Semantic Space Informed Prompt Learning with LLM for Time Series Forecasting

The author proposes the S2IP-LLM framework to align semantic space with time series embeddings, enhancing forecasting performance through prompt learning.