The InChI has become a IUPAC-recognized standard for chemical structure representation, and it provides a compact way to uniquely describe a structure. The InChI is designed to be machine-readable for computer use (e.g., for use in structure databases) but it is also somewhat accessible to analysis by human chemists.
There is not a similar way to recognize a general reaction type. One way to define reactions is to use the Reaction InChI or RInChI, but this only covers specific reactions (e.g., conversion of 1-butanol to butanal) and it provides no information on the general reaction type (e.g., oxidation of a primary alcohol to an aldehyde).
Embedded concisely in the InChI is a great deal of structural information, organized in a consistent way. When one compares the differences in InChIs for related compounds, certain patterns emerge. If the atoms are “numbered” based on their relative position in the structure (rather than by the numbering algorithm of InChI), these patterns can become more consistent between different structures; they now begin to represent the transformation of one structure into the other in a more general way.
The goal of this project is to explore the scope of using InChI differences to represent reactions. If the scope is fairly broad, it will show that this approach may have certain applications that warrant further investigation. If the scope turns out to be very limited, and the problems are insurmountable, then students will still learn much about chemical structures and reactions, and how InChIs relate to these topics. It is also quite possible that the scope is limited, but that “workarounds” might be found that could open up this idea to broader use. For example, at present this approach cannot handle non-fixed hydrogens, but it is very possible that the student might suggest a way these could be incorporated.