The Semantic Days conference, held in Stavanger, Norway earlier this month continues as a flagship event for the industrial use of the World Wide Web consortium’s ‘semantic’ technology. Semantic technology, notably the resource data framework (RDF) was conceived a decade ago as a way of ‘exposing’ machine readable data in web pages and other resources in an application-neutral way. Subsequently, the technology has failed to live up to its initial promise.
So how are semantics faring in oil and gas—and in Norway in particular? The flagship POSC/Caesar Association’s (PCA) ISO 15926 data standard has embedded semantic technology in part (its ‘canonical’ version is in the legacy Express data modeling language) and has been embraced by the US Fiatech standards body as a vehicle for data exchange around capital projects such as oil and gas offshore production platforms and FPSOs. Bechtel’s Darius Kanga provided a status update on the PCA/Fiatech iRING project—a semantic framework that interoperates with multiple engineering source data formats. iRing has support from Bentley, CH2M Hill, Emerson, Hatch, Worley Parsons, Bechtel and others.
David Price (TopQuadrant) presented recent developments around the EPIM Reporting Hub (ERH), a consolidation of production and drilling data reporting for Norway’s oil and gas industry. ERH has leveraged ‘ontologies’ (lists of accepted terminology) derived from the regulator’s NPD Fact Pages and the PCA upstream ontology. Under the hood these are stored in an RDF triple store, AllegroGraph from Franz, Inc. The database is expected to grow to 300 million triples over the next 4 years.
But neither the data providers nor the recipients are as yet tooled-up for semantic delivery or consumption. Incoming data is in XML from a variety of legacy formats. Delivery is in more XML and PDFs generated from the triple store using the Sparql Inference Notation (Spin). Price sees the NPD Facts as a candidate for ‘linked data,’ an open data sharing initiative that is being promoted by W3C guru Tim Berners-Lee. For the moment, data is delivered over the very ‘closed’ Norwegian SOIL network.
Peter Haase’s company, Fluid Operations, has been working with Statoil on scalable end user access to ‘big data’ defined as ‘data sets so large that they are awkward to work with database management tools.’ Production and exploration data falls into this category—with petabytes of relational data described with diverse schema and spread over multiple individual data bases. For Statoil’s 900 expert users, this can mean up to four days to develop and run a new data access query—requiring help from IT.
The ‘Optique’ project is investigating the use of ‘linked, open data’ as a solution to such issues. Using the RDF data model and Sparql for querying, data that is scattered across information silos can be linked into a ‘web of data ontologies.’ Optique components include real time stream processing and scalable query with ‘elastic clouds.’
Emanuele Della Valle of the Milan Polytechnic observed that RDF was not really designed for stream processing. Valle is involved in the EU-backed Large Knowledge Collider, LarKC and is advocating a paradigm shift from ‘one time’ semantics to transient data that is to be consumed on the fly and leveraged in ‘stream reasoning.’ Here, data streams are unbounded sequences of time-varying data elements such as weather observations or oil and gas production data. Valle is working on ‘RDF Streams’ and SPARQL extensions for querying these. Applications overlap with current stream/complex event processing solutions. A streaming linked data framework is being prototyped. Here sensor networks provide data as ‘pairs’ where each pair is made of an RDF triple and a timestamp—i.e. they are neither ‘pairs’ nor ‘triples!’ More from Streamreasoning.
Other contributions of note include Siemens’ Mikhail Roshchin on reasoning with real time data and Kongsberg’s Kaare Finbak (and others) on Statoil’s Integrated environmental monitoring project. These and other presentations are available from the Semantic Days website.
© Oil IT Journal - all rights reserved.