What's in a word...Attensity announces new text analytics suite

Published 25 September 2006

Computational linguistics-based approach ideal for sorting through unstructured data to identify hidden patterns; applications range from managing insurance claims to detecting smuggling and counterfeiting; new product brings analytics and search together under a single label

Think of it as inventory management for information. In the same way that goods piling up in the warehouse hurt a company’s bottom line, the failure to succesfully interpret and manage information can cause companies (and government agencies) to miss valuable opportunities — opportunities they won’t even know they even had. This is particularly true of unstructured text, such as e-mails, letters, and reports, that are not created using a template and are therefore unsortable beyond a simple word search function. According to Palo Alto, California-based Attensity, 85 percent of text in a company is unstructured in this way, one of the reasons why the worldwide market for natural language software isestimated to reach $1.84 billion by 2008.

Attensity relies on computational linguistics — using computers to create statistical and logical models of natural language — to draw out critical data points from unstructured text and place it in a relational format. An insurance company, for instance, might be interested in tracking fire-related claims in order to isolate areas of high risk. Attensity’s software not only would use advanced search technology to find all claim letters that mentioned the word ‘fire’ and all of its synonyms and related terminology, it would use linguistic assessment tools to give adjusters better information about who or what was causing the fires and whether there were any patterns in how they were being reported. There are critical homeland security uses as well. U.S. Customs and Border Protection, for instance, once disrupted a drug-smuggling operation after exhaustive analysis of shipping manifests located a noun that was committing an unusual verb — a banana swimming or a light bulb reading, for instance (company officials, of course, would not provide the actual words used).

This week the company, which received first and second round funding from In-Q-Tel for $3-4 million, announced Attensity 4, an improved version of its text analytics suite that brings together a host of exhaustive extraction tools and search engines into a single, easily interfaced unit. In interviews company officials said they saw government, law enforcement, and insurance and health industries as their target market. The company already works closely with Northrop Grumman, General Dynamics, and other major integrators.

-read more in this company news release; read more about the text analytics market in this IDC report