Background: The Evaluation of NegatedPropertySet in Section 18.5 Property Path Patterns defines the result of evaluating algebraic property path expressions of the form Path(x, NPS(S), y).
Issues: There are three issues in that definition:
- The definition does not restrict the resulting solution mappings in terms of their domain. As a consequence, the resulting solution mappings may have bindings for variables other than x and y (assuming that both x and y are variables). In fact, due to this issue, there can be an unbounded number of resulting solution mappings. The fix to this issue is to add the following condition for μ into the formula in the definition: dom(μ) = { v ∈ {x, y} | v is a variable }.
- The definition should be explicit about the multiplicity of the resulting solution mappings. Is the result of
ppeval(x, NPS(S), y) just a set of solution mappings? (This question is relevant because there may be multiple possible predicates p ∉ S (and, also, x or y may be a blank node).
- For cases in which x or y is a blank node, the definition implicitly relies on the assumption that the set of blank nodes in a query is disjoint from the set of blank nodes in the queried graph. It may be helpful to make this aspect of the definition explicit.
Background: The Evaluation of NegatedPropertySet in Section 18.5 Property Path Patterns defines the result of evaluating algebraic property path expressions of the form Path(x, NPS(S), y).
Issues: There are three issues in that definition:
ppeval(x, NPS(S), y) just a set of solution mappings? (This question is relevant because there may be multiple possible predicates p ∉ S (and, also, x or y may be a blank node).