Granted Patent B2

US 12,547,849 · App. 18/447,506 · Granted Feb 10, 2026

Inventors: Naga A. Ayachitula (Dobbs Ferry, NY); Rohit Khandekar (Jersey City, NJ); Upendra Sharma (Hartsdale, NY)

Assignee: Kyndryl, Inc.

G06F40/40G06F40/289

Abstract

Claims (95)

1 . A computer-implemented method for automated abstractive information technology (IT) summary generation, the computer-implemented method being executed by one or more processors and comprising:

arranging a primary computer system in communication with multiple secondary computer systems via a network, each of the multiple secondary computer systems comprising a computer system configured to manage an IT environment of a client and a ticket repository;

receiving, by the primary computer system via the network, numerous IT records, each one of the numerous IT records comprising text and numerical data from the ticket repository of one of the secondary computer systems;

normalizing, by the primary computer system for each one of the numerous IT records, the numerical data into normalized numerical data;

transforming, by the primary computer system for each one of the numerous IT records, the normalized numerical data into comparative and superlative adjectival terms; and

rewriting, by the primary computer system for each one of the numerous IT records, the text to include the comparative and superlative adjectival terms for output as a rewritten IT record;

generating, by the primary computer system for the client of the one of the secondary computer systems, a summary report of the output as the rewritten IT record for each of the numerous IT records,

wherein the one or more processors use at least one of artificial intelligence (AI) and machine learning (ML) to execute at least the normalizing, the transforming and the rewriting.

2 . The computer-implemented method according to claim 1 , wherein:

the computer-implemented method further comprises performing sentiment analysis on the text of the IT record, and

the rewriting of the text comprises accounting for results of the sentiment analysis.

3 . The computer-implemented method according to claim 2 , wherein the performing of the sentiment analysis comprises:

preprocessing of the text into tokenized text by removing irrelevant information, special characters and stopwords;

analyzing the tokenized text to determine a sentiment associated with each phrase in the tokenized text; and

aggregating the sentiment associated with each phrase in the tokenized text to determine an overall sentiment of the text.

4 . The computer-implemented method according to claim 2 , further comprising performing a coreference resolution with respect to the rewritten IT record to generate a summary report to be output,

wherein the performing of the coreference resolution comprises:

receiving an input of text of the rewritten IT record;

tokenizing of the text of the rewritten IT record;

applying a coreference resolution model to identify and link words and phrases in the rewritten IT record that refer to a same entity; and

outputting indications of the words and phrases in the rewritten IT record that refer to the same entity so that the summary report can include those indications.

5 . The computer-implemented method according to claim 4 , wherein the one or more processors use the at least one of the AI and the ML to execute the performing of the sentiment analysis and the performing of the coreference resolution.

6 . The computer-implemented method according to claim 1 , wherein:

the normalizing comprises min-max scaling, z-score standardization and decimal scaling, and

in normalization by min-max scaling, values are scaled to a fixed range and a scaled_value=(value−min_value)/(max_value−min_value) ensuring that a minimum value in a dataset is mapped to 0 and a maximum value in the dataset is mapped to 1 and preserving relative differences between other values,

in normalization by z-score standardization, z-score standardization transforms values by subtracting a mean and dividing by a standard deviation and z-score=(value-mean)/standard_deviation ensuring that standardized values have a mean of 0 and a standard deviation of 1 and allowing for comparisons based on a number of standard deviations a value is away from the mean, and,

in normalization by decimal scaling, a decimal point of values is shifted to create new scaled values where a scaling factor is determined by a largest absolute value among a dataset and a scaled_value=value/(10{circumflex over ( )}scaling_factor) resulting in values with a reduced range.

7 . The computer-implemented method according to claim 1 , wherein the transforming comprises;

threshold-based comparing in which thresholds or ranges are set based on normalized or scaled values and comparative or superlative adjectives are assigned accordingly;

relative comparing in which the normalized or scaled values are compared with a reference point or benchmark; and

mapping to predefined categories in which the normalized or scaled values are mapped to predefined categories or adjective levels.

8 . A system comprising:

a memory having computer readable instructions; and

one or more processors for executing the computer readable instructions, the computer readable instructions controlling the one or more processors to perform operations for automated abstractive information technology (IT) summary generation comprising:

arranging a primary computer system in communication with multiple secondary computer systems via a network, each of the multiple secondary computer systems comprising a computer system configured to manage an IT environment of a client and a ticket repository;

receiving, by the primary computer system via the network, numerous IT records, each one of the numerous IT records comprising text and numerical data from the ticket repository of one of the secondary computer systems;

normalizing, by the primary computer system for each one of the numerous IT records, the numerical data into normalized numerical data;

transforming, by the primary computer system for each one of the numerous IT records, the normalized numerical data into comparative and superlative adjectival terms; and

rewriting, by the primary computer system for each one of the numerous IT records, the text to include the comparative and superlative adjectival terms for output as a rewritten IT record;

generating, by the primary computer system for the client of the one of the secondary computer systems, a summary report of the output as the rewritten IT record for each of the numerous IT records,

wherein the one or more processors use at least one of artificial intelligence (AI) and machine learning (ML) to execute at least the normalizing, the transforming and the rewriting.

9 . The system according to claim 8 , wherein:

the operations further comprise performing sentiment analysis on the text of the IT record, and

the rewriting of the text comprises accounting for results of the sentiment analysis.

10 . The system according to claim 9 , wherein the performing of the sentiment analysis comprises:

preprocessing of the text into tokenized text by removing irrelevant information, special characters and stopwords;

analyzing the tokenized text to determine a sentiment associated with each phrase in the tokenized text; and

aggregating the sentiment associated with each phrase in the tokenized text to determine an overall sentiment of the text.

11 . The system according to claim 9 , wherein the operations further comprise performing a coreference resolution with respect to the rewritten IT record to generate a summary report to be output,

wherein the performing of the coreference resolution comprises:

receiving an input of text of the rewritten IT record;

tokenizing of the text of the rewritten IT record;

applying a coreference resolution model to identify and link words and phrases in the rewritten IT record that refer to a same entity; and

outputting indications of the words and phrases in the rewritten IT record that refer to the same entity so that the summary report can include those indications.

12 . The system according to claim 11 , wherein the one or more processors use the at least one of the AI and the ML to execute the performing of the sentiment analysis and the performing of the coreference resolution.

13 . The system according to claim 8 , wherein:

the normalizing comprises min-max scaling, z-score standardization and decimal scaling, and

in normalization by min-max scaling, values are scaled to a fixed range and a scaled_value=(value−min_value)/(max_value−min_value) ensuring that a minimum value in a dataset is mapped to 0 and a maximum value in the dataset is mapped to 1 and preserving relative differences between other values,

in normalization by z-score standardization, z-score standardization transforms values by subtracting a mean and dividing by a standard deviation and z-score=(value−mean)/standard_deviation ensuring that standardized values have a mean of 0 and a standard deviation of 1 and allowing for comparisons based on a number of standard deviations a value is away from the mean, and,

in normalization by decimal scaling, a decimal point of values is shifted to create new scaled values where a scaling factor is determined by a largest absolute value among a dataset and a scaled_value=value/(10{circumflex over ( )}scaling_factor) resulting in values with a reduced range.

14 . The system according to claim 8 , wherein the transforming comprises:

threshold-based comparing in which thresholds or ranges are set based on normalized or scaled values and comparative or superlative adjectives are assigned accordingly;

relative comparing in which the normalized or scaled values are compared with a reference point or benchmark; and

mapping to predefined categories in which the normalized or scaled values are mapped to predefined categories or adjective levels.

15 . A computer program product comprising a computer readable storage medium having program instructions embodied therewith, the program instructions executable by one or more processors to cause the one or more processors to perform operations for automated abstractive information technology (IT) summary generation comprising:

arranging a primary computer system in communication with multiple secondary computer systems via a network, each of the multiple secondary computer systems comprising a computer system configured to manage an IT environment of a client and a ticket repository;

receiving, by the primary computer system via the network, numerous IT records, each one of the numerous IT records comprising text and numerical data from the ticket repository of one of the secondary computer systems;

normalizing, by the primary computer system for each one of the numerous IT records, the numerical data into normalized numerical data;

transforming, by the primary computer system for each one of the numerous IT records, the normalized numerical data into comparative and superlative adjectival terms; and

rewriting, by the primary computer system for each one of the numerous IT records, the text to include the comparative and superlative adjectival terms for output as a rewritten IT record;

generating, by the primary computer system for the client of the one of the secondary computer systems, a summary report of the output as the rewritten IT record for each of the numerous IT records,

wherein the one or more processors use at least one of artificial intelligence (AI) and machine learning (ML) to execute at least the normalizing, the transforming and the rewriting.

16 . The computer program product according to claim 15 , wherein:

the operations further comprise performing sentiment analysis on the text of the IT record, and

the rewriting of the text comprises accounting for results of the sentiment analysis.

17 . The computer program product according to claim 16 , wherein the performing of the sentiment analysis comprises:

preprocessing of the text into tokenized text by removing irrelevant information, special characters and stopwords;

analyzing the tokenized text to determine a sentiment associated with each phrase in the tokenized text; and

aggregating the sentiment associated with each phrase in the tokenized text to determine an overall sentiment of the text.

18 . The computer program product according to claim 16 , wherein the operations further comprise performing a coreference resolution with respect to the rewritten IT record to generate a summary report to be output,

wherein the performing of the coreference resolution comprises:

receiving an input of text of the rewritten IT record;

tokenizing of the text of the rewritten IT record;

applying a coreference resolution model to identify and link words and phrases in the rewritten IT record that refer to a same entity; and

outputting indications of the words and phrases in the rewritten IT record that refer to the same entity so that the summary report can include those indications.

19 . The computer program product according to claim 18 , wherein the one or more processors use the at least one of the AI and the ML to execute the performing of the sentiment analysis and the performing of the coreference resolution.

20 . The computer program product according to claim 15 , wherein:

the normalizing comprises min-max scaling, z-score standardization and decimal scaling, and

in normalization by min-max scaling, values are scaled to a fixed range and a scaled_value=(value−min_value)/(max_value−min_value) ensuring that a minimum value in a dataset is mapped to 0 and a maximum value in the dataset is mapped to 1 and preserving relative differences between other values,

in normalization by z-score standardization, z-score standardization transforms values by subtracting a mean and dividing by a standard deviation and z-score=(value−mean)/standard_deviation ensuring that standardized values have a mean of 0 and a standard deviation of 1 and allowing for comparisons based on a number of standard deviations a value is away from the mean, and,

in normalization by decimal scaling, a decimal point of values is shifted to create new scaled values where a scaling factor is determined by a largest absolute value among a dataset and a scaled_value=value/(10{circumflex over ( )}scaling_factor) resulting in values with a reduced range.

21 . The computer program product according to claim 15 , wherein the transforming comprises:

threshold-based comparing in which thresholds or ranges are set based on normalized or scaled values and comparative or superlative adjectives are assigned accordingly;

relative comparing in which the normalized or scaled values are compared with a reference point or benchmark; and

mapping to predefined categories in which the normalized or scaled values are mapped to predefined categories or adjective levels.

Assignments (1)

Continuity (1)

References Cited (16)