ChatTS: Aligning Time Series with LLMs via Synthetic Data for Enhanced Understanding and Reasoning

Summary: ChatTS is the first MLLM treating multivariate time series as a native modality for alignment and reasoning. Attribute-based synthetic series and Time Series Evol-Instruct enable exclusive synthetic-data fine-tuning, yielding strong gains over GPT-4o and LLM baselines. (summarized by gpt-5.6-luna on Jul 24 2026)

Paper ID: 14074
Venue: VLDB
Year: 2025
Pagerank: 5.093636e-05
Overall Rank: 10,868 | 25.44%
DOI: 10.14778/3742728.3742735

Incoming Non-self Citations Over Time

No non-self incoming citations found for this paper in this database.

Authors

1. Zhe Xie (Beijing Institute of Technology; Tsinghua University)
2. Zeyan Li (ByteDance)
3. Xiao He (ByteDance)
4. Longlong Xu (Beijing Institute of Technology; Tsinghua University)
5. Xidao Wen (BizSeer)
6. Tieying Zhang (ByteDance)
7. Jianjun Chen (ByteDance)
8. Rui Shi (ByteDance)
9. Dan Pei (Beijing Institute of Technology; Tsinghua University)

BibTeX Citation

@article{xie_vldb25,
        title = {{ChatTS: Aligning Time Series with LLMs via Synthetic Data for Enhanced Understanding and Reasoning}},
        author = {Xie, Zhe and Li, Zeyan and He, Xiao and Xu, Longlong and Wen, Xidao and Zhang, Tieying and Chen, Jianjun and Shi, Rui and Pei, Dan},
        journal = {PVLDB},
        series = {{VLDB} '25},
        volume = {18},
        number = {8},
        pages = {2385--2398},
        doi = {10.14778/3742728.3742735},
        url = {https://doi.org/10.14778/3742728.3742735},
        year = {2025}
}

Incoming Citations (Sorted by Pagerank)

Showing 0 of 0 citing papers.

Rank	Citing Paper	Year	Venue	Pagerank

Outgoing Citations (Sorted by Pagerank)

Showing 2 of 2 cited papers.

Citations counted here include only citations to other VLDB/SIGMOD/CIDR/PODS papers in this database.

Rank	Cited Paper	Year	Venue	Pagerank
1,920	D-Bot: Database Diagnosis System using Large Language Models	2024	VLDB	9.4846185e-05
8,517	OneShotSTL: One-Shot Seasonal-Trend Decomposition For Online Time Series Anomaly Detection And Forecasting	2023	VLDB	5.4119882e-05

Semantically Similar Papers

#	Overall Rank	Paper	Year	Venue
1	10,627	Revisiting Task-Oriented Dataset Search in the Era of Large Language Models: Challenges, Benchmark, and Solution	2026	VLDB
2	5,038	Time2Feat: Learning Interpretable Representations for Multivariate Time Series Clustering	2023	VLDB
3	11,314	A Demonstration of TENDS: Time Series Management System based on Model Selection	2024	VLDB
4	10,663	Dialogue Benchmark Generation from Knowledge Graphs with Cost-Effective Retrieval-Augmented LLMs	2025	SIGMOD
5	4,886	Chat2Data: An Interactive Data Analysis System with RAG, Vector Databases and LLMs	2024	VLDB
6	4,141	AutoAI-TS: AutoAI for Time Series Forecasting	2021	SIGMOD
7	8,346	TSGBench: Time Series Generation Benchmark	2024	VLDB
8	5,129	AutoCTS: Automated Correlated Time Series Forecasting	2022	VLDB
9	6,443	AutoCTS+: Joint Neural Architecture and Hyperparameter Search for Correlated Time Series Forecasting	2023	SIGMOD
10	7,871	TSGAssist: An Interactive Assistant Harnessing LLMs and RAG for Time Series Generation Recommendations and Benchmarking	2024	VLDB