Time-series databases (TSDBs) have become essential in various time-series data applications. As the size of time-series datasets continues to grow rapidly, efficient system solutions are needed to manage this data effectively. One promising approach is caching systems that reduce response times and alleviate traffic burdens on back-end TSDBs. However, conventional caching strategies are often suboptimal for handling time-series data. To address this challenge, we propose SEMECAF (SEmantic-aware and MEtadata-driven CAching Framework), a novel caching framework designed to enhance the efficiency of time-series data management. SEMECAF introduces three key innovations: a cache framework, a semantic-aware processing mechanism, and metadata-driven management. Experimental results show that a lightweight implementation of SEMECAF significantly outperforms the baseline approaches across various performance metrics.