Fixes the definition of evaluation of property path with OneOrMorePath #350
Conversation
| where <var>V</var> is the set of RDF terms that is returned by the | ||
| following algorithm. | ||
|
|
||
| Let <var>X</var> = <a href="#defn_reachableTerms"><var>reachableTerms</var></a>(<var>x</var>, <var>ppe</var>) |
There was a problem hiding this comment.
Let reachableTerms(x:term, ppe) be the set of RDF terms
reached by repeated matches of ppe,
when starting at RDF term x.
"reacheable" to me means an arbitrary length match e.g. :p*
"Evaluation of OneOrMorePath" uses it as "one step".
I think the definition of reachableTerm is tying to be one step but "reachable", and "repeated matches", imply to me an arbitrary length match.
Maybe "all matches of the ppe" and maybe call it "pathMatches" or "oneStep"/"oneApplication"
"neighbors" might be possible but ppe is not just length 1.
There was a problem hiding this comment.
@afs I agree that this part is misleading, at best. In fact, I would say that the reachableTerms function is ill-defined. Notice, however, that this is not an issue of this PR, not even an issue of something we have done in v1.2 now. The function that is called 'reachableTerms' now was called 'eval' in 1.1 (indeed, like several other "eval" functions in 1.1)---see it being used in the definition of OneOrMorePath in 1.1 and its "definition" in the context of the ALP function in 1.1. I will open a separate issue to address this comment.
Co-authored-by: Andy Seaborne <[email protected]>
Closes #267 (and adds a note for #349)
Preview | Diff