<ul><li>Existing methods for rating time series data quality have shown promise within individual domains, but struggle with diverse data sets spanning different properties.</li><li>TSRating is introduced as a novel framework for rating time series data from various domains, leveraging the knowledge of Large Language Models (LLMs) for quality assessment.</li><li>TSRating utilizes LLMs to compare quality differences in diverse time series data and employs a rating model named TSRater to predict quality efficiently for future samples.</li><li>Through meta-learning and signSGD for inner-loop updates, TSRating demonstrates improved accuracy, efficiency, and adaptability across different domains based on experimental results on benchmark datasets.</li></ul>

TSRating: Rating Quality of Diverse Time Series Data by Meta-learning from LLM Judgment

Discover more