This study presents an automated approach to identify and classify relevant chemical compounds within patent documents, focusing on the chemical compounds essential to the patent’s core inventions. Combining dictionary-based and morphology-based recognition tools, the system extracts chemical entities and classifies their relevance based on context, achieving an F-score of 86% for compound recognition. This method significantly reduces data size by filtering irrelevant compounds, improving patent database utility. The system’s effectiveness was tested on a specialized corpus and aims to enhance the efficiency of databases like Reaxys, enabling faster access to key compounds for scientific and commercial research.