报告题目：One-Class Order Embedding Learning for Dependency Relation Prediction
报告时间：2020年1月3日 星期五 上午10：00
报 告 人：Amy
I'm a research scientist at the Living Analytics Research Centre (LARC), Singapore Management University (SMU). Before joining LARC, I was a senior software engineer at Yahoo Inc. Taiwan. I received my Ph.D. degree in Computer Science from National Chiao Tung University in Jan 2013. I received my Master's degree and Bachelor’s degree in Computer Science from National Chengchi University in 2004 and 2006. I visited University of Illinois Chicago (UIC) in 2011.
My research focuses on three domains: Data Mining, Knowledge Representation Learning, and Urban Computing. The common focus of my research is to gain insights and support Advanced Knowledge Utilization from massive data. My research has a strong technical focus to develop machine learning approaches to unravel knowledge in various forms.
Most of today's representation learning techniques aim to learn entity representations that aresemantics-preserving, i.e., semantically similar entities are mapped into a nearby area in the semantic embedding space. In this study, we aim to learnorder-preservingrepresentations for entities, i.e., antisymmetric relations between two entities are captured in the embedding space. Learning a good order embedding in knowledge graphs pose several challenges due to the existence of partial ordering and missing relations in the observed data. Moreover, most application scenarios do not provide non-trivial negative dependency relation instances. We propose a framework to perform dependency relation prediction by exploring both rich semantic and hierarchical structure information in the data. In particular, we propose several negative sampling strategies based on graph-specific centrality properties, which supplement the positive dependency relations with appropriate negative samples to effectively learn order embeddings. This research not only addresses the needs of automatically recovering missing dependency relations, but also unravels dependencies among entities using several real-world datasets, such as course dependency hierarchy involving course prerequisite relations, job hierarchy in organizations, and paper citation hierarchy. Extensive experiments are conducted on both synthetic and real-world datasets to demonstrate the prediction accuracy as well as to gain insights using the learned order embedding.