Systems and methods for generating codes and code books using cosine proximity
Certain aspects of the disclosure provide a system and method for generating codes associated with clusters of sentiments, comprising receiving a text input, partitioning the text input into one or more segments, generating one or more numerical vectors associated with each of the one or more segments, comparing the one or more numerical vectors to generate a plurality of cosine proximity values associated with the one or more numerical vectors, applying a clustering algorithm to the one or more numerical vectors to generate clusters of segments within one or more cosine proximity ranges, generating one or more codes associated with each of the clusters of segments within the one or more cosine proximity ranges, wherein each cluster represents an overall sentiment, and netting the one or more codes into one or more categories by inputting the one or more codes into a large-language model.
1 . A computer-implemented method for generating codes, comprising:
receiving a text input comprising one or more customer reviews of consumer products and services;
partitioning the text input into one or more segments;
generating one or more numerical vectors associated with each of the one or more segments;
comparing the one or more numerical vectors to generate, using a pre-trained machine learning model comprising a large-language model, a plurality of cosine proximity values associated with the one or more numerical vectors;
applying a clustering algorithm to the one or more numerical vectors to generate clusters of segments within one or more cosine proximity ranges;
generating, using the pre-trained machine learning model, one or more codes associated with each of the clusters of segments within the one or more cosine proximity ranges, wherein each code is a category related to one of the one or more customer reviews;
displaying the one or more codes on a graphical user interface;
netting each of the one or more codes into one or more super-categories by inputting the one or more codes into the large-language model, wherein the large language model provides the one or more super-categories as an output, and each super-category comprises at least one code of the one or more codes; and
displaying the one or more super-categories on the graphical user interface, wherein each super-category is configured to be expanded to display corresponding at least one code.
2 . The method of claim 1 , wherein the cosine proximity values indicate a level of semantic proximity between segments.
3 . The method of claim 1 , wherein the text input is comprised of customer comments of the consumer products and services.
4 . The method of claim 1 , wherein each of the one or more segments comprises a distinct idea.
5 . The method of claim 1 , wherein partitioning the text input uses natural language processing.
6 . The method of claim 1 , wherein partitioning the text input uses artificial intelligence.
7 . The method of claim 1 , wherein partitioning the text input uses the large-language model.
8 . The method of claim 1 , wherein the one or more numerical vectors are generated using a Bidirectional Encoder Representations from Transformers algorithm.
9 . The method of claim 1 , wherein the clustering algorithm is based on criteria selected by the user using the graphical user interface.
10 . The method of claim 1 , wherein the user can perform drag and drop operations on the one or more codes using the graphical user interface.
11 . A computer-program product for grouping customer comments, comprising:
instructions stored on a memory and executable by a computer processor to cause the non-transitory computer-program product to:
receive a text input comprising one or more customer reviews of consumer products and services;
display a graphical user interface comprising the text input;
accept a user command on the graphical user interface that causes the computer-program product to:
partition the text input into one or more segments;
generate one or more numerical vectors associated with each of the one or more segments;
compare the one or more numerical vectors to generate, using a pre-trained machine learning model comprising a large-language model, a plurality of cosine proximity values associated with the one or more numerical vectors;
apply a clustering algorithm to the one or more numerical vectors to generate one or more clusters of segments within one or more cosine proximity ranges, wherein each of the one or more clusters of segments corresponds with a sentiment;
generate, using the pre-trained machine learning model, one or more groups of customer comments associated with each of the clusters of segments within the one or more cosine proximity ranges, wherein each group is associated with a category related to the one or more customer comments and customer reviews; and
display the one or more groups of customer comments to a user on the graphical user interface;
net each of the one or more codes into one or more super-categories by inputting the one or more codes into the large-language model, wherein the large language model provides the one or more super-categories as an output, and each super-category comprises at least one code of the one or more codes; and
display the one or more super-categories on the graphical user interface, wherein each super-category is configured to be expanded to display corresponding at least one code.
12 . The computer-program product of claim 11 , wherein the graphical user interface comprises a coding section, a display section, and an export section.
13 . The computer-program product of claim 12 , wherein the coding section comprises criteria determined by the user using the graphical user interface.
14 . The computer-program product of claim 13 , wherein the criteria generates a codebook that forms a basis for implementation of a clustering algorithm.
15 . The computer-program product of claim 11 , wherein the user can perform drag and drop operations on the one or more groups of customer comments using the graphical user interface.
16 . A system for categorizing codes based on sentiment, comprising:
a processor;
memory coupled with the processor;
instructions stored in the memory and executable by the processor to cause the system to:
receive a text input comprising one or more customer comments and customer reviews of products and services;
partition the text input into one or more segments;
generate one or more numerical vectors associated with each of the one or more segments;
compare the one or more numerical vectors to generate, using a pre-trained machine learning model comprising a large-language model, a plurality of cosine proximity values associated with the one or more numerical vectors;
apply a clustering algorithm to the one or more numerical vectors to generate one or more clusters of segments within one or more cosine proximity ranges, wherein each of the one or more clusters of segments corresponds with a sentiment;
generate, using the pre-trained machine learning model, one or more codes associated with each of the clusters of segments within the one or more cosine proximity ranges, wherein each code is a category related to one of the one or more customer comments and customer reviews;
display the one or more codes to a user on a graphical user interface;
net each of the one or more codes into one or more super-categories by inputting the one or more codes into the large-language model, wherein the large language model provides the one or more super-categories as an output, and each super-category comprises at least one code of the one or more codes; and
display the one or more super-categories on the graphical user interface, wherein each super-category is configured to be expanded to display corresponding at least one code.
17 . The system of claim 16 , wherein the cosine proximity values indicate a level of semantic proximity between segments.
18 . The system of claim 16 , wherein partitioning the text input uses natural language processing, artificial intelligence, or the large language model.
19 . The system of claim 16 , wherein the one or more numerical vectors are generated using a Bidirectional Encoder Representations from Transformers algorithm comprising a machine-learning framework.
20 . The system of claim 16 , wherein the user can perform drag and drop operations on the one or more codes using the graphical user interface.