Temporal Knowledge Graph Question Answering: A Survey

TL;DR

This survey advances the field of Temporal Knowledge Graph Question Answering (TKGQA) by proposing a unified taxonomy of temporal questions and a three-way method categorization: Semantic Parsing-based, TKG Embedding-based, and LLM-based approaches. It formalizes temporal knowledge graphs with $(s,r,o,t)$ quadruples under $\,mathcal{G}= (\\mathcal{E}, \\mathcal{R}, \\mathcal{T}, \\mathcal{F})$ and reviews representative techniques for grounding, representation, and reasoning, including $TComplEx$-style tensor decompositions and time-aware GNNs, as well as RAG and agentic strategies for LLMs. The paper analyzes datasets, evaluation metrics (e.g., Hits@k, MRR, ARI), and leaderboards, documenting that LLM-based methods currently achieve strong performance on many benchmarks while SP-based methods excel in explicit temporal reasoning. It also outlines future directions such as expanding question types and modalities, improving robustness and generalization, and leveraging LLMs for temporal reasoning to drive practical applications in time-sensitive information querying. Overall, the work provides a comprehensive reference to guide researchers in developing more capable, temporally aware QA systems over KGs and their dynamic extensions.

Abstract

Knowledge Base Question Answering (KBQA) has been a long-standing field to answer questions based on knowledge bases. Recently, the evolving dynamics of knowledge have attracted a growing interest in Temporal Knowledge Graph Question Answering (TKGQA), an emerging task to answer temporal questions. However, this field grapples with ambiguities in defining temporal questions and lacks a systematic categorization of existing methods for TKGQA. In response, this paper provides a thorough survey from two perspectives: the taxonomy of temporal questions and the methodological categorization for TKGQA. Specifically, we first establish a detailed taxonomy of temporal questions engaged in prior studies. Subsequently, we provide a comprehensive review of TKGQA techniques of two categories: semantic parsing-based and TKG embedding-based. Building on this review, the paper outlines potential research directions aimed at advancing the field of TKGQA. This work aims to serve as a comprehensive reference for TKGQA and to stimulate further research.

Figures (9)

• Figure 1: Given a temporal question (purple box), the TKGQA task aims to derive the answer from the underlying TKG (green box).
• Figure 2: Taxonomy of temporal questions from three aspects, including (a) Question Content; (b) Answer Type, and (c) Complexity. Each grey box contains an example of a question-answer pair.
• Figure 3: Overall procedure of SP-based methods. The representation forms of ungrounded query and executable query are $\lambda$-calculus and SPARQL, respectively.
• Figure 4: Semantic framework of temporal constraints for question "Which movie did Alfred Hitchcock [$_\mathsf{Event_1}$ direct] [$_\mathsf{Signal_1}$ in] [$_\mathsf{Time_1}$ 1960]?".
• Figure 5: Overall procedure of TKGE-based methods.