The how-to schemas (i.e., HowTo, HowToSection, HowToStep, HowToSupply, HowToTool) also have some interesting conceptual overlaps with: procedural knowledge, computer programming, and automated planning and scheduling.

So, in addition to (1) encoding alternatives, choices, and optionality in procedures, there are to consider: (2) encoding parallel or concurrent sequences of steps, and (3) encoding procedures which might branch, or vary, based upon conditions or upon observations or decisions to be made by those following the procedures.

With respect to Recipe and point 2, above, an example might resemble: "prepare the sauce while cooking the entree."

With respect to Recipe and point 3, above, an example might resemble: "bake the food at 350° for 20 minutes; if the crust is golden brown, at that point, remove it from the oven; otherwise, bake it for another 5 minutes."

Implementation might involve providing instances of HowToSection or HowToStep with @id identifiers (blank-node labels like _:n1) in order to reference them from any new schema types for: fork-points, join-points, or-points, condition-points (if/else; switch/case), observation-points, choice-points or decision-points, and any other kinds of data structures for branching or routing.

Some collections of alternatives, choices, or options could be, could use, or could map with the rdf:Alt type (the choose-1 collection).

In my opinion, it's worth considering both simple HowTo data, which might be capable of being aligned with linear instructional videos, and other more complex HowTo data, which might contain procedural branching points.

That is, some simple representations of a Recipe could be mapped to linear instructional videos containing illustrative presentations of food-preparation procedures. Perhaps hierarchical HowToSection and/or HowToStep instances could make use of something like the Media Fragments model to reference intervals of URL-addressable video resources, to reference video clips or excerpts of videos having both start and end times or both start times and durations.

Meanwhile, other more complex representations of a Recipe might involve uses of nonlinear video, might be more suitably mapped to hypertext document collections, or might be intended for utilization by AI systems for retrieving and conversationally, interactively presenting procedural knowledge.