Search from History and Reason for Future: Two-stage Reasoning on Temporal Knowledge Graphs

Proceedings of the 59th Annual Meeting of the Association for Computational Linguistics and the 11th International Joint Conference on Natural Language Processing, pages 4732–4743 August 1–6, 2021. ©2021 Association for Computational Linguistics 4732 Search from History and Reason for Future: Two-stage Reasoning on Temporal Knowledge Graphs Zixuan Li1,2, Xiaolong Jin1,2, Saiping Guan1,2, Wei Li3, Jiafeng Guo1,2, Yuanzhuo Wang1,2 and Xueqi Cheng1,2 1School of Computer Science and Technology, University of Chinese Academy of Sciences; 2CAS Key Laboratory of Network Data Science and Technology, Institute of Computing Technology, Chinese Academy of Sciences; 3Baidu Inc. {lizixuan,jinxiaolong,guansaiping}@ict.ac.cn liwei85@baidu.com Abstract Temporal Knowledge Graphs (TKGs) have been developed and used in many different ar- eas. Reasoning on TKGs that predicts poten- tial facts (events) in the future brings great challenges to existing models. When facing a prediction task, human beings usually search useful historical information (i.e., clues) in their memories and then reason for future meticulously. Inspired by this mechanism, we propose CluSTeR to predict future facts in a two-stage manner, Clue Searching and Tem- poral Reasoning, accordingly. Specifically, at the clue searching stage, CluSTeR learns a beam search policy via reinforcement learn- ing (RL) to induce multiple clues from histor- ical facts. At the temporal reasoning stage, it adopts a graph convolution network based se- quence method to deduce answers from clues. Experiments on four datasets demonstrate the substantial advantages of CluSTeR compared with the state-of-the-art methods. Moreover, the clues found by CluSTeR further provide in- terpretability for the results. 1 Introduction Temporal Knowledge Graphs (TKGs) (Boschee et al., 2015; Gottschalk and Demidova, 2018, 2019; Zhao, 2020) have emerged as a very active research area over the last few years. Each fact in TKGs has a timestamp indicating its time of occurrence. For example, the fact, (COVID-19, New medical case occur, Shop, 2020-10-2), indicates that a new medical case of COVID-19 occurred in a shop on 2020-10-2. In this paper, reasoning on TKGs aims to predict future facts (events) for timestamp t > tT , where tT is assumed to be the current timestamp (Jin et al., 2020). An example of the task is shown in Figure 1, which attempts to an- swer the query (COVID-19, New medical case oc- cur, ?, 2020-12-23) with the given historical facts. Obviously, such a task may benefit many practical Query: (COVID-19, New medical case occur, ? , 2020-12-23) Stage 1: Clue Searching Candidatas 12-20 10-2 11-5 (COVID-19, New medical case occur, Shop) 12-21 12-21 10-1 12-21 10-14 Stage 2: Temporal Reasoning Shop Police station Bank Police station Shop Shop Bank Police station (The man, Go to, Shop) (COVID-19, New suspected case occur, Bank) (COVID-19, New medical case occur, Police station) (The man, Go to, Police station) The man (COVID-19, Diagnose , The man) -1 (COVID-19, Diagnose , The man) -1 Figure 1: An illustration of the reasoning process in- spired by human cognition. Different colors indicate different relations. r−1 is the inverse relation of r. applications, such as, emerging events response (Muthiah et al., 2015; Phillips et al., …

Content truncated to 500 words