دانلود رایگان مقاله انگلیسی ارزیابی عملکرد پایگاه داده های نمودار RDF برای خدمات شهر هوشمند - الزویر 2018

abstract

Smart cities are providing advanced services aggregating and exploiting data from different sources. Cities collect static data such as road graphs, service description, as well as dynamic/real time data like weather forecast, traffic sensors, bus positions, city sensors, events, emergency data, flows, etc. RDF stores may be used to set up knowledge bases integrating heterogeneous information for web and mobile applications to use the data for new advanced services to citizens and city administrators, thus exploiting inferential capabilities, temporal and spatial reasoning, and text indexing. In this paper, the needs and constraints for RDF stores to be used for smart cities services, together with the currently available RDF stores are evaluated. The assessment model allows a full understanding of whether an RDF store is suitable to be used as a basis for Smart City modeling and applications. The RDF assessment model is also supported by a benchmark which extends available RDF store benchmarks at the state the art. The comparison of the RDF stores has been applied on a number of well-known RDF stores as Virtuoso, GraphDB (former OWLIM), Oracle, StarDog, and many others. The paper also reports the adoption of the proposed Smart City RDF Benchmark on the basis of Florence Smart City model, data sets and tools accessible as Km4City Http://www.Km4City.org, and adopted in the European Commission international smart city projects named RESOLUTE H2020, REPLICATE H2020, and in Sii-Mobility National Smart City project in Italy.

7. Conclusions

The usage of RDF stores to store smart city data is becoming of wide interest for several applications. In this paper we have proposed a Smart City RDF Assessment Model for a comparative study about the state of the art on RDF stores according to their main features and in particular on the SPARQL aspects/features. In addition, the Smart City RDF Benchmark has been proposed. The benchmark is based on (i) some datasets of triples (that are grounded on Km4City ontological model) accessible from http://www.disit. org/smartcityrdfbenchmark, it can be used only for benchmarking purpose; (ii) a set of SPARQL queries declined for different SPARQL constructs. The benchmark has been defined for smart city services to compare results which can be obtained by using different RDF Stores. In the benchmark, particular emphasis has been given to geo-spatial and full text searches, since such aspects have been only partially considered and addressed by the general state of the art benchmarks such as LUBM and BSBM. As a general remark, the produced benchmark can be profitably used in several other contexts where similar aspects are modeled.

The comparison addressed a number of well-known RDF stores such as Virtuoso, GraphDB, StarDog, and Oracle for the performance aspects. As a general consideration about performance, it should be noted that Virtuoso performs better in presence of less selective queries, thus providing a higher number of results. On the contrary, GraphDB performs better when specific results are searched, thus when a smaller number of results are requested. As to Virtuoso, some small problems have been detected. For example, with the st_intersect function that, in order to get results, constrained us to rewrite queries using st_distance function. It seems that in this case the spatial indexing structure is not used and the optimizer does not exploit it as the starting point for a selection.