Event Prediction in Event-Centric Knowledge Graphs Using BERT

Takanori Ugai, Shusaku Egami, Ken Fukuda

Abstract: The dataset provided by the Knowledge Graph Reasoning Challenge is an event-centric knowledge graph representation of avatar behavior in a virtual space. Therefore, it contains information about the entire virtual space, including parts of the virtual space that are hidden in the video. However, some information cannot be obtained by any means; for example, when sensor data from the real world are collected. In this paper, we report a method for predicting hidden parts using BERT and the results of experiments using this method, using the Challenge dataset as complete data for training.