Text mining and content categorization

Text Analytics for P&C Claims Industry

Occasional Contributor
Posts: 7

Text Analytics for P&C Claims Industry

I have developed some software using only BASE SAS to read documents (text, csv, excel, pdf, etc) created by P&C Insurance Claim Adjusters.  These documents range from a short paragraph to literally hundreds of pages (60 lines x 80 char columns) to mark these documents on over 400 different claim issues from "is there subrogation opportunity being overlooked", "is the claim an Exploder", "Could claim be Fraud bound", "is the Claim a trend towards a new unforeseen class action type issue", "would this claim make a Sr. VP Claim Executive sleepless", "would this claim be an ideal example to train new claim adjusters", "which of my millions of claims need to be audited", "should this claim be routed to a claim adjuster specialist in this type of claim", and the list goes on.  And the list of 400 flags can be expanded as part of the application.  My Ngram Table consists of over 200,000 ngrams that reflect low False Positives.  The application builds dynamic Stories based upon what you are interested in "finding" - what I refer to as a "CONCEPTUAL SEARCH" which are customizable - you search for your "topic" which could consist of literally hundreds of word phrases behind the "topic" specific to your needs/desires.

Looking for clients and/or SAS TM Product Manager to discuss taking this application commercially or incorporating what I have done into the existing SAS TM product.

Results thus far has shown an ROI ranging from just 2% upwards to 25% in processing several years of claims from a small list of previous clients which can not be shared.

Any ideas, comments or suggestions are welcomed.

Best Regards,

Charles Patridge

Charles_S_Patridge AT Prodigy Dot Net

Ask a Question
Discussion stats
  • 0 replies
  • 1 in conversation