AmLaw 10 Firm AutoClassifies Extensive Data Backlog While Maintaining Strict Country-wide Data Security Protocols

Case Study

 

PowerHouse AutoClassifies On Premise documents half a world away

Situation

Seeking to tackle over 20 million accumulated files of attorney work product, this global, AmLaw 10 law firm needed a solution beyond asking their 500+ attorneys to classify files by hand. 

Solutions Applied:

AutoClassification

Document Analytics

Electronic File Processing

OCR & Text Extraction

Analytics & Data Mining

Products Used:

PowerHouse

h

Download the PDF

Challenge

With an uncertain target Document Management System in mind, the firm sought a solution via AutoClassification.  However, the firm, and all of its data, is located outside the United States, and all data, systems and software must live on premise within its country borders.

Solution

Working together with the firm’s Application and IT departments, Valora installed PowerHouse On Premise in their offices outside of the United States, providing all configuration and upgrade efforts remotely, sight unseen. 

Working via secure, remote access, Valora installed and configured PowerHouse across a suite of virtual servers and processing machines to match prior hand-performed classification of older files, using the older data as guidance.  PowerHouse extracted and created a set of tailored rich metadata for its Rules Engine to levy.  Its Rules Engine now automatically classifies documents on an ongoing basis, which eliminates the need for manual document coding. Utilizing the rich metadata, as well as textual content for analysis, the PowerHouse Rules Engine classifies each file into its proper Matter Number and repository location. 

Results

Valora’s accuracy scores on this project exceeded 90% for the correct selection of client matter numbers, with a choice of Best Match and two alternates.  The high results mean thousands of hours of saved billable attorney time, plus savings in easy retrieval based on matter number, and a host of other rich metadata.