If you're an IT leader who wants to reduce the time and costs integration projects take, then you might want to sit down and talk with your development or data management team about mapping.
If you’re not familiar with the nitty-gritty of integration work, mapping is when you’re matching elements of one data set to the other data set. You’re basically making sure the software knows, as a very simple example, that column A is where you’ve stored a customer’s last name in one dataset, but in the other dataset, the last names are in column B. This is a key part of integration, and making sure it lines up the way you want is a starting point for data quality.
That makes mapping a big deal. In fact, mapping can account for 40 percent or more of the costs on B2B integration projects, according to Liaison Technologies.
So … theoretically, if you can simplify mapping, you could cut your integration costs and the time to deploy significantly.
Now, I've no doubt there are lots of ways to accomplish this and I'm sure I'll soon be receiving emails from all sorts of vendors on the topic. But recently, Liaison Technologies shared what I thought was an unusual approach to this problem, and now I'm sharing it with you.
Liaison decided to apply artificial intelligence to data mapping.
Bayesian Artificial Intelligence techniques formed the foundation for Liaison's new mapping tool, Map Intuition, according to Robert Fox, director of B2B/EAI Software Development at Liaison Technologies.
Fox is a veteran of simplifying integration, particularly in the B2B space. He was an original contributor to the ebXML 1.0 specification and the former chair of marketing and business development for ASC ANSI X12. He also co-founded and co-chaired the Connectivity Caucus, a group devoted to improving EDI Network interconnectivity.
During a recent email interview, Fox explained how Map Intuition's “inference” engine differs from other automapping techniques:
Many companies attempt to speed up the mapping process by using an ontological or vocabulary approach. By creating a dictionary of canonical business terms, and associating these terms to the interfaces, the mapping tool can automatically map matching terms. This approach takes a "vertical" problem, and makes it "horizontal" – meaning, user spends less time mapping top-down, but more time defining the vocabulary (picture adding terms to an existing interface as horizontal activity).
The traditional approach works well if the interfaces are reused multiple times and the mappings are one-to-one.
“In the real world, this equates to disappointing return on investment,” he said.
Map Intuition looks at the data and the model together to infer mappings. You can rank terms, demoting or promoting matches, to fine tune the proposed mappings.
“This allows the user to very quickly see what a proposed mapping is doing with source and target data to finish the map. The more data source/target sample files you provide, the better the mappings,” he said.
For Liaisons, this pays off for B2B integration projects, which often involve translating maps from third parties.
Map Intuition is part of Liaison’s Contivo Vocabulary Management Solution (VMS). To learn more, check out our full interview.
Disclosure: I also write a blog post about B2B integration and supply chain management at B2B.com, a vendor-neutral site that is owned by Software AG, which also does B2B integration.