A global media organization that provides fixed-line internet IP TV to some 10 million customers had a new business initiative that was going to require it to gain some insight into its client base. After some 15 years in business, though, it’s not surprising to learn that that information exists – and re-exists in many different forms – across many legacy applications, and trying to map those customers’ old purchase relationships to a new product catalog as part of a new payment and sales platform could have been just the thing to slow down the company.
Does that situation sound familiar? If your company’s been in business for some length of time, the answer probably is a resounding yes. Like this media business, you may well be in a market with plenty of competitive threats, meaning that unless you constantly innovate, your bread and butter is threatened. And so, you too, probably always are turning to your IT infrastructure team with new requirements.
“And it can be hard for them to build what they need to deliver,” says Carl Bray, product manager at Ontology Systems.
He knows firsthand, having worked with the media company to get things in order, fast and with greater flexibility.
Bray, who will be discussing the case study at the upcoming Semantic Technology and Business Conference in NYC, says the job married agile project methodology for defining the data access approach and semantic search. “The tradition of a migration and integration project normally takes a waterfall approach,” he says, requiring developing a big data dictionary, studying it, understanding the schemas, and conducting a big exercise to map schemas and relationships together – all at design time before looking at the data.
“We say look at the data first, bring it in, make it searchable,” he says. “In the semantic world, because the data is there, you just put a model on top of it and you take way all the constraints” that can inhibit more traditional approaches, such as having to go through the various gating processes to test the hypothesis. With an agile approach for migrating and integrating data for new efforts, models can be incrementally built and, as they are formed, you can very quickly show things that don’t fit, “and it is searchable and viewable and you can point out there and then these things to your customers in the data.”
At the same time, not every piece of data needs to be modeled as semantic data to be searched. It’s possible, he says to link in unstructured information with RDF triples to present to users as an end user view of data. “This lets us deal with more data and with a lot more variety of things,” he says, and the benefits of being able to ask questions that users don’t necessarily know that they’ll want to ask when a new project begins remain.
In the case of this company, another benefit was that users was that the ability to semantically search the new system took the load off its scarce resource of subject matter experts distributed around the country. “Using the semantic search engine, they can do a better search themselves and get more information about the proper domain so that they can ask a more specific question of the subject matter expert, or avoid asking him or her a question in the first place, freeing up their time,” he says.
Experiences like this, Bray says, make the point that it’s time for people to rethink how to manage massive projects for greater success, by taking the agile route. “Traditional tools just aren’t receptive to agility. They all are designed for big design process upfront and they make change difficult,” he says. “Turn that on its head: Be agile upfront. Embrace change, and do that by delivering semantic tools to let that happen.”