It’s a pattern: SAS buys text mining specialist Teragram

Extending its coverage from statistical to text analytics, business intelligence (BI) specialist SAS Institute has acquired Teragram, a 40-person organization based in Cambridge, Mass. The deal, which brings SAS the kinds of natural language analytic tools it has lacked, will allow the company to first focus on analytic applications targeting customer touch points, such as warranty and afterma...
By Tony Baer, senior contributing editor (tbaer@tbaer.com) June 1, 2008

Extending its coverage from statistical to text analytics, business intelligence (BI) specialist SAS Institute has acquired Teragram, a 40-person organization based in Cambridge, Mass.

The deal, which brings SAS the kinds of natural language analytic tools it has lacked, will allow the company to first focus on analytic applications targeting customer touch points, such as warranty and aftermarket service calls.

The merger brings together two companies that have not heavily dealt with each other in the past, as SAS previously relied on a relationship with Inxight, a text analytics provider later acquired by Business Objects .

And although SAS plans to operate Teragram as a separate business unit, it also intends to develop integration points between its core analytics offerings.

Teragram’s tools provide context-based searches that complement SAS’ existing text mining tools, which conduct pattern matching searches but currently lack capabilities to infer similarities between related but different text patterns.

With Teragram adding higher-level context, a context-enhanced SAS Text Miner could elicit semantic relationships among the data it unearths, where for instance, it could correlate diagnoses with context on how those diagnoses were arrived at, or whether the diagnoses were made using consistent logic.

According to Gaurav Verma, director of BI marketing for SAS, the goal is to synchronize analyses of structured and unstructured data earlier in the process, rather than after the fact, which is current practice.

Verma says it will also provide for natural language processing to refine hybrid queries involving text and structured data before generating reports.