Prescriptive analytics platform and polarity analysis engine

• US 10,235,336 B1
• Filed: 09/15/2016
• Issued: 03/19/2019
• Est. Priority Date: 09/14/2016
• Status: Active Grant

• Current Assignee: Clearsense Acquisition 1 LLC, Clearsense, LLC
• Original Assignee: Compellon, Inc. (Clearsense, LLC)
• Inventors: Bir, Adrian Marc, Liachenko, Nikolai Nikolaevich, Presley, Daniel Brooks
• Primary Examiner(s): Hershley, Mark E
• Application Number:
  US15/266,992
• Time in Patent Office: 915 Days
• Field of Search: 707723
• CPC Class Codes: G06F 16/2455 Query execution
• CPC Class Codes: G06F 16/24578 using ranking
• CPC Class Codes: G06F 16/3329 Natural language query formulation or dialogue systems
• CPC Class Codes: G06F 16/3344 using natural language analysis

Abstract

Embodiments of the systems and methods disclosed herein provide a prescriptive analytics platform and polarity analysis engine in which a user can identify a target objective and use the system to find out whether the user'"'"'s objectives are being met, what predictive factors are positively or negatively affecting the targeted objectives, as well as what recommended changes the user can make to better meet the objectives. The systems and methods may include a polarity analysis engine configured to determine the polarity of terms in free-text input in view of the target objective and the predictive factors and use the polarity to generate the recommended changes.

Claims

1. A prescriptive analytics system for conducting a polarity analysis on unstructured, free text data, the prescriptive analytics system comprising:

a first electronic database storing a set of response data, the set of response data comprising a structured data set and a free text data set, and wherein the set of response data is based at least partly on an aggregated customer feedback dataset;

a second electronic database storing a first objective corresponding to the set of response data, the first objective selected by a user, wherein the first objective is associated with one or more objective values such that each objective value is associated with a degree of impact on one or more outcomes, wherein the one or more outcomes are based at least partly on the first objective selected by the user;

a hardware processor is configured to execute computer-executable instructions in order to;

access from the first electronic database the set of response data;

analyze the structured data set of the structured data to identify key predictive factors within the structured data set that are correlated to the first objective and indicate one or more behavior patterns associated with the first objective;

generate a correlated structured data set from the structured data set using the identified key predictive factors;

convert a plurality of terms within the free text data set of the response data into a plurality of lemmas corresponding to the plurality of terms to generate a standardized lemma data set;

filter the standardized lemma data set to remove terms that are not associated with actionable words to create a filtered lemma data set, wherein actionable words include nouns and adjectives selected by the user;

aggregate the filtered lemma data set with the correlated structured data set;

generate scores for the filtered lemma data set in view of the first objective to create a scored lemma data set;

analyze the distribution of the scored lemma data set against a distribution of the first objective in the correlated structured data set to determine a first number of occurrences of lemmas of the plurality of lemmas associated with a positive objective value and to determine a second number of occurrences of lemmas of the plurality of lemmas associated with a negative objective value;

balance the distribution of the scored lemma data set against aggregate frequency of terms in the scored lemma data set to create a distributed lemma data set; and

assign polarity values to terms in the distributed lemma data set, wherein the polarity values indicate whether each of the terms is a positive term or a negative term, based at least partly on a determination of the terms as being associated with a degree of impact on one or more outcomes;

generate a recommendation action based at least in part on the polarity values.

2. The computing system of claim 1, wherein the hardware processor is configured to execute computer-executable instructions in order to generate recommendations for achieving an improvement to the first objective using the polarity values and the distributed lemma data set.

3. The computing system of claim 1, wherein the generation of scores for the filtered lemma data set in view of the first objective to create a scored lemma data set includes incorporating a user-specified amount of requested change in the objective.

4. The computing system of claim 3, wherein the user-specified amount of requested changes in the objective corresponds to a change to increase the amount of a positive response.

5. The computing system of claim 3, wherein the user-specified amount of requested changes in the objective corresponds to a change to decrease the amount of a negative response.

6. The computing system of claim 1, wherein the hardware processor is configured to generate a data packet that includes a graphical representation of at least a subset of the distributed lemma data set that includes a graphical representation of whether a term in the at least a subset is tied to a negative sentiment or a positive sentiment, the data packet configured for display on a remote computing device.

7. The computing system of claim 1, wherein the hardware processor is configured to generate a data packet that includes a graphical representation of at least a subset of the distributed lemma data set that includes a graphical representation of the frequency and strength of terms in the subset, the data packet configured for display on a remote computing device.

8. The computing system of claim 1, wherein the hardware processor is configured to generate a data packet that includes a graphical representation of at least a subset of the distributed lemma data set that includes a graphical representation of whether a term in the subset is tied to a negative sentiment or a positive sentiment and the frequency and strength of terms in the subset, the data packet configured for display on a remote device.

9. A computer-implemented method of conducting a polarity analysis on unstructured, free text data, the computer-implemented method comprising:

as implemented by one or more computing devices configured with specific computer-executable instructions,

accessing from a first electronic database a set of response data that includes a structured data set and an unstructured data set, wherein the set of response data is based at least partly on an aggregated customer feedback dataset;

analyzing the structured data set to identify key predictive factors within the structured data set that are correlated to a first objective and indicate one or more behavior patterns associated with the first objective, wherein the first objective corresponds to the set of response data and is selected by a user, and wherein the first objective is associated with one or more objective values such that each objective value is associated with a degree of impact on one or more outcomes, wherein the one or more outcomes are based at least partly on the first objective selected by the user;

generating a correlated structured data set from the structured data set using the identified key predictive factors;

converting a plurality of terms within the free text data set of the response data into a plurality of lemmas corresponding to the plurality of terms to generate a standardized lemma data set;

filtering the standardized lemma data set to remove terms that are not associated with actionable words to create a filtered lemma data set, wherein actionable words include nouns and adjectives selected by the user;

aggregating the filtered lemma data set with the correlated structured data set;

generating scores for the filtered lemma data set in view of the first objective to create a scored lemma data set;

analyzing the distribution of the scored lemma data set against a distribution of the first objective in the correlated structured data set to determine a first number of occurrences of lemmas of the plurality of lemmas associated with a positive objective value and to determine a second number of occurrences of lemma of the plurality of lemmas associated with a negative objective value;

balancing the distribution of the scored lemma data set against rarity and aggregate frequency of terms in the scored lemma data set to create a distributed lemma data set; and

assigning polarity values to terms in the distributed lemma data set, wherein the polarity values indicate whether each of the terms is a positive term or a negative term, based at least partly on a determination of the terms as being associated with a degree of impact on one or more outcomes;

generating a recommendation action based at least in part on the polarity values.

10. The computer-implemented method of claim 9, further comprising generating recommendations for achieving an improvement to the first objective using the polarity values and the distributed lemma data set.

11. The computer-implemented method of claim 9, wherein the generation of scores for the filtered lemma data set in view of the first objective to create a scored lemma data set includes incorporating a user-specified amount of requested change in the objective.

12. The computer-implemented method of claim 11, wherein the user-specified amount of requested changes in the objective corresponds to a change to increase the amount of a positive response.

13. The computer-implemented method of claim 11, wherein the user-specified amount of requested changes in the objective corresponds to a change to decrease the amount of a negative response.

14. The computer-implemented method of claim 9, further comprising generating a data packet that includes a graphical representation of at least a subset of the distributed lemma data set that includes a graphical representation of whether a term in the at least a subset is tied to a negative sentiment or a positive sentiment, the data packet configured for display on a remote computing device.

15. The computer-implemented method of claim 9, further comprising generating a data packet that includes a graphical representation of at least a subset of the distributed lemma data set that includes a graphical representation of the frequency and strength of terms in the subset, the data packet configured for display on a remote computing device.

16. The computer-implemented method of claim 9, further comprising generating a data packet that includes a graphical representation of at least a subset of the distributed lemma data set that includes a graphical representation of whether a term in the subset is tied to a negative sentiment or a positive sentiment and the frequency and strength of terms in the subset, the data packet configured for display on a remote device.

17. Non-transitory computer storage having stored thereon a computer program, the computer program including executable instructions that instruct a computer system to at least:

access from a first electronic database a set of response data that includes a structured data set and an unstructured data set, wherein the set of response data is based at least partly on an aggregated customer feedback dataset;

analyze the structured data set to identify key predictive factors within the structured data set that are correlated to a first objective and indicate one or more behavior patterns associated with the first objective, wherein the first objective corresponds to the set of response data and is selected by a user, and wherein the first objective is associated with one or more objective values such that each objective value is associated with a degree of impact on one or more outcomes, wherein the one or more outcomes are based at least partly on the first objective selected by the user;

generate a correlated structured data set from the structured data set using the identified key predictive factors;

convert a plurality of terms within the free text data set of the response data into a plurality of lemmas corresponding to the plurality of terms to generate a standardized lemma data set;

filter the standardized lemma data set to remove terms that are not associated with actionable words to create a filtered lemma data set, wherein actionable words include nouns and adjectives selected by the user;

aggregate the filtered lemma data set with the correlated structured data set;

generate scores for the filtered lemma data set in view of the first objective to create a scored lemma data set;

analyze the distribution of the scored lemma data set against a distribution of the first objective in the correlated structured data set to determine a first number of occurrences of lemmas of the plurality of lemmas associated with a positive objective value and to determine a second number of occurrences of lemma of the plurality of lemmas associated with a negative objective value;

balance the distribution of the scored lemma data set against rarity and aggregate frequency of terms in the scored lemma data set to create a distributed lemma data set; and

assign polarity values to terms in the distributed lemma data set, wherein the polarity values indicate whether each of the terms is a positive term or a negative term based at least partly on a determination of the terms as being associated with a degree of impact on one or more outcomes;

generate a recommendation action based at least in part on the polarity values.

18. The non-transitory computer storage of claim 17, the executable instructions further comprising generating recommendations for achieving an improvement to the first objective using the polarity values and the distributed lemma data set.

19. The non-transitory computer storage of claim 17, wherein the generation of scores for the filtered lemma data set in view of the first objective to create a scored lemma data set includes incorporating a user-specified amount of requested change in the objective.

20. The non-transitory computer storage of claim 19, wherein the user-specified amount of requested changes in the objective corresponds to a change to increase the amount of a positive response.

21. The non-transitory computer storage of claim 19, wherein the user-specified amount of requested changes in the objective corresponds to a change to decrease the amount of a negative response.

22. The non-transitory computer storage of claim 17, the executable instructions further comprising generating a data packet that includes a graphical representation of at least a subset of the distributed lemma data set that includes a graphical representation of whether a term in the subset is tied to a negative sentiment or a positive sentiment, the data packet configured for display on a remote computing device.

23. The non-transitory computer storage of claim 17, the executable instructions further comprising generating a data packet that includes a graphical representation of at least a subset of the distributed lemma data set that includes a graphical representation of the frequency and strength of terms in the subset, the data packet configured for display on a remote computing device.

24. The non-transitory computer storage of claim 17, the executable instructions further comprising generating a data packet that includes a graphical representation of at least a subset of the distributed lemma data set that includes a graphical representation of whether a term in the subset is tied to a negative sentiment or a positive sentiment and the frequency and strength of terms in the subset, the data packet configured for display on a remote device.