Authors: Xinxun Zhang, Yuxiang Wang, Shanshan Feng, Jun Zhang, Tieyun Qian, Binhang Yuan, Kai Yin, Xiao Yan, Jiawei Jiang
Organizations: School of Computer Science, Wuhan University · Tencent Inc. · Hong Kong University of Science and Technology · School of Computer Science, Central China Normal University
Recent advances in large language models (LLMs) have spurred growing interest in temporal knowledge graph question answering (TKGQA). Most existing methods rely on semantic retrieval augmented generation (RAG), which suffers from fundamental limitations in complex temporal reasoning. On the one hand, semantic-based RAG methods struggle to capture the multi-hop causal knowledge required for complex temporal reasoning. On the other hand, these methods introduce excessive redundant knowledge, trapping LLMs in spurious reasoning chains that potentially result in erroneous answers. To address these problems, we propose a Temporal Causal Chain-guided RAG (TCC-RAG) to retrieve a concise yet sufficient set of causal knowledge to support complex temporal reasoning. Specifically, we reformulate the temporal knowledge graph into a multi-view causal cue graph, with causal cues inducing connections that explicitly encode temporal evolution and causal dynamics. Based on the causal cue graph, we construct a query-specific temporal causal chain to guide the retrieval process, enabling the model to focus on a concise set of causally relevant knowledge. Finally, we introduce a causal evidence filter to refine the retrieved knowledge, ensuring its causal consistency with the underlying temporal logic of the query. Experiments show that TCC-RAG achieves substantial performance gains of 10.72% and 8.40% on two TKGQA benchmarks, and delivers a remarkable 62.97% improvement for multi-step temporal reasoning.
Figures & tables
Fig. 1: The limitations of semantic-based RAG methods and design of our temporal causal chain-guided RAG approach.
Fig. 2: The framework of TCC-RAG. (a) Temporal facts are organized into a Temporal Event Graph (TEG) and two temporally sorted indices: the Subject Causal Cue Index (SCCI) and Object Causal Cue Index (OCCI). (b) NER and LLM extraction ground query entities through matching and consistency checking. Relation, cue, and role parsing then guide view routing and temporal chain retrieval. (c) The Causal Evidence Filter removes equivalent duplicates and selects a sufficient subset of query-consistent facts. (d) The LLM reasons over the filtered evidence to generate answers. When structured retrieval yields no facts, semantic retrieval provides fallback candidates that also pass through the filter.
Structure
Storage
Construction
Per-query
TEG ( Gt )
O(∣F∣)
O(∣F∣)
O(deg)
SCCI ( Ds )
O(∣F∣)
O(∣F∣logdmax)
O(logdmax+k)
OCCI ( Do )
O(∣F∣)
O(∣F∣logdmax)
O(logdmax+k)
TABLE I: Complexity of the multi-view causal cue graph. ∣F∣ : total facts; dmax : maximum posting list length; k : number of cue-consistent events per query. Per-query measures index retrieval with grounded keys and a resolved temporal boundary (see Section IV-E ).
Fig. 3: The prompt template for causal chain-guided temporal reasoning based answer generation.
Category
Train
Dev
Test
Single
Equal
135,890
18,983
17,311
Before/After
75,340
11,655
11,073
First/Last
72,252
11,097
10,480
Multiple
Equal Multi
16,893
3,213
3,207
After First
43,305
6,499
6,266
Before Last
43,107
6,532
6,247
TABLE II: Statistics of question categories in MULTITQ.
Category
Train
Dev
Test
Explicit
2,724
1,302
1,311
Implicit
651
291
292
Temporal
2,657
1,073
1,067
Ordinal
938
570
567
Total
6,970
3,236
3,237
TABLE III: Statistics of question categories in TimeQuestions.
Model
Hits@1
Hits@10
Overall
Question Type
Answer Type
Overall
Question Type
Answer Type
Multiple
Single
Entity
Time
Multiple
Single
Entity
Time
BERT
0.083
0.061
0.092
0.101
0.040
0.441
0.392
0.461
0.531
0.222
DistilBERT
0.083
0.074
0.087
0.102
0.037
0.482
0.426
0.505
0.591
0.216
ALBERT
0.108
0.086
0.116
0.139
0.032
0.484
0.415
0.512
0.589
0.228
EmbedKGQA
0.206
0.134
0.235
0.290
0.001
0.459
0.439
0.467
0.648
0.001
TABLE IV: Results of TCC-RAG and baselines on Hits@1 and Hits@10 on MULTITQ. The best and second best results are marked in bold and underlined, respectively. The bottom row is the relative improvement of TCC-RAG over the best baseline.
Model
Overall
Multiple
Single
Entity
Time
w/o CCG
0.647
0.127
0.837
0.517
0.933
w/o PQP
0.778
0.653
0.824
0.759
0.821
w/o CEF
0.841
0.664
0.910
0.791
0.958
TCC-RAG
0.847
0.691
0.911
0.801
0.959
TABLE VI: Ablation studies (Hits@1) of TCC-RAG on MULTITQ.
Fig. 4: A comparison of the retrieved facts between semantic-based RAG methods and TCC-RAG on MULTITQ. Facts highlighted in red are causally essential evidence.
Dataset
Path
Coverage
Hits@1
MULTITQ
Structured
84.46%
0.912
Fallback
15.54%
0.492
TimeQuestions
Structured
91.82%
0.738
Fallback
8.18%
0.398
TABLE VII: Retrieval-path coverage and conditional Hits@1.
Fig. 5: Experiment results of Hits@1 across multiple temporal granularities on the MULTITQ dataset.
Fig. 6: Comparison of token cost and inference time between TCC-RAG and RTQA on MULTITQ.
Knowledge Graph Question Answering answers natural-language questions by finding and reasoning over entities and relations in structured knowledge graphs.
Temporal Knowledge Graph Question Answering (TKGQA) is challenging because it requires multi-hop reasoning under complex temporal constraints. Recent LLM-based approaches have improved semantic modeling for this task, but many still rely on fixed reasoning workflows or costly post-training, which can limit adaptability and make error recovery difficult. We show that enabling an off-the-shelf Large Language Model (LLM) to determine its next action is already effective in a zero-shot setting. Based on this insight, we propose AT2QA, an Autonomous and Training-free Agent for TKG Question Answering. AT2QA empowers the LLM to iteratively interact with the TKG via a generic search tool, inherently enabling autonomous exploration and dynamic self-correction during reasoning. To further elicit the LLM's potential for complex temporal reasoning, we introduce a training-free experience mining mechanism that distills a compact few-shot demonstration library from successful self-generated trajectories. AT2QA also yields a transparent audit trail for every prediction. Experiments on three challenging benchmarks -- MultiTQ, Timeline-CronQuestion, and Timeline-ICEWS-Actor -- show that AT2QA achieves new state-of-the-art performance, surpassing the strongest baselines by 10.7, 4.9, and 11.2 absolute points, respectively. Our code is available at https://github.com/AT2QA-Official-Code/AT2QA-Official-Code
Xufei Lv, Jiahui Yang, Haoyuan Sun +5
Tsinghua University · National University of Defense Technology
Retrieval-Augmented Generation (RAG) has shown strong effectiveness in grounding Large Language Models (LLMs) with external knowledge. However, existing RAG and Graph RAG frameworks largely treat knowledge as static or associate time with coarse-grained timestamps or metadata, failing to capture rich temporal structures such as duration, overlap, and containment. We propose IA-RAG, a hierarchical temporal RAG framework that models knowledge as time intervals and performs retrieval under formal temporal constraints. IA-RAG represents facts as Interval Event Units (IEUs) and organizes them into a hierarchical Thematic Forest, where temporal dependencies are governed by Allen's Interval Algebra. To handle incomplete or uncertain temporal boundaries, IA-RAG further introduces a Sub-graph Time Tightening mechanism that refines fuzzy intervals through logical constraints within connected event subgraphs. In addition, IA-RAG supports implicit temporal semantic retrieval through interval-algebra-guided traversal. Experiments on multiple temporal question answering benchmarks, including TimeQA, TempReason, and ComplexTR, demonstrate that IA-RAG achieves strong temporal retrieval and reasoning performance, particularly on complex compositional temporal reasoning tasks. Our code is released at https://github.com/xiaoAugenstern/LogicalRAG_TemporalQA.
Xiaoman Wang, Yaoze Zhang, Wenzhuo Fan +7
East China Normal University · Shanghai Artificial Intelligence Laboratory · University of Shanghai for Science and Technology +1
Question Answering over Temporal Knowledge Graphs (TKGQA) requires reasoning over time-sensitive facts, yet existing embedding-based methods struggle with multi-step queries due to single-pass reasoning pipelines. We propose SABET-QA, a framework that iteratively refines reasoning states across multiple hops via a bidirectional entity-temporal scoring mechanism and a slot-aware contextualization module that aligns question semantics with temporal KG embeddings. A differentiable working memory enables progressive hypothesis refinement, while auxiliary temporal boundaries serve as coarse supervision when available. Experiments on CronQuestions, Complex-CronQuestions, MultiTQ, and TimeQuestions demonstrate consistent improvements over strong baselines, particularly on complex multi-step temporal queries.
Brahim Touayouch, Mirette Moawad, Dmitry Akulov
QuickSort Research, Paris, France · ENS Paris-Saclay, École Polytechnique, France