Joint Multi-Facts Reasoning Network For Complex Temporal Question Answering Over Knowledge Graph

TL;DR

This work tackles complex temporal question answering over knowledge graphs by moving beyond single-temporal-fact assumptions. It introduces JMFRN, a Joint Multi-Facts Reasoning Network that retrieves multiple temporal facts and uses entity-aware and time-aware attention to fuse information for joint reasoning, supplemented by an answer-type discrimination task. The model achieves state-of-the-art performance on TimeQuestions, especially for questions involving multiple entities, demonstrating the effectiveness of jointly reasoning over temporal facts and leveraging type constraints. Overall, JMFRN advances practical TKGQA by robustly handling temporally related facts and reducing incorrect answers through explicit answer-type guidance.

Abstract

Temporal Knowledge Graph (TKG) is an extension of regular knowledge graph by attaching the time scope. Existing temporal knowledge graph question answering (TKGQA) models solely approach simple questions, owing to the prior assumption that each question only contains a single temporal fact with explicit/implicit temporal constraints. Hence, they perform poorly on questions which own multiple temporal facts. In this paper, we propose \textbf{\underline{J}}oint \textbf{\underline{M}}ulti \textbf{\underline{F}}acts \textbf{\underline{R}}easoning \textbf{\underline{N}}etwork (JMFRN), to jointly reasoning multiple temporal facts for accurately answering \emph{complex} temporal questions. Specifically, JMFRN first retrieves question-related temporal facts from TKG for each entity of the given complex question. For joint reasoning, we design two different attention (\ie entity-aware and time-aware) modules, which are suitable for universal settings, to aggregate entities and timestamps information of retrieved facts. Moreover, to filter incorrect type answers, we introduce an additional answer type discrimination task. Extensive experiments demonstrate our proposed method significantly outperforms the state-of-art on the well-known complex temporal question benchmark TimeQuestions.

Figures (4)

• Figure 1: An example of a complex temporal question. The answer of the question is Louis Mclane.
• Figure 2: The overview of of our proposed JMFRN.
• Figure 3: Performance of different numbers of grounded entities.
• Figure 4: Convergence process of JMFRN with different $\lambda$.