Because different sources index the literature in different ways, performing complementary searches using multiple tools will provide a more comprehensive view of a particular area of science; as a result, the main strengths of Penn's collection of chemistry databases lies in its size and diversity. Like most other researchers, chemists can search the literature using keywords and phrases, but the chemistry databases also allow them to enter the primary journal literature through a search for specific chemicals, general substructures, and even partial or complete reactions. In addition to article indices, the chemistry database collection includes databases of chemicals and their properties, structural chemistry and crystallography databases, and databases of reactions, and many of the largest tools in the collection perform all of these functions and more. In recent years, Penn's reaction databases have expanded to include tools that help chemists determine the best pathways to make molecules; called "retrosynthesis tools," these databases employ artificial intelligence and machine learning strategies to propose effective synthesis routes based on the chemist's input parameters.