Thursday, March 16, 2017

Working Towards an Industrial Ontology Foundry to Facilitate Interoperability

By KC Morris and Serm Kulvatunyou, Systems Integration Division, NIST

Ontologies for industrial problems have been a topic of research for several years.  Recently they are also appearing in activities by consortia groups working on standards.  These efforts, though overlapping, have been disjoint.  Wouldn’t it be nice to have a single place, a foundry, to go to find a community-vetted ontology suitable for your industrial needs or at least as a starting point for your work?  That was exactly the topic of the workshop at the National Institute of Standards and Technology (NIST) last December when a group of industry, government, and academic researchers met to discuss the possibilities for such a place. 

Per Wikipedia, Ontology in Information Science is “a formal naming and definition of the types, properties, and interrelationships of the entities that exist for a particular domain of discourse.” Generally, ontologies serve the purpose of mediating the semantics of information between different applications.  An ontology compartmentalizes the variables needed for computations and establishes the relationships between them. It is a requirement for problem solving and analytics in many domains. Nearly all projects in the European Union’s (EU) Horizon 2020 Factories of the Future program have adopted ontology as a component.  Similarly, in the US NIST’s smart manufacturing projects also have ontology as a component, and over the last two decades, NIST has developed several ontologies in manufacturing and engineering domains.

Existing industrial ontologies serve various overlapping objectives. Objectives should be coordinated but loosely related to avoid bottlenecks on independent parallel projects.  If the commonality between the ontologies could be aligned, through something like an ontology foundry, long term interoperability between the different engineering, manufacturing, and supply chain disciplines would be better served.  Perhaps more importantly, the development of new ontologies would become both easier and the results would be more compatible with each other.  Value would be gained through a better ability to leverage the work of others.  The effort of developing an ontology for new areas would be reduced through the ability to add to already existing work rather than starting anew.  New ontologies would become more robust and reliable by building on existing foundations.

With this idea in mind, NIST organized a workshop to explore the idea of a framework for curating ontologies—an Industrial Ontologies Foundry (IOF).  The goal for the workshop was to identify industry needs for an IOF, to develop consensus around the idea and what form the IOF might take, and to identify the issues that need to be addressed to move the activity forward.  The workshop with consensus that the IOF idea would be beneficial.  Several participants at the workshop expressed their business drivers, including a very motivational keynote talk by Dominique Florack, President, Research and Development of Dassault Systèmes.  While numerous challenges were identified, two action groups were formed to begin to address the most challenging of those:  1) the development of scenario-driven demonstrations to show the value of an IOF and 2) formulation for a business and governance model.  Follow up virtual meetings are ongoing to work on these issues.

In preparation for the workshop, participants contributed and reviewed more than 25 existing manufacturing-related ontologies to see what lessons could be gained from these.  (Other industrial-related ontologies exist and some were presented at the workshop.)  The reviewed ontologies fell into the following categories:  manufacturing capability modeling, product engineering, supply chain, materials data, manufacturing engineering, operations, and “other”.  An analysis of these ontologies showed two different approaches to ontology development in the area.  The first approach is to formalize the reality.  The focus of these efforts is to fully capture “reality” and tend to be very large in scope.  Such efforts are often the product of visionaries or academic research.  While this approach has the potential for broad impact, many barriers to adoption exist.  The other approach is to engineer the ontology to the need.  These ontologies have a much narrower scope in that they address a specific business need and focus on improving a particular capability, functionality or quality.  These ontologies are much more readily adopted since they typically have a customer in mind during development; however, they may be much less reusable and the result is a single point solution. 

Participants also reported on the reasons that they are interested in seeing an industrial ontology foundry.  The most reported need was to address interoperability including tool interoperability, data interoperability, and interoperability between supply chain partners and their applications.  Other drivers were information linking, formalization of requirements through information constraints, incorporation of business process aspects, and quality and traceability considerations.  Capabilities to address these needs seem achievable and ontologies would be a good technical approach to addressing them. 

The success of other semantics-based projects suggests that an Industrial Ontologies Foundry would be both achievable and useful.  The Open Biomedical Ontologies (OBO) Foundry is one example.  The OBO ontologies have demonstrated significant value in supporting the sharing of biomedical data and knowledge.  In the building and construction industry the Building Information Model (BIM) with the supporting Industry Foundation Classes (IFCs) is another example of a widely adopted semantic model which brought significant value to the industry and have enabled much of the building automation that is becoming available today.  If these successes could be replicated for industry-related ontologies, the impact would be greater interoperability and better business intelligence.  

If this is an idea that you find intriguing and think it could help your business, now is your chance to get in on the ground floor of its development.  The project will need broad and diverse support.  It should enable easier integration of industrial data and better coordination of the activities across relevant industrial systems.  The Industrial Ontologies Foundry will be the focus of one of the parallel sessions at the upcoming NIST/OAGi workshop on Enabling Composable Service-Oriented Manufacturing Systems April 10 & 11.  Come learn more, express your opinions and offer your insights as to what could make it succeed!  If you are interested in participating in the workshop, please contact one of the session chairs:  Dimitris Kiritsis, EPFL, Switzerland, dimitris.kiritsis@epfl.ch or Paul Witherell, NIST, USA, paul.witherell@nist.gov.  If you don’t want to wait for the workshop and would like to jump right in and join the weekly conference calls, please contact Serm Kulvatunyou, NIST, USA, serm@nist.gov.

Related articles from prior workshop:
Boonserm Kulvatunyou, Nenad Ivezic, K.C. Morris, Simon Frechette: Drilling down on Smart Manufacturing – enabling composable apps. Manufacturing Letters 08/2016; 10., DOI:10.1016/j.mfglet.2016.08.004
Nenad Ivezic, Boonserm Kulvatunyou, Yan Lu, Yunsu Lee, Jaehun Lee, Albert W. Jones, Simon P. Frechette: OAGi/NIST Workshop on Open Cloud Architecture for Smart Manufacturing. Report number: NISTIR 8124, Affiliation: NIST, http://dx.doi.org/10.6028/NIST.IR.8124




No comments: