BIG 2017 CUP

Capturing human knowledge and understanding of the world has been a holy grail of semantic computing. As we continue to make progress in knowledge base technologies, we keep pushing the frontiers of machine learning to capture relationships among concepts and ideas as understood by humans and to step beyond human capabilities, to reveal patterns and insights that are not easily observable and thus yet unknown. BIG 2017 supports such efforts and, in view of its focus on “Deep Learning and Big Data Analytics”, it has organized a challenge for researchers and practitioners in partnership with Microsoft. Microsoft has built a concept graph and software toolkits, Microsoft Cognitive Toolkit and Microsoft Graph Engine in order to explore ways to capture meaning in human communication through deep learning and network graph processing.

One of the key issues in the creation and usage of concept graphs extracted from free text is annotation of concepts with tags that capture concept relationships. Microsoft has released the data that established IsA relationship among concepts mined from the Web. The objective of the BIG2017 CUP is to encourage exploration of relationship tagging methods. The task is to predict whether a pair of given concepts has IsA relationship. Innovative approaches, combining deep learning and big graph processing techniques, are encouraged.

The BIG2017 CUP is organized in partnership with Microsoft

Slides of the presentation

  • The winners are:
    Team Members Affiliation Accuracy
    1st winner Dakgalbi Gyeongbok Lee, Hyunsouk Cho Yonsei University, Postech 96%
    2nd winner ProbasePlus Jiaqing Liang Fudan University 87.33%
    3rd winner Patron Seungtaek Choi, Kyungjae Lee Yonsei University 84.67%