System for mending through automated processes

Patent metadata

Field	Value
Publication number	US-10049155-B2
Application number	US-201615001918-A
Country	US
Kind code	B2
Filing date	Jan 20, 2016
Priority date	Jan 20, 2016
Publication date	Aug 14, 2018
Grant date	Aug 14, 2018

Abstract

Systems and methods are provided for transforming historical data collected in response to one or more triggering events, in order to classify textual values. Embodiments access a plurality of textual values from historical transaction data; identify one or more distinct patterns within the plurality of textual values; group the textual values based on the one or more distinct patterns, thereby forming one or more clusters; apply a similarity gauge to the textual values of each of the clusters to determine similarity or dissimilarity among the textual values of each cluster; and filter the textual values of each cluster to determine which textual values belong in each cluster, wherein the textual values that belong are cluster values. Some embodiments also remove undesired characters from the textual values, and in some cases identifying the distinct patterns includes comparing pronunciations and/or phonetics of the textual values.

First claim

What is claimed is:

1. A system for transforming historical data collected in response to one or more triggering events, in order to classify textual values, the system comprising:
a computer apparatus including a processor and a memory; and
a software module stored in the memory, comprising executable instructions that when executed by the processor cause the processor to:
access a plurality of textual values from historical transaction data;
remove undesired characters from the plurality of textual values;
implement a clustering algorithm to the plurality of textual values to identify one or more distinct patterns within the plurality of textual values, wherein the clustering algorithm comprises:
a primary process for coding the plurality of textual values into one or more phonetic components, thereby reducing the plurality of textual values into a combination of consonant sounds, wherein identifying the one or more distinct patterns within the plurality of textual values comprises comparing pronunciations and phonetics of the plurality of textual values; and
a secondary process for identifying and classifying, based on an Internet search, one or more of the plurality of textual values unable to be classified by the primary process;

create one or more clusters by grouping the plurality of textual values based, respectively, on the one or more distinct patterns output by the primary process and the Internet search of the secondary process;
apply a similarity gauge to the textual values of each of the clusters to determine similarity or dissimilarity among the textual values of each cluster;
filter the textual values of each cluster to determine which textual values belong in each cluster and which textual values do not belong in each cluster, wherein the textual values that belong are cluster values;
pass the cluster values for each cluster to a reference table;
store the cluster values for each cluster in the reference table for future access; and
in response to a need for classification of a future set of textual values, access the reference table and lookup the future set of textual values in the reference table to determine whether any of the future set of textual values are cluster values.

2. The system of claim 1 , wherein comparing pronunciations and/or phonetics of the textual values comprises:
applying a double metaphone algorithm to the textual values.

3. The system of claim 1 , wherein applying a similarity gauge to the textual values comprises:
determining a Jaccard distance score among the textual values of each cluster.

4. The system of claim 1 , wherein the instructions when executed further cause the processor to:
connect the textual values that belong in each cluster; and
remove the textual values that do not belong in each cluster.

5. The system of claim 4 , wherein connecting the textual values that belong in each cluster comprises:
applying an OPTNET algorithm to the textual values of each cluster.

6. The system of claim 1 , wherein filtering the textual values of each cluster comprises:
determining a Jaccard distance score threshold;
comparing the Jaccard distance score to the Jaccard distance score threshold for each of the textual values of each cluster, thereby filtering textual values based on their similarity and/or dissimilarity.

7. The system of claim 1 , wherein the instructions when executed further cause the processor to:
apply a standardized value aggregate to the cluster values of each cluster.

8. A computer program product for transforming historical data collected in response to one or more triggering events, in order to classify textual values, the computer program product comprising:
a non-transitory computer readable storage medium having computer readable program code embodied therewith, the computer readable program code comprising:
computer readable program code configured to access a plurality of textual values from historical transaction data;
computer readable program code configured to remove undesired characters from the plurality of textual values;
computer readable program code configured to implement a clustering algorithm to the plurality of textual values to identify one or more distinct patterns within the plurality of textual values, wherein the clustering algorithm comprises:
a primary process coding the plurality of textual values into one or more phonetic components, thereby reducing the plurality of textual values into a combination of consonant sounds, wherein identifying the one or more distinct patterns within the plurality of textual values comprises comparing pronunciations and phonetics of the plurality of textual values; and
a secondary process for identifying and classifying, based on an Internet search, one or more of the plurality of textual values unable to be classified by the primary process;

computer readable program code configured to create one or more clusters by grouping the plurality of textual values based, respectively, on the one or more distinct patterns output by the primary process and the Internet search of the secondary process;
computer readable program code configured to apply a similarity gauge to the textual values of each of the clusters to determine similarity or dissimilarity among the textual values of each cluster;
computer readable program code configured to filter the textual values of each cluster to determine which textual values belong in each cluster and which textual values do not belong in each cluster, wherein the textual values that belong are cluster values;
computer readable program code configured to pass the cluster values for each cluster to a reference table;
computer readable program code configured to store the cluster values for each cluster in the reference table for future access; and
computer readable program code configured to, in response to a need for classification of a future set of textual values, access the reference table and lookup the future set of textual values in the reference table to determine whether any of the future set of textual values are cluster values.

9. The computer program product of claim 8 , wherein comparing pronunciations and/or phonetics of the textual values comprises:
applying a double metaphone algorithm to the textual values.

10. The computer program product of claim 8 , wherein applying a similarity gauge to the textual values comprises:
determining a Jaccard distance score among the textual values of each cluster.

11. The computer program product of claim 8 , the computer readable program code comprising:
computer readable program code configured to connect the textual values that belong in each cluster; and
computer readable program code configured to remove the textual values that do not belong in each cluster.

12. The computer program product of claim 11 , the computer readable program code comprising:
computer readable program code configured to apply an OPTNET algorithm to the textual values of each cluster.

13. The computer program product of claim 8 , the computer readable program code comprising:
computer readable program code configured to determine a Jaccard distance score threshold; and
computer readable program code configured to compare the Jaccard distance score to the Jaccard distance score threshold for each of the textual values of each cluster, thereby filtering textual values based on their similarity and/or dissimilarity.

14. The computer program product of claim 8 , the computer readable program code comprising:
computer readable program code configured to apply a standardized value aggregate to the c

Assignees

• Bank Of America

Inventors

• Neela Raghuram
• Amin Seyamak
• Joa David
• Pawar Samir B
• Mandala Krishna Reddy
• Maddila Santosh Siva Kumar
• Maradla Santosh Kumar

Classifications

• G06Q40/02

  Banking, e.g. interest calculation or account maintenance (credit or loans G06Q40/03) · CPC title

• G06Q30/0255

  based on user history · CPC title

• G06Q40/00

  Finance; Insurance; Tax strategies; Processing of corporate or income taxes · CPC title

• G06F40/40

  Processing or translation of natural language (natural language analysis G06F40/20; semantic analysis G06F40/30) · CPC title

• G06F40/166

  Editing, e.g. inserting or deleting · CPC title