Core Technology

Text-Mining Automation: Based on proprietary pattern recognition technology, the Praedea DEP utilizes genetic algorithm-enhanced machine learning systems to extract specific, targeted data from documents. In a two-stage process, the Praedea DEP first parses the document(s) into a hierarchy of generic components to decompose the often lengthy text into a logically connected set of similar text elements. Next, the Praedea DEP applies a statistical approach to identify and extract the targeted data from the parsed document(s), in which genetic algorithms are invoked to compute optimal data location determinants. These location determinants include both distance and key phrase indicators. Unlike traditional parsing methodologies that merely look at a limited number of specific examples or string occurrences in attempting to locate targeted data, the Praedea DEP observes hundreds or even thousands of key phrases and text pattern layouts that surround the targeted data to provide for greatly heightened data extraction accuracy and success rates.

While the text-mining technology and underlying processes of the Praedea DEP are extremely complex, the Praedea DEP presents users with a simple user-interface utilizing a step-by-step “Term Model Wizard.” The Term Model Wizard enables users to independently build and optimize text-mining term models – all without the need for in-house engineering or text-mining expertise.

Technology Platform: The Praedea DEP runs on operating systems compliant with the JavaTM version 1.3 or 1.4 technology and conforms to the published specifications for the SunTM SolarisTM, Red Hat® Linux®, HP-UX, and Microsoft® Windows NT®, Windows® 2000 and Windows® XP platforms.

The Praedea DEP is broadly compatible with most database platforms. Customers are required to have their own database license(s) when the Praedea DEP is implemented in a production environment or, alternatively, to utilize open source offerings such as the MySQL® database. The Praedea DEP requires a JDBC-capable RDBMS install on a database server machine. This machine can co-reside with the Praedea server or be IP addressable from the Praedea server.