Page 31 - Straive eCoffee Table Book: Data Solutions

What were some of the client challenges?

There were three critical challenges:

Yearly manual
processing of more than Inconsistencies in the
3 million auto accident complexity and quality of the

reports was inefficient scanned reports with more than
and expensive. 50 different formats (one for
each county) and hundreds of
Long turnaround times for data points to be extracted from

processing reports led to each document.
sub-optimal customer
experiences.

Straive created an integrated solution, replacing the manual process by configuring and deploying the Straive Data Platform
and augmenting it with a minimal human layer for validation and processing of irregularities. Straive’s team overcame several
challenges, such as more than 50 different structures, noise, and poor-quality scanned reports.

Straive adopted the following two-phased approach:

• Phase 1: Auto-extraction powered by optical character recognition (OCR) and business rules with an auto-zoning layer

• Phase 2: Custom machine learning models and noise reduction strategies to increase extraction accuracy delivered by
Phase 1 solution

Straive customized and deployed an end-to-end solution in the client environment in fewer than 90 days.

Transforming unstructured textual data | Data Solutions | 31