We derived and implemented the linear classification algorithm Olfactory Weighted Sum (OWSum) to predict molecules’ odors. The presented approach relies solely on structural patterns of the molecules as features for algorithmic treatment. In addition to the prediction of molecular odor, OWSum provides insights into properties of the dataset and allows to understand how algorithmic classifications are reached. OWSum achieves an accuracy of 90.5% on our database consisting of molecules belonging to six different odor classes. Using OWSum, we can quantitatively assign structural patterns to odors and identify the most important ones, giving chemists an intuitive understanding of underlying interactions. To deal with ambiguities of the natural language used to describe odor, we introduce a descriptor overlap as a metric for the quantification of semantic overlap between descriptors. This allows for grouping of descriptors and derivation of higher-level descriptors.