Master Data Management (MDM) systems are far from immune to error and often encounter trouble from duplicate data entries. If a company has more than one customer from a single organization or somebody misspells the name of a customer who purchased a second product, it can be embarrassing when an MDM system doesn't make the connection.
This was the theme of Tibco's latest update to its Collaborative Information Manager product. The company integrated an algorithm from Netrics meant to discover and clean duplicate master data living in the system. It attempts to sort out misspellings, find links between entries and put it all in a global perspective.
"Fundamentally, what MDM suffers from is that the database is looked at as the center of the world," said Sid Suri, senior product marketing manager at Tibco. "When you do that you take SOA, which is a distributed environment, and turn it into a centralized hub."
Tibco took a distributed approach to MDM since acquiring and absorbing Velosel from that space in 2005. The company's approach was to develop for service-oriented architectures and keep the MDM system data type-independent. To keep things in line with SOA principals, Tibco incorporated in Collaborative Information Manager the distributed caching technology it developed for the financial services industry. This keeps the data flowing faster than in systems that must frequently write to a disc.
"It looks at two datasets, either fields or records, and tells you the probability that they are the same, or how different they are," said Suri. "It's about matching and linking records that have slight differences in them."
Matches appear in Collaborative Information System's GUI and are fed through the products governance system where a user can manipulate the results and take an action.