I’m in a deep technological mood these days. As you probably noticed, I’m attending Semantic Technology & Business conferencein beautiful, but cold San Francisco. SemTech 2012 covers an interesting technological space that covers a variety of topics related to data, data management, big data, semantics, linked data and semantic web. So, the environment of the conference and some presentations made me think about some modern trends in data management related to data federation. It probably goes a bit beyond the technological level of this blog, but I found it interesting and insightful.
Distributed Data Architecture
Our world is getting more and more distributed. The time when you was able to concentrate the data in a single computer and/or databases almost became a history. We are moving towards something bigger that can scale to the level of web. The following two examples show a potential role of semantic technologies in support of federated data environment:
Andrew Sunderland of Spry Inc presented enterprise data management options. Here is the interesting quote explaining his presentation:
Companies are looking for methods to quickly expose data sources for federated data access, while at the same time developing a robust, executable enterprise ontology. Data profiling tools can be leveraged to profile data sources and bootstrap ontologies and mappings. This talk will showcase how Spry is leveraging these tools to quickly expose data sources, while in developing an enterprise ontology
Another example is coming from FluidOps – Transformation of Enterprise Data Islands into Linked and Living Knowledge. Information Workbench environment coming from FluidOps. The discussion focus was on the transformation of enterprise data islands into linked and living knowledge and elaborates on the costs and benefits of managing information in a unified semantic space.
The following picture shows Information Workbench architecture and the role of semantic technologies to achieve the role of data unification.
Data Federation and Asymmetric Computing
I had a chance to attend the presentation of Bryan Thompson of Systap discussing the bigdata® architecture. His presentation was focused on the computing side of distributed data environment and federation. The following slide presents the role of RDF and graph as a unified model for heterogeneous data sources.
How is that related to PLM?
Now, you can ask me- how it is related to PDM and Engineering and Manufacturing world. Here is my take. IT infrastructure of manufacturing companies is extremely complicated these days. It includes existing data management and enterprise systems, content and document management vaults, unmanaged files and other data sources. Nowadays, cloud and web are coming as an additional data places companies target for data. The overall environment is global and distributed. Existing PLM systems are striving towards centralization of data into a singe data. The single database architecture might be not sufficient, cost of data transition might be too high, cloud and globalization is another dimension of complexity. Distributed and federated data management capable to scale to the level of web – logically and physically can be an interesting platform option to discover.
What is my conclusion? The history gave us many examples when large companies missed new technological trends, and it cost them to lose their leadership position. At the same time, we can see how web companies built their infrastructure and disrupted many existing domains. What will be the technological foundation that can support challenges manufacturing and engineering companies are facing today? What will be the role of semantic data technologies in the future of these systems is a right question to ask these days.
Best, Oleg
Wasn’t ISO 15926 mentioned on the conference? That’s the real application of SW technologies to PLM world.
Victor, thanks for commenting! Can you elaborate a bit how do you think ISO 15926 is related to semantic technologies? It is mostly about data exchange standards, STEP, etc.. Thanks, Oleg
Oleg,
The relation is rather strict. The Part 8 of the standard specifies OWL serialisation of ISO 15926 data, which preserves (or slightly modifies) logical relationships of Parts 2 and 7, thus bringing it fully into the Semantic Web world. Part 3 draft, which is about geometry data, already contains OWL ontology, STEP is fully replaced by OWL now. And data exchange is architectured as SPARQL endpoint federation of Big Data sources.
ISO 15926 was broadly represented at Semantic Days 2012, you can find list of speakers at https://www.posccaesar.org/wiki/PCA/SemanticDays2012 (among them was TopQuadrant, whom you have probable met at SF).
Here are two most interesting presentations with a number of links between ISO 15926 and SW:
https://www.posccaesar.org/svn/pub/SemanticDays/2012/Presentations/May9/10_David_Price.pdf
https://www.posccaesar.org/svn/pub/SemanticDays/2012/Presentations/May10/05_Darius_Kanga.pdf
And here is mine, http://www.slideshare.net/vvagr/search-language-of-dot15926-at-semantic-days-2012 fully dedicated to the topic of SW data.
Oleg, I’ve answered your question but the comment disappeared somehow and I can not repeat it.
Victor,
thanks for your comment and information! It originally came to spam, because of heavy links.
I wasn’t aware about such a deep conversion from STEP to OWL. I don’t think it is reflected in what vendors support. Have you heard about any CAD vendor started to produce RDF/OWL instead of STEP?
In my view, the idea of conversion from STEP to RDF/OWL is an interesting. However, my concern here is mostly about how to make 15926 open. To use RDF/OWL as a language won’t solve the problem of openness in standards.
Just my thoughts… I’d be interested to discuss more.
Best, Oleg
Oleg,
To be more exact, STEP (geometry) data is not yet converted to OWL by anyone. Part 3 of the ISO 15926 contains very good mapping from STEP to OWL, but as far as I know only XMpLant team is preparing its tools to work with geometry in OWL.
All others are for a time quite satisfied with the abilities to transfer plant attribute data (piping, equipment, HVAC, civil eng., metal constructions) and basic topology for piping (PID diagrams).
Only Bentley (Open Plant family of products) have built-in button ”OWL export”, with other systems you have to rely on external adapters built mostly on iRING platform, or with XMpLant tools, or with our dot15926 tools.
Frankly speaking I don’t understand what is so close about ISO 15926? Open SPARQL endpoints (3 major ones), open source iRING tools, many educational and information resources on the web. Community itself is also quite open, although commercial sponsors of the JORD project are putting real money tn its development.
Victor, thanks for the information. In my view, openness defined by AAA semantic principle – Anyone can say Anything about Any topic. The core ideas of semantic web, RDF and OWL are built on this principle. At the moment of time, you introduce ISO (and any other standard) in the game, the limitation comes from the standard. Any standard is a result of “agreement”, and this is not open, in my view. It doesn’t matter if agreement is “ISO” or another group of people. To me, semantic web technologies are about how to solve the limitation. Just my thoughts… Oleg
Oleg,
The problem is in the Anything requirement. You need a shared conceptualization in the domain – that’s the main idea behind the ontology use. Therefore you have RDFS on top of RDF, OWL on top of RDFS, ISO 15926 on top of OWL, and even that is not the top of the stack. You have limitations on what you can say – but you have a benefit of understanding by outside people you have never talked with before.
We can see ISO 15926 used along with other more general or more specific ontologies, like in the IBM Reference Semantic Model. ISO 15926 is part of the solution to the problem you state,not part of the problem
Victor, I think you’ve raised a very important question related to ontology building. Yes, ontologies can be very helpful to define domain concepts. To have rich and meaningful ontologies is very important. However, even a very good ontology doesn’t bring a solution – software does. If software requires the existence of “shared” and “agreed” ontology, it locks all down to the agreement. And this is where “walled garden” begins, in my view. Just my thoughts… Best, Oleg