STELLA: Guiding Large Language Models for Time Series Forecasting with Semantic Abstractions

2025-12-04

AI Summary

Current large language models (LLMs) struggle to effectively model dynamic temporal patterns in time series forecasting, and existing prompting strategies lack both global corpus-level and instance-level semantic guidance. To address these limitations, we propose STELLA—a novel framework that first performs dynamic semantic abstraction to decompose raw time series into trend, seasonality, and residual components. It then constructs hierarchical semantic anchors by jointly integrating corpus-level priors and fine-grained behavioral prompts as prefix tokens, enabling zero-shot or few-shot LLM inference without parameter tuning. STELLA innovatively unifies time series decomposition, semantic feature extraction, and hierarchical prompt engineering into a co-optimized pipeline. Extensive experiments across eight benchmark datasets demonstrate that STELLA achieves state-of-the-art performance on both long-term and short-term forecasting tasks, with strong generalization capability. Ablation studies confirm the critical role of semantic anchors in performance gains.

Technology Category

• Machine Learning: Large Multimodal Models (LMMs)
• Natural Language Processing: (Large) Language Models
• Planning, Routing, and Scheduling: Planning with Language Models

Application Category

• Semantics and Knowledge: Data modeling to support human-machine intelligence, including LLMs agents, intelligent system behavior, explanations, and user-friendly interactions
• Search and Retrieval-Augmented AI: Search Tool Learning with LLM: Teaching LLMs to invoke search and make use of retrieved information
• User Modeling, Personalization and Recommendation: Large Language Models (LLM) for user modeling and recommendation

Abstract

Recent adaptations of Large Language Models (LLMs) for time series forecasting often fail to effectively enhance information for raw series, leaving LLM reasoning capabilities underutilized. Existing prompting strategies rely on static correlations rather than generative interpretations of dynamic behavior, lacking critical global and instance-specific context. To address this, we propose STELLA (Semantic-Temporal Alignment with Language Abstractions), a framework that systematically mines and injects structured supplementary and complementary information. STELLA employs a dynamic semantic abstraction mechanism that decouples input series into trend, seasonality, and residual components. It then translates intrinsic behavioral features of these components into Hierarchical Semantic Anchors: a Corpus-level Semantic Prior (CSP) for global context and a Fine-grained Behavioral Prompt (FBP) for instance-level patterns. Using these anchors as prefix-prompts, STELLA guides the LLM to model intrinsic dynamics. Experiments on eight benchmark datasets demonstrate that STELLA outperforms state-of-the-art methods in long- and short-term forecasting, showing superior generalization in zero-shot and few-shot settings. Ablation studies further validate the effectiveness of our dynamically generated semantic anchors.

Problem

• Enhance LLM reasoning for time series forecasting
• Inject structured semantic and temporal context
• Guide LLMs with dynamic behavioral abstractions

Innovation

• Dynamic semantic abstraction decouples series into components
• Hierarchical Semantic Anchors provide global and instance-level context
• Prefix-prompts guide LLM to model intrinsic time series dynamics