Solving for Text Intelligence: Straive enabling accident data reporting

Output & TAT


TAT of

24

hours
with

99.5%

accuracy

Cost Savings


34%

cost saving annually

The power that data holds is unprecedented. This is especially true in a world that is swiftly moving towards a full-fledged data revolution. It is not a mere conversation anymore. It is the only way to operate and differentiate today. Data, when understood correctly, can create significant business and customer experience opportunities. However, data can come in all shapes and sizes and can create challenges for enterprises looking to tap their real potential. One such example is that of property and casualty insurance providers, who are constantly on the lookout to enhance their knowledge of their customers to differentiate in customer experience and offerings. This is their ultimate edge over their competition. It includes understanding consumer driving patterns across individuals, needs, regulations, seasons, vehicles, cities, state, and many other such factors. This means they want to capture the driving and driving-related data that is floating around in different formats and can change at any moment. Unstructured data of this kind are spread in an ocean of tiny specs of information waiting to be strung together for its most effective use. They are waiting someone to connect the dots!

One of Straive’s long-standing clients owns a leading platform for analysis of national accident reports, which primarily serves law enforcement agencies, individuals, authorized parties, and insurers. Their end users derive insights about driving habits and behaviors and are input to models for insurers to better understand their customers’ driving patterns. This data was extracted primarily through manual keying as the input files from each state had varying formats and came from scanned image PDFs. Given the volume of reports and data points that continue to evolve, manual data acquisition was not only time consuming but also created significant quality issues. The client needed a data solution that would support their current needs (in terms of data points, coverage, quality, and speed) but would also continue to scale as the need for additional data points and the variety of sources to capture these data points increase. There was a need for a solution that could not only handle all these challenges but could also scale up in a short timeframe so the client could quickly learn the applicability and impact of the extracted data. That is where Straive’s data platform was perfectly poised to help.

Leveraging our data expertise, we implemented Straive Data Platform (SDP) to automate the data acquisition, enrichment, and delivery operations in a way that would scale with minimal manual intervention. In order to deal with variability of data points, sources, and volume while maintaining quality, SDP’s auto extraction feature was deployed with both a rules-based and ML engine. The platform extracted critical data elements from police auto-accident report images coming from 64 different state agencies, and automatically adjusted to various constantly changing formats. In addition, SDP’s annotation layer ensured that through human curation, a data accuracy rate of 99.8% was maintained, while any exception processing was fed as inputs back into the ML model for better auto-extraction accuracy. This was all set up and scaled in a three-month timeframe.

Straive has been providing specialized data solutions — solving data intelligence problems in the unstructured data domain — for more than a decade. With the results and benefits we see in our client engagements and their impact on customer stories, product discoveries or operational improvements leveraging all dark data and public data, we have sharpened our focus with additional investments in people and platforms. If you have a data challenge, we have the solution. Join us on our quest to elevate knowledge to solve the critical problems of our time.