The opening keynote sessions at SemTech this week made one thing abundantly clear: Semantic technology is for business, and it’s time to start putting it in practice there.
“Semantics is a game changer in the B-to-C model,” said Bill Guinn, CTO Product Enablers, Amdocs Product Business Unit. The company’s focus is on delivering customer care and experience systems in the telco space, but Guinn’s address was centered on “applying semantics in any situation that involves complex and recurring relationships between business and consumers,” with the aim of improving revenue, reducing costs, and retaining customers.
For Amdocs, the focus was on intelligent decision automation at the operational level. Think of it in one respect as a way past the complications of consumers aborting those long-winded IVR (Interactive Voice Response) activities and winding up in the hands of customer service reps who then have no context of why people are calling. That leads to their having to ask a bunch of questions to figure that out, and then probablyre-routing them to a more qualified agent to resolve the issue. It’s all a recipe for customer dissatisfaction. At the same time, by using semantics to make intelligence available to applications, the door opens to better things. That includes proactively addressing problems before the customer churns; resolving issues more efficiently and so more cost-effectively; and even reaching out into consumers’ social spheres to influence their connections. Know what consumers like right now in their shopping carts, and how great an influencer they are within their circles, and target to that -- doing so means you have an 11 times greater chance of making two additional sales, he said.
Amdocs was able to create a new semantic platform to enable this without ripping and replacing the 60 million lines of code in its portfolio. “No one can afford to eviscerate a bunch of applications to put intelligence in them,” Guinn said. Instead, its platform subscribes to transactions and events that make up the customer experience, and stores data into a triple store to track that experience. Several billion triples a day and 24 by 7 operations are its requirements. It used Franz’ AllegroGraph to create a customer ontology that goes beyond tracking transactions to take meaning from them, and to infer what each transaction means in isolation and in combination with the overall experience customers have. That, in combination with its historical view, gives context to determine appropriate action to take with customers.
That could mean, for example, to send them a video explaining why their bill is higher than usual for a particular month, with tips on how to reduce that – before they call up to inquire or complain about it. Or it could mean to smarten the IVR system so that it knows enough to ask first thing if they are calling because their bill is higher than usual, and then immediately gets them to the right agent. It could also mean that a consumer's dropped calls and complaints become a trigger to indicate in real-time their unhappiness and to do something about it immediately, rather than waiting to review data at month's end,by which time it may be too late.
“So the view of the customer had to supercede past stuff. It’s not just knowing what you did with a customer but what that meant,” Guinn said. “Hence, semantics.” Missed payments, for instance, could mean a customer is unhappy with your service, or it could mean they are financially burdened, and the action you take will differ tremendously depending on which it is.
The Business Matters
But probably the most important aspect of making any of this possible is getting the business to be a part of the process. “The key is being able to model the business and what is important to the business in the ontology," he said. "Let the business establish this, define the concepts in the ontology, and manage the event-condition-action paradigm.” And that means Guinn doesn’t talk to his customers about ontologies, reasoners and OWL, “but about having a history of the customer experience, about giving a holistic customer experience. It’s not the language of semantics.” Guinn says he just throws that word in once or twice, “so they realize this is different and not based on the old technologies they are used to, and so they see it is innovative.”
Over at the BBC, where John O’Donovan led the development of the World Cup site that was a major implementation of Semantic Web technologies, it was also important that the business users – the journalists themselves – understood that there was a value to them of taking this approach to producing and managing content. The site had a large number of pages – for each team, each player, each match, including BBC-produced and non-BBC produced content – that would have been impossible to manage without automation, he said. The trick was in separating content and metadata; in creating content assets, managing them as objects, using ontologies to bring them together, and outputting from there. “Content abstraction is a big part of this as it makes it easier to apply semantics and RDF to content vs. using huge numbers of dropdowns and taxonomies to apply manually,” he said. “If the metadata store has all the information of where content is and how it is being managed, that is all that matters.”
This did require a change in the workflow journalists were accustomed to. But once the advantages were explained to them – that they could have a site with 800 detailed pages to account for every player, build content aggregations on the fly, and more – they were on board. “They immediately grasped the workflow and understood it. But there is a workflow change implied—you do more metadata upfront to save a lot of effort later,” said O’Donovan, who now is Director of Architecture and Development at the Press Association. Working in his favor is that journalists understand in principle the value of linking people, places and organizations to stories, “so they just get it. And that maps directly through to how you implement it and that is a compelling thing.”
But for any organization that has to deal with content, he recommends that the sell approach to the business be this: “The key to selling this is that if you value content you will manage it semantically because it is the best way to do it.”
The final keynote presenter at Tuesday’s session, Dennis Wisnosky CTO and Chief Architect, Business Mission Area, U.S. Department of Defense, put this spin on why IT needs to adopt semantics for the sake of the business. The DoD's Enterprise Information Web (EIW) project is a pioneering adoption of semantic technology and approaches, setting a new path for enterprise business intelligence and solution architectures at the DoD.
Wisnosky says each major wave of technology has about a 40-year lifespan, and relational database management systems have crossed that chasm. “So you can spend a lot more money and not go much further. Do you want to be on that curve?” he said. Right now, about 12 years into the semantic space, “it’s about to take off. It’s an inflection point. That’s where we are right now.”
But mastering the opportunity semantic technology offers, he says, “is based on first understanding the business.”