Accuracy and utility of spectral analysis has plateaued, but techniques are still are not good enough for many environmental forensics applications (such as microplastic analysis). The primary limitations are a lack of reference data on material types in the environment and the complexity of correctly analyzing spectra. We propose an open source software, reference library, and spectroscopy community to address these problems by leveraging Semi-Supervised Artificial Intelligence Techniques on shared reference data.