Granted Patent B2

US 11,574,125 · App. 17/039,752 · Granted Feb 7, 2023

Inventors: Madhusudan Therani (San Jose, CA); Anil Mathews (Bangalore, IN)

Assignee: NEAR INTELLIGENCE HOLDINGS, INC.

G06F40/295G06F16/90332G06F40/35

Abstract

Claims (38)

1. A processor-implemented method for automatically parsing an unstructured conversation using natural language processing to determine target entities for an engagement activity, the method comprising:

obtaining at least one unstructured conversation associated with an asset between a plurality of entities using a dialogue manager from at least one communication channels;

parsing said at least one unstructured conversation associated with said asset using a natural language processing (NLP) model to obtain a target information associated with said asset;

extracting at least one attribute of a plurality of target entities from said target information associated with said asset using a natural language understanding (NLU) model of domain of said at least one unstructured conversation;

generating, using said at least one attribute of said plurality of target entities, a definition of a cohort and at least one criteria for curation of said cohort for said asset by converting said at least one unstructured conversation into a structured target information, wherein said definition of said cohort comprises said at least one attribute of said plurality of target entities; and

determining, using a hypercube estimator, a size and characteristics of said plurality of target entities in said cohort using said definition of said cohort and said at least one criteria for said asset for an engagement activity by (i) identifying a dimensional factor for each of a plurality of spatiotemporal dimensions of said plurality of target entities by processing the at least one unique entity identifier and a timestamp data updated in a geolocation of a key value data structure, and (ii) determining said size of said plurality of target entities in said cohort based on said definition of said cohort, said dimensional factor and a base cardinality of each or combinations of said plurality of spatiotemporal dimensions.

2. The method of claim 1 , wherein said method further comprises:

testing, in real-time, said cohort by running a test engagement activity with at least one unique entity identifier associated with at least one target entity in said plurality of target entities; and

refining, in real-time, said plurality of target entities of said cohort for said asset based on said test engagement activity and curating said plurality of target entities for said engagement activity based on cohort response from said test engagement activity.

3. The method of claim 1 , wherein said at least one attribute of said plurality of target entities and said characteristics of said plurality of target entities of said cohort comprise at least one of a countable attribute, a categorical attribute, an ordinal attribute, a location, a spatial attribute, or a temporal behavior of the plurality of target entities, wherein said cohort comprises metadata of said cohort, wherein said metadata comprises data associated with said at least one of the categorical attribute, the spatial attribute, or the temporal behavior of the plurality of target entities.

4. The method of claim 3 , wherein said countable attribute of said plurality of target entities comprises at least one of (i) spend levels, or (ii) a frequency of visits to a location, wherein said categorical attribute of said plurality of target entities comprises at least one of (i) a gender, (ii) an age-group, (iii) a content, or (iv) a content type, wherein said spatial attribute of said plurality of target entities comprises at least one of (i) residential areas, (ii) regions of interest, or (iii) place categories.

5. The method of claim 1 , wherein said method further comprises training the natural language understanding (NLU) model using at least one of a definition of said cohort data associated with said asset, or a domain information associated with said asset.

6. The method of claim 1 , wherein said method further comprises determining, using a target audience estimator, said size of said plurality of target entities in said cohort based on said definition of said cohort.

7. The method of claim 1 , wherein said plurality of spatiotemporal dimensions comprises a location, individual attributes, a time window or activity or a combination thereof.

8. The method of claim 1 , wherein said method further comprises automatically generating a response for said engagement activity based on said at least one unstructured conversation associated with said asset.

9. One or more non-transitory computer-readable storage medium storing the one or more sequence of instructions for automatically parsing an unstructured conversation using natural language processing to determine target entities for an engagement activity, which when executed by a processor cause:

obtaining at least one conversation associated with an asset between a plurality of entities using a dialogue manager from at least one communication channels;

parsing said at least one unstructured conversation associated with said asset using a natural language processing (NLP) model to obtain a target information associated with said asset;

extracting at least one attribute of a plurality of target entities from said target information associated with said asset using a natural language understanding (NLU) model of domain of said at least one unstructured conversation;

generating, using said at least one attribute of said plurality of target entities, a definition of a cohort and at least one criteria for curation of said cohort for said asset by converting said at least one unstructured conversation into a structured target information, wherein said definition of said cohort comprises said at least one attribute of said plurality of target entities; and

determining, using a hypercube estimator, a size and characteristics of said plurality of target entities in said cohort using said definition of said cohort and said at least one criteria for said asset for an engagement activity by (i) identifying a dimensional factor for each of a plurality of spatiotemporal dimensions of said plurality of target entities by processing the at least one unique entity identifier and a timestamp data updated in a geolocation of a key value data structure, and (ii) determining said size of said plurality of target entities in said cohort based on said definition of said cohort, said dimensional factor and a base cardinality of each or combinations of said plurality of spatiotemporal dimensions.

10. A system for automatically parsing an unstructured conversation using natural language processing to determine target entities for an engagement activity, the system comprising:

a processor;

a memory that stores set of instructions, which when executed by the processor, causes to perform:

obtaining at least one unstructured conversation associated with an asset between a plurality of entities using a dialogue manager from at least one communication channels;

parsing said at least one unstructured conversation associated with said asset using a natural language processing (NLP) model to obtain a target information associated with said asset;

extracting at least one attribute of a plurality of target entities from said target information associated with said asset using a natural language understanding (NLU) model of domain of said at least one unstructured conversation;

generating, using said at least one attribute of said plurality of target entities, a definition of a cohort and at least one criteria for curation of said cohort for said asset by converting said at least one unstructured conversation into a structured target information, wherein said definition of said cohort comprises said at least one attribute of said plurality of target entities; and

determining, using a hypercube estimator, a size and characteristics of said plurality of target entities in said cohort using said definition of said cohort and said at least one criteria for said asset for an engagement activity by (i) identifying a dimensional factor for each of a plurality of spatiotemporal dimensions of said plurality of target entities by processing the at least one unique entity identifier and a timestamp data updated in a geolocation of a key value data structure, and (ii) determining said size of said plurality of target entities in said cohort based on said definition of said cohort, said dimensional factor and a base cardinality of each or combinations of said plurality of spatiotemporal dimensions.

11. The system of claim 10 , wherein said processor further:

tests, in real-time, said cohort by running a test engagement activity with at least one unique entity identifier associated with at least one target entity in said plurality of target entities; and

refines, in real-time, said plurality of target entities of said cohort for said asset based on said test engagement activity and curating said plurality of target entities for said engagement activity based on cohort response from said test engagement activity.

12. The system of claim 10 , wherein said at least one attribute of said plurality of target entities and said characteristics of said plurality of target entities of said cohort comprise at least one of a countable attribute, a categorical attribute, an ordinal attribute, a location, a spatial attribute, or a temporal behavior of an entity, wherein said cohort comprises metadata of said cohort, wherein said metadata comprises data associated with said at least one of the categorical attribute, the spatial attribute, or the temporal behavior of the plurality of target entities.

13. The system of claim 12 , wherein said countable attribute of said plurality of target entities comprises at least one of (i) spend levels, or (ii) a frequency of visits to a location, wherein said categorical attribute of said plurality of target entities comprises at least one of (i) a gender, (ii) an age-group, (iii) a content, or (iv) a content type, wherein said spatial attribute of said plurality of target entities comprises at least one of (i) residential areas, (ii) regions of interest, or (iii) place categories.

14. The system of claim 10 , wherein said processor further trains the natural language understanding (NLU) model using at least one of a definition of said cohort data associated with said asset, or a domain information associated with said asset.

15. The system of claim 10 , wherein said processor further determines, using a target audience estimator, said size of said plurality of target entities in said cohort based on said definition of said cohort.

16. The system of claim 10 , wherein said plurality of spatiotemporal dimensions comprises a location, individual attributes, a time window/activity, or a combination thereof.

17. The system of claim 10 , wherein said processor further automatically generates a response for said engagement activity based on said at least one unstructured conversation associated with said asset.

Assignments (10)

Continuity (1)