[dotNetRDF-Develop] Unexpected SPARQL Query Results

[Fedja A. <ad...@ad...>]

Hello dotNetRDF Team,

I am using the current dotNetRDF version from NuGet (1.0.7.3471) and am 
experiencing something that appears to be a bug in SPARQL Query 
evaluation. Consider the following RDF Database:

[ Link ] <http://pastebin.com/2aYNdZ7j>

It describes the relationship between sheep and wolves in the context of 
a fairytale. There are no other data sources, these are all existing 
triples, a total of seven.

I am querying this tiny database using SparqlQueryParser, 
InMemoryDataset and LeviathanQueryProcessor. Now consider the following 
queries:

[ Query A ] <http://pastebin.com/T73SVdBY> retrieves the subject URIs of 
everything that is a Sheep. As expected with this dataset, the result is 
only Frame:Sheep itself, but it could be specific kinds of sheep in 
different datasets.

[ Query B ] <http://pastebin.com/PMKmgUME> retrieves the subject URIs of 
everything that a Wolf preys upon. Since a Wolf only has a PredatorOf 
relationship to Sheep, the expected result would still be only the Sheep 
URI. Instead, however, the resultset contains /both/ Sheep /and/ Wolf. 
In fact, the value of ?prey is, as expected, Sheep in both cases, *but 
**?obj**is Sheep in one of the results, and Wolf in the other*.

The first line of the WHERE condition clearly states that ?obj is 
required to be a ?prey, which is correct when both are Sheep (because a 
Sheep is a Sheep), *but a Wolf is certainly not a Sheep*. Both are 
animals, but that relationship goes the other way around.

I have verified the above queries with the above database in Protégé 
<http://protege.stanford.edu/> and get the expected results, which leads 
me to the conclusion that this is in fact a bug in dotNetRDF.

Can I expect a fixed NuGet package anytime soon? Is there a workaround?

Regards,
Adam