on which Personalized PageRank is incorporated 17 to compute entity coherence. The final results are derived by combining local mention-entity similarity 18 and global entity coherence. The experimental studies validate the superiority of our method. Our 19 proposal not only improves the performance of list-only entity linking, but also opens up the bridge 20 between list-only entity linking and conventional entity linking solutions. 21