[1F1] [General Session] 4. Knowledge Utilization and Sharing

[1F1-01] The First Knowledge Graph Completion Challenge 2018

- Toward the Interpretable AI -

〇Takahiro Kawamura¹, Shusaku Egami², Shinichi Nagano³, Ikki Ohmukai⁴, Takeshi Morita⁵, Yasunori Yamamoto⁶, Koji Kozaki⁷ (1. Japan Science and Technology Agency, 2. University of Electro-Communications, 3. Toshiba Corp., 4. National Institute of Informatics, 5. Keio University, 6. Database Center for Life Science, 7. Osaka University)

Keywords:Knowledge Graph, Reasoning, Interpretability

This paper describes a new challenge for addressing knowledge graph completion held in Japan, 2018. Deep learning techniques have opened a new vista of Artificial Intelligence and promoted applying the AI-related systems to a wide variety of social systems. However, security and safety of such AI systems in addition to the interpretability of AI decision would become extremely important accordingly, as already warned in several communities. Therefore, the Special Interest Group (SIG) on Semantic Web and Ontology of JSAI organized the challenge, in which a knowledge graph showing well-known novels of Sherlock Holmes is openly constructed, and techniques for reasoning or completing the missing information about criminals that was intentionally removed from the knowledge graph are recruited from the general public. The SIG has started the challenge at the 32nd JSAI Annual Conference 2018, and hopes many participants concerned will have an active discussion to this end.