Time Series Event Detection (TSED) aims to localize semantically meaningful events in time series data, with critical applications in high-stakes domains. Unlike statistical anomalies, events are often defined by natural-language descriptions with internal temporal-logic structures across multiple physical channels. However, in real-world settings, dense event annotations are expensive to obtain, making purely supervised learning difficult. We introduce Language-guided TSED, a setting where a model is given textual event descriptions and must ground them to intervals in multivariate signals with little or no labeled data. To address this problem, we propose Event Logic Tree (ELT), a knowledge representation framework that converts linguistic descriptions into structured temporal logic over signal primitives. Building on ELT, we present SELA, a neuro-symbolic VLM agent framework that iteratively grounds primitives from signal visualizations and composes them under ELT constraints, producing both event intervals and faithful tree-structured explanations. We further release a real-world benchmark across energy and climate domains with expert knowledge and annotations. Experiments show that SELA improves over supervised fine-tuning and existing zero/few-shot time series reasoning baselines.
Figures & tables
Figure 1: An illustration of the Language-guided Time Series Event Detection task.
Operator
Natural Language Description
Formalization
TS Instance
SEQ
“A spike in A is followed by a drop in B .” “A drop in B occurs after a spike in A .”
Table 2: Statistics by Dataset: PT = Pressure Test, CH = Climate Hazard. Lengths are event durations in samples; TS Sample is the series length. Ratio is the percentage of samples containing that class.
Table 4: Comparison of different models and methods on the two KITE datasets. Zs=Zero-shot; Fs=Few-shot In-Context Learning.
Pressure Test ( M -F1)
Climate Hazard ( M -Coverage)
GPT-5
GPT-4.1
Avg.
GPT-5
GPT-4.1
Avg.
Variant
@0.5
@0.9
@0.5
@0.9
F1
F0.5
F1
F0.5
SELA
66.8
60.0
43.9
32.7
50.9
50.7
57.1
52.2
53.9
53.5
Event-logic structure
w/o ELT (Geo. Mean)
41.8 ↓ 25.0
39.1 ↓ 20.9
41.5 ↓ 2.4
32.2 ↓ 0.5
38.7 ↓ 12.2
29.4 ↓ 21.3
34.5 ↓ 22.6
48.5 ↓ 3.7
49.7 ↓ 4.2
40.5 ↓ 13.0
w/o ELT (Arith. Mean)
41.8 ↓ 25.0
39.2 ↓ 20.8
42.9 ↓ 1.0
32.7 ↑ 0.0
39.2 ↓ 11.7
29.6 ↓ 21.1
34.7 ↓ 22.4
47.4 ↓ 4.8
48.6 ↓ 5.3
40.1 ↓ 13.4
Table 5: ELT ablations on Pressure Test and Climate Hazard. Avg. averages the four backbone–metric entries. Bold and underline denote the best and second-best results per column.
Figure 4: Case study from Climate Hazard, showing predictions from different methods against the ground truth, together with the instantiated primitive instances and Event Logic Tree (ELT).
D1: Hierarchical Representation
D2: Semantic Quantification
D3: Topological Elasticity
SAX / ABBA
✗
▲
✗
Logical-Shapelets
▲
▲
✓
Z-Time
▲
▲
✓
Chronicle / STL
✓
▲
✗
Event Logic Tree
✓
✓
✓
Table 6: Comparison of frameworks against our desiderata. ( ✓ : Satisfied; ▲ : Limited / Partially satisfied; ✗ : Not satisfied)
Appendix figures & tables1 asset
Supplementary material from the paper’s appendix.
Appendix
Zero-shot
Few-shot
Set
Method
Tokens
Cost
Tokens
Cost
PT
SELA (GPT-5)
233.6
0.623
-
-
SELA (GPT-4.1)
125.9
0.189
-
-
Num. (GPT-5)
33.6
0.081
91.9
0.143
Num. (GPT-4.1)
29.2
0.059
92.2
0.185
VL-T. (GPT-5)
5.1
0.033
6.7
0.031
Appendix
Table 7: Average per-sample token usage (K) and inference cost ($) for LLM-based methods. The slightly lower few-shot cost of VL-Time with GPT-5 on Pressure Test results from reduced output-token usage despite the longer input prompt.