Hazardous chemicals in textiles represent a serious health issue. This is mainly due to missing data on the used chemicals and/or on their hazard, which prevents proper chemical risk assessment. Although identifying and filling these data gaps is crucial, the myriad chemicals used for textile production and multiple data sources make it extremely difficult to manually collect and process all the data. Here, we propose a machine learning-based approach to tackle this issue. First, we identify the relevant sources and data that can be analyzed with machine learning. Then, we propose knowledge graphs as a tool to organize and analyze the data. We finally provide specific examples and detail the expected outcomes of our approach.