Citation Context

...ClassOf, w)[n] (w, subClassOf, u)[m] (v, subClassOf, u)[n ⊗ m] f-rdfs12 (v, type, ContainerMembershipProperty)[n] (v, subPropertyOf, member)[n] f-rdfs13 (v, type, Datatype)[n] (v, subClassOf, Literal)=-=[1]-=- f-rdfp1 (p, type, FunctionalProperty)[n] (u, p, v)[m] (u, p, w)[l] (v, sameAs, w)[l ⊗ m ⊗ n] f-rdfp2 (p, type, InverseFunctionalProperty)[n] (u, p, w)[m] (v, p, w)[l] (u, sameA, v)[l ⊗ m ⊗ n] f-rdfp3...

Citation Context

...mory. 4.1 Datasets Since there is no real fuzzy RDF data available, we generate synthesis fuzzy ontology, called fpdLUBM 3 , for experimental purpose. Our system is based on a fuzzy extension of LUBM =-=[2]-=-, called fLUBM, which is used for testing querying 1 http://hadoop.apache.org/ 2 http://hadoop.apache.org/hdfs/ 3 Available at http://apex.sjtu.edu.cn/apex wiki/fuzzypdability under fuzzy DL-Lite sem...

Citation Context

... major representation standards for the Semantic Web. RDF Schema (RDFS) is used to describe vocabularies used in RDF descriptions. However, RDF and RDFS only provide a very limited expressiveness. In =-=[3]-=-, a subset of Ontology Web Language (OWL) vocabulary (e.g., owl:sameAs) was introduced, which extends the RDFS semantics to the pD ∗ fragment of OWL. Unlike the standard OWL (DL or Full) semantics whi...

Citation Context

...l. [5] decompose and merge RDF molecules using MapReduce and Hadoop. They perform SPARQL queries on the data but performance is reported over a dataset of limited size (70,000 triples). Urbani et al. =-=[13]-=- develop the MapReduce algorithms for materializing RDFS inference results. In [12], they further extend their methods to handle OWL pD ∗ fragment, and conduct experiment over a dataset containing 100...

Citation Context

...echnique for parallel OWL inferencing through data partitioning. Experimental results show good speedup but only on very small datasets (1M triples) and runtime is not reported. In Weaver and Hendler =-=[14]-=-, straightforward parallel RDFS reasoning on a cluster is presented. But this approach splits the input to independent partitions. Thus this approach is only applicable for simple logics, e.g. RDFS wi...

Citation Context

...oning algorithm for the new semantics. Since fuzzy pD ∗ semantics is targeted to handle large scale semantic data, it is critical to provide a scalable reasoning algorithm for it. Earlier works (e.g. =-=[12]-=-) have proved that MapReduce is a very efficient framework to handle the computation of the closure containing up to 100 billion triples under pD∗ semantics. One may wonder if it is helpful to scalabl...

Citation Context

...kes into account of fuzzy information. Schlicht and Stuckenschmidt [8] show peer-to-peer reasoning for the expressive ALC logic but focusing on distribution rather than performance. Soma and Prasanna =-=[9]-=- present a technique for parallel OWL inferencing through data partitioning. Experimental results show good speedup but only on very small datasets (1M triples) and runtime is not reported. In Weaver ...

Citation Context

...vocabularies,150 Time (minutes) 120 90 60 30 0 1/128 1/64 1/32 1/16 Inverse of the number of units Fig. 1. Time versus inverse of number of mappers such as TransitiveProperty and SameAs. In [11] and =-=[6]-=-, a more general framework called annotated RDF to represent annotation for RDF data and a query language called AnQL were proposed. As far as we know, this is the first work on applying the MapReduce...

Citation Context

...eful OWL vocabularies,150 Time (minutes) 120 90 60 30 0 1/128 1/64 1/32 1/16 Inverse of the number of units Fig. 1. Time versus inverse of number of mappers such as TransitiveProperty and SameAs. In =-=[11]-=- and [6], a more general framework called annotated RDF to represent annotation for RDF data and a query language called AnQL were proposed. As far as we know, this is the first work on applying the M...

Citation Context

...LUBM, which is used for testing querying 1 http://hadoop.apache.org/ 2 http://hadoop.apache.org/hdfs/ 3 Available at http://apex.sjtu.edu.cn/apex wiki/fuzzypdability under fuzzy DL-Lite semantics in =-=[7]-=-. The fLUBM dataset adds two fuzzy classes, called Busy and Famous. The fuzzy degrees of how an individual belongs to these classes are generated according to the number of courses taught or taken by ...

Citation Context

...pproach splits the input to independent partitions. Thus this approach is only applicable for simple logics, e.g. RDFS without extending the RDFS schema, where the input is independent. Newman et al. =-=[5]-=- decompose and merge RDF molecules using MapReduce and Hadoop. They perform SPARQL queries on the data but performance is reported over a dataset of limited size (70,000 triples). Urbani et al. [13] d...

Citation Context

...me should be inversely proportional to the speed, after eliminating the effect of the platform overhead, the system’s performance speeds up linearly to the increase of number of units. 5 Related Work =-=[10]-=- is the first work to extend RDFS with fuzzy vagueness. In [4], we further propose the fuzzy pD ∗ semantics which allows some useful OWL vocabularies,150 Time (minutes) 120 90 60 30 0 1/128 1/64 1/32...

Citation Context

...em over large scale fuzzy pD ∗ ontologies. We briefly discuss some related work on scalable reasoning in OWL and RDF. None of them takes into account of fuzzy information. Schlicht and Stuckenschmidt =-=[8]-=- show peer-to-peer reasoning for the expressive ALC logic but focusing on distribution rather than performance. Soma and Prasanna [9] present a technique for parallel OWL inferencing through data part...

Citation Context

...with degree at least 0.8. However, Fuzzy RDF (or fuzzy RDFS) has limited expressive power to represent information in some real life applications of ontologies, such as biomedicine and multimedia. In =-=[4]-=-, we extended the OWL pD ∗ fragment with fuzzy semantics to provide more expressive power than fuzzy RDF(S). In that work, we focused on some theoretical problems, such as the complexity issues, witho...