Case Study: SRA International achieves re-indexing in hours and 25% higher accuracy with SAS Text Analytics

SRA re-indexed its collection faster, more accurately — and discovered new documents along the way

SRA International was hired by a government agency to solve a findability and information-organization problem for a digitized library of reports spanning more than 100 years. Manual re-indexing and metadata updates were impractical given changes in information organization and limited resources, and the agency needed a taxonomy and an automated, integrable solution to consistently apply metadata across hundreds of thousands of documents.

SRA used SAS Text Analytics and the SAS Content Categorization Studio to build a semantic taxonomy and rule-based categorization system. The collection was re-indexed in hours instead of months, new documents can be indexed in seconds, accuracy rose to about 90% (a 25% improvement over manual indexing), indexers were redeployed as analysts to refine rules, and previously unsearchable historical documents were discovered.

SRA International

Bill McKinney

Taxonomy Manager