TDWI | Training & Research | Business Intelligence, Analytics, Big Data, Data Warehousing

Business Objects’ Data Integration Coup

Business Objects—the industry’s Accidental Data Integration vendor—continues to flex its muscles.

• By Stephen Swoyer
• June 6, 2007

Late last month, Business Objects SA acquired text mining and unstructured information access specialist Inxight Software Inc. What’s a big business intelligence (BI) and performance management (PM) power like Business Objects doing picking up a specialty vendor such as Inxight?

Call it a case of Business Objects—the industry’s Accidental Data Integration vendor—continuing to flex its muscles.

Among the major data integration powers, Business Objects sticks out like a sore thumb. Market leaders Informatica Corp. and IBM Corp. focus mostly on data integration middleware. (For the record, Informatica did make an ill-advised foray into reporting and analytics, while Big Blue likes to promote its DB2 RDBMS as "middleware," too.)

Like Business Objects, SAS Institute Inc. is a bona-fide business intelligence (BI) powerhouse. But SAS’ ETL dominance is to an extent an outgrowth of its decades of leadership in both statistical analysis and data mining.

Business Objects first became a data integration player five years ago, when it acquired ETL specialist Acta. Few knew what to make of Business Objects’ purchase at the time, but almost no one expected it to mature the former Acta technology—which it re-branded Data Integrator—into a competitive enterprise ETL offering. Along the way, Business Objects picked up enterprise information integration (EII, via its acquisition of the former Medience) and data quality (thanks to its purchase of Firstlogic), too.

Inxight adds to this simmering data integration stew. In addition to its best-of-breed text analytic engine, Inxight boasts unstructured information access, federated search, and even data visualization capabilities.

"There’s a lot to the Inxight product, because so many capabilities are brought to bear in a single tool," agrees Philip Russom, senior manager of TDWI Research. "Inxight is known for its text mining capabilities, which applies natural language processing to discovering and documenting topics that are discussed in textual documents," he continues, adding that Inxight’s text mining facility does a lot more than just parse unstructured text.

"The resulting output is usually a topic cluster, which reveals clusters of topics found in the body of documents mined," he says. "More than a mere listing of topics, the relative weight of topics and relationships among clusters are revealed, too. Under the covers, the clusters and their relations are modeled in a rich neural net data structure."

Russom also lauds Inxight’s end user (client) experience. "The user interface of the Inxight end-user tool represents topic clusters, weights, and relations in a compelling visual manner, so the end-user can see at a glance which topics appear most often or most deeply in the documents mined," he indicates. "Hence, the visualization is a visual analysis of the semantic content of the body of documents….with a mouse click, the end-user can interact in real time with the visualization to navigate among topics and drill into topics, summaries of them, and the source documents themselves. So, the visualization is also a user interface for navigating semantic structures and discovering source text."

Juliette Sultan, vice-president of Business Objects’ information discovery and delivery (IDD) business unit, says Inxight’s text analytic capability was the primary driver behind the acquisition. "We were thinking way ahead trying to understand how we could address the unstructured data in a corporation. Obviously, through business intelligence we are the master of handling the structured data—databases, data warehouses—but we’re also talking to customers who want to understand how they can leverage data they access to search, they have to find a way to aggregate it. Whether it comes from old emails, discussions, [or] blogs, all of the text fields, the notes, all of the wisdom of the corporate employees—they want to have a way to capture that."

Details of the acquisition weren’t disclosed, but—given all that it brings to the table—Inxight couldn’t have come cheap. The company boasts a large OEM business, for starters (Sultan claims that it has 300 OEM partners), along with a wholly owned subsidiary, Inxight Federal System Groups, that sells into federal agencies across the globe.

Once it’s exposed to the rest of the Business Objects stack, Sultan says, Inxight can inject a "qualitative" aspect into executive decision making. "Imagine being a patient in a hospital. You obviously provide a lot of your data that is structured: your name, your address, your social security number, your plan ID, but also what allergies you have, what medications you’re on, and so forth. This is treated as structured information," she comments.

"But once you’re a patient [in a hospital], you have a patient chart at the end of the bed," Sultan continued. "That’s really where all of your information is captured from nurses or doctors. [Inxight has been] correlating with pharmaceutical databases and capturing those types of insights. These are [insights] that would otherwise just not be captured in an organized form. In one case, they noticed that blood sugar levels of diabetes patients can be correlated to their rate of infection after cardiac surgery."

Because the acquisition hasn’t yet closed, Sultan declined to talk about integration road maps, product support schedules, or other post-acquisition intangibles. She does suggest that Business Objects shouldn’t have too much trouble bringing Inxight into the Enterprise XI fold. "Inxight is working with a lot of different software development kits. That’s why they have more than 300 different OEMs. We also have started working with Inxight over the last year, so we have a good grasp of how we can integrate this technology."

Russom, for his part, says the deal is a coup for the BI giant.

"Business Objects has acquired mature technology for text mining, which obviously gives them a new unstructured data access layer for their concept of enterprise information management," he concludes. "But they also have a new neural-net-based data visualization capability that will complement their other set-based approaches. This is fully consistent, since two areas that Business Objects is pressing forward with of late are: EIM—as an extended data integration platform—and visualization."