Kofax, the world's largest information capture vendor, has acquired San Diego-based automated text classification and extraction developer Mohomine in a cash transaction.

Kofax is the product-development subsidiary of DICOM Group plc.

"This acquisition will accelerate growth in our traditional capture markets by further automating the capture of unstructured textual data, whether paper-based or electronic, which will greatly enhance the economic feasibility for unstructured data capture," said Kofax President Rick Murphy.

"Until now, unstructured data capture required heavy use of expensive, labor-intensive classification and indexing, making many projects infeasible. Mohomine technology changes the equation.

"It will also extend our reach into the large and rapidly expanding markets for these technologies which overlap little with our traditional markets."

According to research firm Gartner Group, approximately 80 percent of all enterprise data is unstructured. Unstructured data are documents where the location within the document of salient information cannot be easily predicted. Examples of unstructured data are e-mail, Web pages, PDF files and paper contracts.

Capture technology has been able to address an extremely small percentage of this data and a somewhat larger proportion of semi-structured and structured documents such as forms.

"This market is wide open because no one has been able to solve the problem of effectively capturing this information for immediate workflow, transactions, CRM and decision-support applications," said Kofax Vice President of Marketing Anthony Macciola.

Kofax expects to release the first products incorporating Mohomine technologies later this year.

"Mohomine's classification and extraction technologies have not been designed as standalone applications," said Mohomine Chief Technology Officer Sameer Samat. "Thus Mohomine has focused on licensing its technologies to enterprise software vendors such as IBM, Oracle, Peoplesoft and U.S. and international security agencies.

"However, Mohomine technologies dovetail beautifully with Kofax capture technologies such as document scanning, XML capture, distributed capture and its deep, API-level integration with about 100 content and document management applications.

"When we complete the integration of Kofax and Mohomine technologies, we will have a powerful, end-to-end solution for automatically capturing extremely large volumes of unstructured data for immediate action."

Mohomine will continue to license and support its technologies for its current customers as well as continue to actively seek new customers for licensing both within and outside of Kofax's traditional markets.

Mohomine was originally funded by Windward Ventures, Hamilton Technology Ventures and In-Q-Tel, a venture group funded by the U.S. Central Intelligence Agency.

Mohomine brings two patented technologies to Kofax: MohoClassifier and MohoExtractor. Key differentiating characteristics shared by both of these technologies are:

-- Highly Scalable: The pattern recognition techniques used by Mohomine can process huge volumes of text data on low-end, inexpensive hardware. Humans can classify 20 to 100 documents per hour. Mohomine software can classify 20 to 100 documents per second. Another way to look at it: The MohoClassifier can categorize between 50-100 megabytes of text a minute on desktop hardware, which approximates a 300-page novel per second.

-- Language Independence: Unlike many natural language processing approaches, Mohomine's pattern recognition software doesn't rely on understanding each language. It has been used successfully with many European languages, Arabic and Chinese.

-- Highest Accuracy: Accuracy ranges from the mid-60 percent to the high-90 percent range, depending on the type of document. According to Gartner, a classifier which achieves 60-percent accuracy justifies the cost of installing and maintaining the system.

-- Ease of Deployment and Integration: The learn-by-example architecture, combined with easy-to-understand and use APIs enables the Mohomine software to be packaged within existing Kofax products and deployed by non-classification and extraction experts quickly and at low cost. Competing, rules-based architectures require the labor-intensive development and maintenance of ontologies for each unique application.

About Kofax

Kofax Image Products Inc. (www.kofax.com), a division of the DICOM Group plc, is a leading developer of application software and image-processing products for the Electronic Data and Document Capture (EDC) market. EDC is essential to helping document-intensive organizations economically, reliably and securely collect, transform and deliver information from all sources into their electronic business processes and archives.