Leveraging Distributed Human Computation and Consensus Partition for Entity Coreference

Entity coreference is important to Linked Data integration. User involvement is considered as a valuable source of human knowledge that helps identify coreferent entities. However, the quality of user involvement is not always satisfying, which significantly diminishes the coreference accuracy. In this paper, we propose a new approach called coCoref, which leverages distributed human computation and consensus partition for entity coreference. Consensus partition is used to aggregate all distributed user-judged coreference results and resolve their disagreements. To alleviate user involvement, ensemble learning is performed on the consensus partition to automatically identify coreferent entities that users have not judged. We integrate coCoref into an online Linked Data browsing system, so that users can participate in entity coreference with their daily Web activities. Our empirical evaluation shows that coCoref largely improves the accuracy of user-judged coreference results, and reduces user involvement by automatically identifying a large number of coreferent entities