When Does Text Inform? Benchmarking Information-Theoretic Metrics for Multimodal Time-Series Forecasting

arXiv:2609.11282v1 Announce Type: new Abstract: Multimodal forecasting models that combine time series with text annotations promise richer prediction through textual context, but how do we know whether a text annotation meaningfully contributes to the forecasters prediction? This is an information-theoretic question, but to evaluate whether information-theoretic metrics can reliably measure the predictive value an annotation provides, a ground truth benchmark is needed, and none currently exist. We create a synthetic time series signal with annotations in three categories: semantically correct, incorrect, and irrelevant. Because the data generation process is fully controlled, ground-truth information content is known exactly, enabling principled evaluation of six complementary mutual information estimators (KSG, MINE, InfoNCE, CCA, PID and V-information). We show that all six estimators identify correct annotations as most informative, and are able to audit the quality of mixed text corpora, choosing the annotations that result in the best downstream forecasting results without the need for model training. Our benchmark identifies limitations of each estimator, and these are validated on seven real-world datasets, which show how estimator performance differs on weak signals. Finally, we establish practical rules for implementing these metrics for annotation auditing and fusion selection.

Related

• NeST: Neighborhood-aware semantic alignment and temporal modulation for LLM based time series forecasting
• GITCO: Gated Inference-Time Context Optimization in TSFMs
• Prototypicality Bias Reveals Blindspots in Multimodal Evaluation Metrics
• A Structured Benchmark for Text-Guided Anomaly Detection: When Language Stops Conditioning the Decision

Source: arXiv cs.AI | 2026-09-12