It's 2010. Your company survived the economic downturn earlier in the millennium by smartly relocating product sub-assembly to Brazil, hiring third-parties to manage non-mission-critical IT functions and retraining your best production managers to oversee quality assurance. A service-oriented architecture (SOA) was also deployed to enable tighter integration of departmental and factory systems.
But true interoperability didn't come until two years ago when new Web-based technologies were deployed that, without the need for proprietary adapters or custom code, enabled in-house and business partner systems to interact with one another using common definitions of terms such as "customer" and "order."
Today, when your company's enterprise resource planning (ERP) system senses that an important component needed to assemble your highest-margin product is almost out of stock, it dispatches a software agent to a trio of approved suppliers' catalogs to find the most parts on the best possible terms. Once the part is located, your ERP system places an order directly in the chosen supplier's order-entry system, which sends a shipping invoice to your ERP accounts receivable module. The order ships the next day, arriving at your sub-assembler within 48 hours, where it is scanned and accounted for in your company's inventory management system (which was expecting the part following a communication from the procurements module). Payment is scheduled in the accounts payable module for 30 days out -- all without any human intervention.
Sound like nirvana? Maybe. But if leading Web researchers have anything to do with it, it will be reality. Web-based applications interoperability standards are in place and technologies are being tested to prove how this type of interoperable, hands-free computing can reduce everything from fat finger re-keying errors to human misinterpretation of data. For manufacturers, taking advantage of these technologies would bring significant -- even disruptive -- competitive advantages including better integrated and less costly supply chains as well as more seamless engineering, production and distribution processes.
Key to all of this is the Semantic Web, whose technological underpinnings have been in development since the mid 1990s when commercial Internet applications were still a gleam in the eyes of many business people. In its raw essence, the Semantic Web is an extension of the technologies that brought the World Wide Web to life -- the browser, the Internet Protocol for transport, extensible markup language (XML) and HTTP, which enables the linking of hypertext within Web pages. Throw in some artificial intelligence (rules-based constructs) and agent technology and, the theory goes, businesses can create applications that have the inherent ability to interpret meaning from, or even reason with, one another. To many, the Semantic Web -- conceived by Web progenitor Tim Berners-Lee -- is the next step in the Internet's evolution; one that manufacturers must prepare for or risk getting left behind.
Today, a great deal of human interpretation is required to understand much of the information conveyed as hypertext on Web pages such as product description, part numbers, pricing and shipping data. For businesses that want to extract meaning from online catalogs and automate business processes, today's Web presents "technical challenges [that] are quite staggering," notes Ora Lassila, a Boston-based research fellow with Nokia Research Center who, with Berners-Lee, co-authored the seminal article "The Semantic Web" in the May 2001 issue of Scientific American.
"Computers are bad at understanding things," adds Dr. Steve Ray, manufacturing systems integration division chief at the National Institute of Standards & Technology. "They need [semantic] information in computable form -- once you have that, the [Web] revolution seen in society will similarly impact business systems."
REALITY BYTES
"The Web was formed with no semantic tags, and there's a realization that HTML was not the best way to deliver business value," says Steve Ross-Talbot, chair of Web Service Activity at the Worldwide Web Consortium (W3C) and chief scientist at Enigmatec Corp. (London), a data center management software developer. XML separates "content from presentation, but the vendors have shown a complete inability to share equivalence between tags," he adds.
To overcome XML's shortcomings, the W3C has brought forth RDF (resource description framework), which enables developers, using tags, to attach unambiguous meaning to content using descriptive rules or logic that represents knowledge about things, properties and values. In addition, the Semantic Web community has standardized around the Web ontology language, or OWL, which formally defines the relationship between terms and concepts. By adopting ontologies that consistently describe business terminology (e.g. a product catalog, parts tolerance specifications, delivery schedules) in a database, business partners could enable their computers to autonomously negotiate all types of transactions.
Say, for example, that a manufacturer wanted to get better at forecasting demand, projecting revenues and enabling regional fulfillment. Using Semantic Web technologies, customer addresses in the company's ERP system would be associated with specific locations so that, when products were shipped to known locations, the company could understand in real time where demand was generated. Then, because it understood the meaning of those forecasts, a warehouse management system would be able to automatically tap this knowledge to figure out in which warehouse particular products should be stored to speed delivery.
No one argues with the concepts, as mind-numbing as they may be, but the adoption curve of the Semantic Web's underlying technologies has come at a glacial pace. The reason: minimal developer training and competency in the new technologies, a lack of industrial strength tools for building Web-based ontologies and resistance by developers to infuse applications with the ability to essentially think for themselves. Another hindrance: The Semantic Web is a leap of faith greater than the one required for manufacturers to expose application interfaces to business partners interested in creating Web services accessible outside the corporate firewall.
"We're just now starting to see early success stories," says Dr. James Hendler, a computer sciences professor at the University of Maryland (College Park, MD), the third co-author of the influential Scientific American article. He points to pilot projects at government agencies such as NASA and the Defense Advanced Research Projects Agency (DARPA) where the Semantic Web's constructs are being tested. The real win, he says, will come when enterprises begin to use the technologies to enhance interoperability among company applications. So far, however, few are charging forward.
"It's like the guy with the first telephone; he's not happy until someone else has one," Hendler says. "We're just getting past that chicken and egg problem -- but at least we have a few chickens and a few eggs out there."
BABY STEPS
Network Inference (Carlsbad, CA) is one company that has made headway primarily working with systems integrators to implement applications built on Semantic Web technologies. One showcase project involves a major West Coast electronics manufacturer, which has requested to remain anonymous. The company collects massive amounts of data on the markets it sells into and the different parts that go into the various products it makes, but wanted to better understand the contribution each component made to its financial results, notes Vickie Farrell, Network Inference's vice president of marketing. The problem: Ever-changing product definitions rendered the company's parts-classification scheme inadequate to make this calculation.
Network Inference provided semantic technologies based on RDF and OWL that were used to build product classification vocabularies so the company could understand the contribution each part made to its financial results. "Our technology makes sure those cross linkages are accurate and dynamic," she says.
"Faster reporting shortens decision cycles," notes Alexander Linden, research vice president of emerging technologies at Gartner Inc., who says this project illuminates the role semantic technologies can play in helping companies more cost-effectively compile flexible and accurate financial reports with "fewer errors and much less uncertainty."
SEMANTIC WEB & SOA
Hendler, like other Semantic Web researchers, believes the Semantic Web's utility will be unlocked when coupled with Web services and SOAs. Ontologies will enable applications built for various vendors' SOAs to more easily interoperate with one another, removing the need for proprietary tools. Once companies can start enabling the discovery of a service that represents a particular type of business activity (e.g. searching a supply catalog), which can then be connected with a bidding mechanism activated with the click of a single button, the Semantic Web's momentum will gather, says Hendler.
However, the desire by some enterprise applications vendors to lock out the competition could make Semantic Web adoption a perpetual work in progress. For instance, says Bill Regli, a computer science professor at Drexel University (Philadelphia), established CAD and PLM vendors like to retain control of the data generated by their systems. The Semantic Web, he continues could "liberate the data from proprietary formats," which would undo the PLM vendor's lock-in effect.
The fact that the Semantic Web's standards for RDF and OWL have been blessed by the W3C should help drive industry's embrace of these technologies, researchers say. NIST's involvement in the standards process, moreover, should help manufacturers feel more at ease working with them. For example, NIST is driving a standards community push to map ISO standards such as STEP (the Standard for the Exchange of Product Model Data) -- which describes how to represent and exchange digital information -- to OWL and RDF. Doing this will help manufacturers capture the Semantic Web's richness, says NIST's Ray.
But no matter what the interoperability advantages of the Semantic Web are, its adoption will be dictated by larger companies imposing rules of electronic engagement on smaller players -- as with RFID and EDI -- researchers contend.
The Semantic Web will eventually become reality, NIST's Ray believes. But as with most potentially disruptive technologies, only the most innovative companies will jump on board early. "It's only the very forward-looking companies investing in the Semantic Web," he says, noting that in some industries, such as chemicals, companies already have shared vocabularies. Where these companies need help is in communicating these vocabularies to other industries such as automotive (where NIST is providing guidance).
Smaller companies should get going with XML. "They can do well with XML with less chance of misunderstanding within their own community," Ray says. "It's if you do business globally or across industry sectors where you run into the trouble of misunderstanding."
Gartner's Linden believes the "Holy Grail" of interoperability, backed by emerging standards, will create the scale and cost efficiencies required to give the Semantic Web a fair marketplace hearing. "The Semantic Web is an important piece of the puzzle -- not the most important -- but critical to making things more cost effective," he concludes.