Method and system for classifying documents

• US 7,849,030 B2
• Filed: 05/31/2006
• Issued: 12/07/2010
• Est. Priority Date: 05/31/2006
• Status: Expired due to Fees

• Current Assignee: Hartford Fire Insurance Company (The Hartford Financial Services Group, Inc.)
• Original Assignee: Hartford Fire Insurance Company (The Hartford Financial Services Group, Inc.)
• Inventors: Ellingsworth, Martin E.
• Primary Examiner(s): HOLMES, MICHAEL B
• Application Number:
  US11/443,760
• Publication Number:
  US 20070282824A1
• Time in Patent Office: 1,651 Days
• Field of Search: 706/20
• US Class Current: 706/20
• CPC Class Codes: G06F 16/313 Selection or weighting of terms for indexing
• CPC Class Codes: G06F 16/353 into predefined classes

Abstract

The invention provides a method and system for classifying insurance files for identification, sorting and efficient collection of subrogation claims. The invention determines whether an insurance claim has merit to warrant claim recovery efforts utilizing software code for partially describing a set of documents having unstructured and structured file data containing terms and phrases having contextual bases, code for transforming the terms and phrases, code for iterating a classification process to determine rules that best classify the set of documents based upon context, code for incorporating the rules into an induction and knowledge representation, thesauri taxonomies and text summarization to classify subrogation claims; code for calculating a base score and a concept vector to identify the selected claims that demonstrate a given probability of subrogation recovery.

Claims

1. A computer-implemented method for determining whether a claim has merit to warrant claim recovery, the method comprising:

describing, using a computer, a set of file documents having data containing a context;

iterating upon a classification process to create rules that classify the set of documents based upon the context;

incorporating the rules into a classification schema to classify the claims;

generating said classification schema using said computer;

wherein the schema calculates a score and a concept vector to identify the claims that demonstrate a threshold probability for recovery.

2. A computer-implemented method for determining whether a claim has merit to warrant claim recovery, the method comprising:

describing, using a computer, a set of documents having unstructured and structured file data containing terms and phrases having one or more contextual bases;

transforming the terms and phrases for iterating upon a classification process to create rules that classify the set of documents based upon the context;

incorporating the rules into classification schema to classify the claims;

generating said classification schema using said computer;

wherein the schema calculates at least one score and one or more concept vectors to identify the claims that demonstrate a threshold probability for recovery.

3. The method of claim 2 wherein the step of transforming includes using a template for parsing the terms and phrases into an electronic data stream.

4. The method of claim 2 wherein the classification process includes the step of incorporating one or more of a rule induction and knowledge representation, thesauri and taxonomy and a text summarization derived from the set of documents.

5. The method of claim 2 including the step of routing the set of documents.

6. The method of claim 2 including the step of processing a first phase where representative subrogation files serve as test cases to establish the parameters and rules and a production process phase wherein files are classified according to score, a concept vector and a threshold probability of subrogation recovery.

7. The method of claim 2 including the step of processing a first phase having representative insurance files comprising information related to an accident.

8. The method of claim 7 wherein the information related to an accident includes one of a file number, a payment indication, medical payments, indemnity payments, expense payments, amounts reserved, progress notes and task diaries, correspondence, transcriptions, recorded statements, photos, and physical evidence.

9. The method of claim 2 wherein the step of processing unstructured data includes utilizing an N-Gram analysis.

10. The method of claim 4 wherein the step of processing unstructured data includes utilizing a keyword process to flag terms to be applied to a classification analysis.

11. A computer program product embodied in a computer readable medium for determining whether an insurance claims has merit to warrant claim recovery comprising:

code for causing a computer to partially describing a set of documents having unstructured and structured file data containing terms and phrases having contextual bases;

code for causing a computer to transform the terms and phrases;

code for iterating a classification process to determine rules that classify the set of documents based upon context;

code for causing a computer to incorporate the rules into one or more or an induction and knowledge representation;

thesauri taxonomy and text summarization to classify claims;

code for causing a computer to calculate a score and a concept vector to identify the selected claims that demonstrate a threshold probability of recovery.

12. A computer system for determining whether a claim has merit to warrant claim recovery comprising:

a computer implemented means for describing a set of documents containing terms and phrases having contextual bases;

a means for transforming the terms and phrases;

a computer implemented means for iterating a classification process to determine rules that best classify the set of documents based upon context;

a computer implemented means for incorporating the rules into an induction and knowledge representation;

a thesauri taxonomy and a text summarization to classify claims;

a computer implemented means for calculating a base score and one or more concept vectors to identify the selected claims that demonstrate a given probability of recovery.

13. The computer system of claim 12 further comprising one or more means for matching on a term or phrase utilizing an N-Gram process;

means for finding terms or phrases within a Levenshtein edit distance;

wildcard queries;

synonym matches;

soundex;

span queries;

range queries;

Naï

ve Bayes;

categorization based on term and phrase frequency and parts of speech.

14. The computer system of claim 12 further comprising one or more of means for one of a statistical analysis or regression analysis where a statistic proves or disproves a null hypothesis regarding a class in which to include the unstructured textual data.

15. The computer system of claim 12 further wherein the claim is an insurance claim.

16. A computer process for sorting files into one or more categories comprising:

analyzing, under control of a computer, associated electronic records utilizing at least one N-Gram technique and at least one Levenshtein technique to sort other associated data records into one or more categories;

producing, using a computer, a score, a concept vector and a threshold that classifies the files into collection strategies.

17. The computer process of claim 16 further comprising the further step of determining that a level of accuracy has not been achieved and adding N-Grams to improve classification performance.

18. The computer process of claim 16 comprising the further step of assessing the accuracy of classification until an assigned error between a classification and a defined accuracy level is achieved.

19. The computer process of claim 16 comprising the further step of indexing terms and phrases found in a file.

20. The computer process of claim 16 comprising the further step of searching a dictionary of all N-Grams by computing the edit distance between one of a string and a list of terms found in a file.

21. The computer process of claim 20 comprising the further step of performing a Naives Bayses analysis on one of the string and a list of terms found in a insurance file to form a subclass of high potential one or more classes of files having collection potentials.

22. A computer system for determining whether a claim has merit to warrant claim recovery comprising:

a file system for electronic claim files to serve as input to a search and indexing engine;

a database for concepts and term data created from a classification process for uploading to a scoring engine;

an indexer having at least one index database to store groups of synonyms each associated with a concept wherein;

(1) each concept forms an element in a vector; and

(2) a file stored in the file system is compared against the elements of the vector to determine whether the file contains the stored concept element; and

(3) if the file contains an element of the concept vector the event is flagged.

23. A computer process for determining whether a claim has merit to warrant recovery comprising:

inputting electronic claim files to a search and indexing engine controlled by a computer;

creating a dictionary of term data having synonyms relating to the claim files and uploading the data to a scoring engine associated with a computer;

storing at least one database of the synonyms each associated with one or more concepts wherein;

(1) one or more concept form into an element in a vector;

(2) a claim file database of the synonyms is compared against the elements of the vector to determine whether the file contains the stored concept element.

24. The computer process of claim 23 wherein if the file contains one of an element of the concept vector or a business rule it is not further tested.

25. The computer process of claim 23 wherein a score derived from an N-Gram analysis forms an element for deciding if a claim ahs merit to warrant claim recovery efforts.

26. The computer process of claim 25 wherein each concept contained in a vector forms a pattern.

27. The computer process of claim 26 wherein each pattern forms an element for deciding if a claim has merit to warrant claim recovery efforts.

28. The computer process of claim 25 wherein the score and the pattern is compared against a pre assigned threshold.

29. The computer process of claim 25 wherein the score and the pattern is compared against a pre assigned threshold and if the score is greater then the pre assigned threshold the file is referred to one or more destinations.

30. The computer process of claim 25 further including the step of incorporating term data in a dictionary of phrases.

31. The computer process of claim 26 further including the step of pointing through a trace to a claim in which the phrases occur and linking them to the concepts that are indicative of a collection claim.