Using Crowdsourcing for Fine-Grained Entity Type Completion in Knowledge Bases

摘要: Recent years have witnessed the proliferation of large-scale Knowledge Bases (KBs). However, many entities in KBs incomplete type information, and some are totally untyped. Even worse, fine-grained types (e.g., BasketballPlayer) containing rich semantic meanings more likely to be incomplete, as they difficult obtained. Existing machine-based algorithms use predicates birthPlace) infer their missing types, limitations that may insufficient types. In this paper, we utilize crowdsourcing solve problem, address challenge controlling cost. To end, propose a hybrid machine-crowdsourcing approach for entity completion. It firstly determines “representative” via then infers remaining based on results. support approach, first an embedding-based influence inference which considers not only distance between embeddings but also distances embeddings. Second, new difficulty model selection can better capture uncertainty machine algorithm when identifying We demonstrate effectiveness our through experiments real platforms. The results show method outperforms state-of-the-art by improving completion at affordable