| Question answer
|
#('''Jesse Weaver''') In section one, the …
#('''Jesse Weaver''') In section one, the authors state: 'In rule based reasoners, the OWL ontology definitions are first compiled into a set of rules.' I believe this is what you described as the 'latter approach' in your question. While the paper does not explicitly and unambiguously state how they are handling the rules, it is my assumption that they are doing exactly as you described. This does matter because their system only handles 'single-join rules' (rules having two subgoals that join on one variable), and it is assumed that such a compilation of the OWL ontology under OWL Horst semantics will result in only single-join rules. (I believe this is shown in a previous paper of theirs, cited as (8) in this paper.) Yes, this would possibly result in a large number of rules. However, that may actually be advantageous for the rule-partitioning approach, since they cite as a weakness of rule-partitioning the lack of number of rules to distribute among computational nodes.
#('''Jesse Weaver''') I, too, am concerned with their choice of data sets. You are correct in that they do not discuss the OWL expressivity of the datasets; it would be nice to know what kinds of rules were compiled from the respective ontologies. LUBM and UOBM are both synthetic datasets, and MDC is unknown to me. (The authors refer to MDC as their 'own data-set' in section six.) This is especially a concern when comparing domain-specific data partitioning with graph data partitioning. In figure five, the authors show that domain-specific data partitioning speeds up nearly as well as graph data partitioning, but this seems like it benefits from the lack of interrelations in LUBM data. If it were more interrelated/interconnected, then it seems like it would be more difficult to come up with a good domain-specific data partitioning. a good domain-specific data partitioning.
|
| Question asked
|
#authors are working within the OWL Horst …
#authors are working within the OWL Horst semantics and in section III-B they mention of Rule-base partitioning. I believe rules in OWL Horst are of the type {p a owl:TransitiveProperty, s1 p o1, o1 p o2 --> s1 p o2}. This rule can be re-written for each transitive property as {s1 tp o1, s2 tp o2 --> s1, tp, o2}, where tp is declared to be transitive in the KB. I couldn't clearly interpret the kind of rules the authors are referring. First of all does it matter? Although from the sole example they present it looks like they take the latter approach, and I believe rule base partitioning makes sense only in the latter approach. (Please correct me). Although latter approach means that rule base is comparatively huge because corresponding to each instance of the axiom they would have a rule.
#I am concerned of the choice of data sets. First of all authors do not mention anything about the OWL expressivity of LUBM, UOBM or MDC (UOBM uses a more expressive OWL). Secondly, LUBM (similarly UOBM) is a benchmark for traditional OWL reasoning. By nature the individual university data sets are totally unrelated, and therefore speed-up on data partitioning is not a surprise. I do not mean to say that real life data sets are any different. Do you think that it would have been helpful if authors could have referred to some study that discussed characteristics of real-life rdf graphs (owl data) in terms of connectedness/ partitions etc. or at least give some sense of real rdf graphs? least give some sense of real rdf graphs?
|
