example project: rule-based credential mapping

A niche interest of mine is the composition of rule systems, by which I mean processing pipelines that can take individually composed or graph-composed sets of rules and use them to validate complex logic, assign metadata, or make decisions. I have designed my own rule systems for several projects, usually written to be saved as JSON objects with various query directives supported (e.g., derived Mongo-like parameters like instructor.degrees: {$count: {$gte: 2}}, but extending to more complicated comparisons, with a decision-tree kind of branching to subrules).

The most significant project for which I worked out a robust rule system was credential mapping for reaccreditation. In that case, the rules could be easily expanded directly by stakeholders willing to write justifications for each one, and the final set fully described a sophisticated logic for relating all teaching areas and degrees.

In other cases, I have used a similar principle to bring a more logical validation and correction strategy to other areas of often-irregular institutional data, like the full set of circumstances that need to be met in order to cross-list courses. In many of these cases, the prior status quo from administrators on campus took the wildly inefficient approach of long, difficult to debug sql query runs, and I have always advocated for the advantage of clearly auditable rules that can show their sequences of tests and failures per object in the domain.