mlsanityData to 7 Oct 2026

STReasoner: Empowering LLMs for Spatio-Temporal Reasoning in Time Series via Spatial-Aware Reinforcement Learning

arXiv 2601.03248 ↗First posted 6 Jan 2026cs.CLcited by 7

On the map: LLM reasoning → Spatial and embodied reasoning

Abstract (arXiv)

Spatio-temporal reasoning in time series involves the explicit synthesis of temporal dynamics, spatial dependencies, and textual context. This capability is vital for high-stakes decision-making in systems such as traffic networks, power grids, and disease propagation. However, the field remains underdeveloped because most existing works prioritize predictive accuracy over reasoning. To address the gap, we introduce ST-Bench, a benchmark consisting of four core tasks, including etiological reasoning, entity identification, correlation reasoning, and in-context forecasting, developed via a network SDE-based multi-agent data synthesis pipeline. We then propose STReasoner, which empowers LLM to integrate time series, graph structure, and text for explicit reasoning. To promote spatially grounded logic, we introduce S-GRPO, a reinforcement learning algorithm that rewards performance gains specifically attributable to spatial information. Experiments show that STReasoner achieves average accuracy gains between 17% and 135% at only 0.004X the cost of proprietary models and generalizes robustly to real-world data.

Key facts. Cited by 7 papers in the map (data through 7 Oct 2026). Placed in Spatial and embodied reasoning, one of 17 directions in LLM reasoning. Builds on 56 of its 99 references that are in the map. First posted on arXiv 6 Jan 2026.

Builds on

56 of 99 references.

Closest ancestors · 4

Same direction, newest first

  1. 2025UrbanLLaVA: A Multi-modal Large Language Model for Urban Intelligence with Spatial Reasoning and Understanding29 Jun
  2. 2025A Modular Multitask Reasoning Framework Integrating Spatio-temporal Models and LLMs25 Jun
  3. 2025Benchmarking Spatiotemporal Reasoning in LLMs and Reasoning Models: Capabilities and Challenges16 May
  4. 2025Open3D-VQA: A Benchmark for Comprehensive Spatial Reasoning with Multimodal Large Language Model in Open Space14 Mar

Other references · 52

  1. 2024DeepSeekMath: Pushing the Limits of Mathematical Reasoning in Open Language Modelscited by 8,851
  2. 2025Qwen3 Technical Reportcited by 8,195
  3. 2025DeepSeek-R1: Incentivizing Reasoning Capability in LLMs via Reinforcement Learningcited by 6,146
  4. 2024LLaVA-OneVision: Easy Visual Task Transfercited by 2,669
  5. 2022A Time Series is Worth 64 Words: Long-term Forecasting with Transformerscited by 1,981
  6. 2024LlamaFactory: Unified Efficient Fine-Tuning of 100+ Language Modelscited by 1,442
  7. 2025Search-R1: Training LLMs to Reason and Leverage Search Engines with Reinforcement Learningcited by 1,371
  8. 2017Spatio-Temporal Graph Convolutional Networks: A Deep Learning Framework for Traffic Forecastingcited by 1,125
  9. 2019Graph WaveNet for Deep Spatial-Temporal Graph Modelingcited by 820
  10. 2023Time-LLM: Time Series Forecasting by Reprogramming Large Language Modelscited by 680
  11. 2025Vision-R1: Incentivizing Reasoning Capability in Multimodal Large Language Modelscited by 653
  12. 2025Video-R1: Reinforcing Video Reasoning in MLLMscited by 464
38 more
  1. 2025RAGEN: Understanding Self-Evolution in LLM Agents via Multi-Turn Reinforcement Learningcited by 357
  2. 2024Large Language Models for Mathematical Reasoning: Progresses and Challengescited by 283
  3. 2024Time-MoE: Billion-Scale Time Series Foundation Models with Mixture of Expertscited by 280
  4. 2025Logic-RL: Unleashing LLM Reasoning with Rule-Based Reinforcement Learningcited by 260
  5. 2025Time-R1: Post-Training Large Vision Language Model for Temporal Video Groundingcited by 138
  6. 2023Large Models for Time Series and Spatio-Temporal Data: A Survey and Outlookcited by 109
  7. 2024ChatTS: Aligning Time Series with LLMs via Synthetic Data for Enhanced Understanding and Reasoningcited by 95
  8. 2025Time-MQA: Time Series Multi-Task Question Answering with Context Enhancementcited by 74
  9. 2025V-STaR: Benchmarking Video-LLMs on Video Spatio-Temporal Reasoningcited by 66
  10. 2025MTBench: A Multimodal Time Series Benchmark for Temporal Reasoning and Question Answeringcited by 50
  11. 2024TimeSeriesExam: A time series understanding examcited by 47
  12. 2024Towards Time Series Reasoning with LLMscited by 36
  13. 2025ITFormer: Bridging Time Series and Natural Language for Multi-Modal QA with Large-Scale Multitask Datasetcited by 35
  14. 2025Time Series Forecasting via Reasoning: A Slow-Thinking Approach with Reinforcement Fine-Tuned LLMscited by 34
  15. 2024How Can Large Language Models Understand Spatial-Temporal Data?cited by 31
  16. 2025OpenTSLM: Time-Series Language Models for Reasoning over Multivariate Medical Text- and Time-Series Datacited by 30
  17. 2025CausalRivers -- Scaling up benchmarking of causal discovery for real-world time-seriescited by 24
  18. 2025TimeOmni-1: Incentivizing Complex Reasoning with Time Series in Large Language Modelscited by 24
  19. 2025Time-R1: Towards Comprehensive Temporal Reasoning in LLMscited by 23
  20. 2025Can Slow-thinking LLMs Reason Over Time? Empirical Studies in Time Series Forecastingcited by 19
  21. 2025TimeMaster: Training Time-Series Multimodal LLMs to Reason via Reinforcement Learningcited by 16
  22. 2024A Survey of Generative Techniques for Spatial-Temporal Data Miningcited by 14
  23. 2024UrbanLLM: Autonomous Urban Activity Planning and Management with Large Language Modelscited by 13
  24. 2025TimeDistill: Efficient Long-Term Time Series Forecasting with MLP via Cross-Architecture Distillationcited by 12
  25. 2025Inferring Events from Time Series using Language Modelscited by 10
  26. 2025Learning with Calibration: Exploring Test-Time Computing of Spatio-Temporal Forecastingcited by 9
  27. 2024Are Synthetic Time-series Data Really not as Good as Real Data?cited by 7
  28. 2025TempoGPT: Enhancing Time Series Reasoning via Quantizing Embeddingcited by 7
  29. 2025Time-RA: Towards Time Series Reasoning for Anomaly Diagnosis with LLM Feedbackcited by 7
  30. 2025CaTS-Bench: Can Language Models Describe Time Series?cited by 7
  31. 2025Beyond Chunks and Graphs: Retrieval-Augmented Generation through Triplet-Driven Thinkingcited by 6
  32. 2025Can Large Language Models Adequately Perform Symbolic Reasoning Over Time Series?cited by 6
  33. 2025U-Cast: Learning Hierarchical Structures for High-Dimensional Time Series Forecastingcited by 5
  34. 2025Exploring Accuracy Law for Deep Time Series Forecasters: An Empirical Studycited by 4
  35. 2025How Different from the Past? Spatio-Temporal Time Series Forecasting with Self-Supervised Deviation Learningcited by 4
  36. 2025Pre-training Epidemic Time Series Forecasters with Compartmental Prototypescited by 3
  37. 2025Towards Interpretable and Trustworthy Time Series Reasoning: A BlueSky Visioncited by 2
  38. 2024Ergodic Network Stochastic Differential Equationscited by 1

50 most cited of 52 All 52 (JSON)

Built on by

Cited by year (2026 to 7 Oct)

Citing papers per year: 2026: 7.

7 papers.

  1. 2026Mixture-of-Experts Knowledge Graph Retrieval-Augmented Generation for Multi-Agent LLM-based Recommendationcited by 2
  2. 2026STAR: Failure-Aware Markovian Routing for Multi-Agent Spatiotemporal Reasoningcited by 1
  3. 2026EpiEvolve: Self-Evolving Agents for Streaming Pandemic Forecasting under Regime Shiftscited by 1
  4. 2026Exposing Vulnerabilities in Explanation for Time Series Classifiers via Dual-Target Attackscited by 0
  5. 2026SURGE: An Event-Centric Social Media Sentiment Time Series Benchmark with Interaction Structurecited by 0
  6. 2026Learning Spatio-Temporal Foundation Models from Pure Synthetic Datacited by 0
  7. 2026TimeThink: Eliciting Compositional Reasoning in Timeseries Large Language Modelscited by 0

Similar, not cited

Close in wording, no citation either way

  1. 2025USTBench: Benchmarking and Dissecting Spatiotemporal Reasoning of LLMs as Urban Agents
  2. 2026MACRO-LLM: LLM-Empowered Multi-Agent Collaborative Reasoning under Spatiotemporal Partial Observability
  3. 2025A Survey of Reasoning and Agentic Systems in Time Series with Large Language Models
  4. 2026Time Series Reasoning via Process-Verifiable Thinking Data Synthesis and Scheduling for Tailored LLM Reasoning
  5. 2026Visual Reasoning over Time Series via Multi-Agent System
  6. 2025From Building Blocks to Planning: Multi-Step Spatial Reasoning in LLMs with Reinforcement Learning
  7. 2026CastFSR: A Fast--Slow--Reflect Agentic Reasoning Framework for Context-Aware Time Series Forecasting
  8. 2026TemporalBench: A Benchmark for Evaluating LLM-Based Agents on Contextual and Event-Informed Time Series Tasks