One of the surprising things about big data analytics is just how much free stuff is on offer. From open data sets via the UK government to open source software for everything from data visualisation to graph databases, you can pick up most of the tools you need without having to reach for your wallet. The exception is support in converting that toolkit into an actual solution to a business problem.
“Analysts have consistently told us that chief data officers - as well as chief information officers and chief financial officers - are interested in using big data analytics to create business intelligence, but they don’t know where to start,” said Rod Fontecilla, VP and global lead, advanced analytics, Unisys told DataIQ in an interview. “They don’t have a roadmap for where to go and what to expect.”
As a consequence, investments into new resources like data science have either stalled because of a lack of a compelling business case or have been viewed as disappointments, often as a result of over-inflated expectations about their impact or the timescales involved. “CDOs are asking for help with their strategy - where to start and what to do,” said Fontecilla.
In a remarkable move, Unisys is launching in mid-May a new Artificial Intelligence Centre of Excellence (CoE) that will allow users to upload data sets, have machine learning algorithms applied and get a viable proof of concept delivered back by a team of data scientists - all for free. “The CoE is a suite of resources that are very hot in the market,” he explained. “They are very complex to build because they are a combination of complicated processes, data, algorithms and data science. We have put them all together so they can be delivered as a toolkit.”
When version 1.0 goes live online, it will, for example, be able to ingest bulk uploads of documents, extract entities such as names, numbers or addresses, then identify links between them across those documents. That kind of machine learning is at the cutting edge of law enforcement currently, where it is used to spot connections between a stolen van and an explosion as detailed in police reports. Said Fontecilla: “We can do that sort of thing very easily.”
The basis for Unisys’s skills in this area is its 15-year-old contract with US Homeland Security Coast and Border Patrol. It analyses the entry and exit documents for everybody going in or out of the United States to predict which ones might be bad actors. The solution runs on the world’s biggest Oracle data warehouse, handling 1.3 billion transactions daily and generating 700,000 predictive models.
The solution runs 24/7, 365 and needs to be resilient given the mission-critical nature of its task. While the CoE will not operate at that scale, “that methodology is folded into the centre.” Linked analytics of this sort is one of four tasks sets which the analytics resource has been set up to handle. Machine learning will be available to apply algorithms from the firm’s library in order to see how predictive they are. “You won’t need to be a data scientist to do it because we do the hard work behind the scenes,” explained Fontecilla.
Natural language processing is another part of the proposition, especially to allow for testing of AI bots for customer services. Fontecilla noted that there is strong interest in these from the public sector. Location-driven internet of things experiments will also be supported. “The only thing the user will have to do is buy a couple of iBeacons, download the Beacon Bully app, and they will be able to track customers inside their stores. We’ll provide the analytics and dashboards to show that the process works,” he said.
Running proofs of concept has been an important new business tool for Unisys, which just won a contract with California State University to build its own centre of excellence. Fontecilla admits that he has no idea what the level of demand is likely to be once the CoE goes on-stream, but the company’s teams of data scientists around the globe - half of whom have PhDs - are on standby.
The next process to get delivered will be social analytics. “We want users to be able to go to our website and enter the name of a company, like United Airlines, and see what people are saying about it, then combine that with their particular information to give a picture that includes context,” he said. A partnership with a third-party rights holder to those social media data feeds will provide the vehicle for this solution.
With a promise to “never talk about the technology, only the business problem,” the CoE is a surprising venture for the firm. But Fontecilla readily admits that, “in Unisys, we are like a start-up. Analytics is the smallest service line in the compnay and we have not been visible, despite our 15 years working with Homeland. We are the best-kept secret.” Until May, when it seems likely that word is likely to spread - and fast.