Towards A Question Answering System over Temporal Knowledge Graph Embeddings

Question Answering (QA) over knowledge graphs is a vital topic within information retrieval. Questions with temporal intent are a special case of questions for QA systems that have received only limited attention so far. In this paper, we study using temporal knowledge graph embeddings (TKGEs) for temporal QA. Firstly, we propose a microservice-based architecture for building temporal QA systems on pre-trained TKGE models. Secondly, we present a Bayesian model average (BMA) ensemble method, where results of several link prediction tasks on separated TKGE models are combined to find better answers. Within the system built using the microservice-based architecture, the experiments on two benchmark datasets show that BMA provides better results than the individual models.