Automated discovery of migration notes

A system for performing a system migration in a computing system, comprising:

• a memory;
• and
• at least one processor coupled to the memory and configured to:
• receive a note from the computing system, wherein the received note fails to provide a solution a migration incident, and wherein the migration incident occurs during the system migration;
• extract incident data associated with the received note, wherein the incident data comprises a data object of the migration incident;
• convert the data object of the migration incident into a description of the migration incident, wherein to convert the data object, the at least one processor is configured to:
• concatenate a title field of the data object and a detailed description field of the data object into a concatenated description;
• recognize, using a trained noise model, at least one word within the concatenated description as noise based on a similarity between the at least one word and labeled noise data, wherein the trained noise model is configured to remove noise from the incident data using the labeled noise data, and wherein the labeled noise data comprises text unrelated to the migration incident from prior received notes;
• remove the at least one word from the concatenated description, thereby creating a filtered description of the incident data;
• tokenize the filtered description into a set of tokens;
• and
• store the set of tokens in the description of the incident;
• generate a trained machine learning model to determine whether the incident data is related to the migration incident, the trained machine learning model generated by calculating, based on the description of the migration incident, a degree of proximity between the incident data and labeled-incident data based on an average distance between the labeled-incident data and a number of neighbor-labeled data identified as migration data, wherein the labeled-incident data indicates whether the incident data is related to the migration incident of the computing system undergoing the migration, wherein the degree of proximity is calculated by:
• transforming labeled-incident data into a vectorized representation corresponding to the labeled-incident data;
• clustering, using a clustering algorithm, the neighbor-labeled data based on an accuracy metric, wherein the accuracy metric identifies a precision in which the neighbor-labeled data is related to the migration;
• determining, using the accuracy metric, the average distance between the labeled-incident data and the neighbor-labeled data by comparing the vectorized representation of the labeled-incident data with a vectorized representation of the neighbor-labeled data, wherein the comparing outputs a score corresponding to the average distance;
• and
• determining whether the score corresponding to the average distance is less than a threshold distance, thereby calculating the degree of proximity;
• determine, using the generated trained machine learning model, that the incident data is related to the migration incident based on the calculated degree of proximity;
• based on the determination that the incident data is related to the migration incident, extract, from a plurality of notes previously received from the computing system and using a mapping identifier of the data object, a migration note associated with the migration incident, wherein the migration note comprises one or more actions to be taken towards a system failure experienced by the computing system;
• determine, by applying relevancy criteria to the migration note, that the one or more actions of the migration note provides the solution for the migration incident of the computing system;
• and
• continue the system migration of the computer system by resolving the migration incident for the migration, wherein the resolving comprises causing execution of the one or more actions to address the system failure of the computing system.

A non-transitory computer-readable medium for performing a system migration in a computing system, having instructions stored thereon that, when executed by at least one computing device, cause the at least one computing device to perform operations comprising:

• receiving, by at least one processor, a note from the computing system, wherein the received note fails to provide a solution for a migration incident, and wherein the migration incident occurs during the system migration;
• extracting incident data associated with the received note, wherein the incident data comprises a data object of the migration incident;
• converting the data object of the migration incident into a description of the migration incident, wherein the converting further comprises:
• concatenating a title field of the data object and a detailed description field of the data object into a concatenated description;
• recognizing, using a trained noise model, at least one word within the concatenated description as noise based on a similarity between the at least one word and labeled noise data, wherein the trained noise model is configured to remove noise from the incident data using the labeled noise data, and wherein the labeled noise data comprises text unrelated to the migration incident from prior received notes;
• removing the at least one word from the concatenated description, thereby creating a filtered description of the incident data;
• tokenizing the filtered description into a set of tokens;
• and
• storing the set of tokens in the description of the migration incident;
• generating a trained machine learning model to determine whether the incident data is related to the migration incident, the trained machine learning model generated by calculating, based on the description of the migration incident, a degree of proximity between the incident data and labeled-incident data based on an average distance between the labeled-incident data and a number of neighbor-labeled data identified as migration data, wherein the labeled-incident data indicates whether the incident data is related to the migration incident of the computing system undergoing the migration, wherein the degree of proximity is calculated by:
• transforming labeled-incident data into a vectorized representation corresponding to the labeled-incident data;
• clustering, using a clustering algorithm, the neighbor-labeled data based on an accuracy metric, wherein the accuracy metric identifies a precision in which the neighbor-labeled data is related to the migration;
• determining, using the accuracy metric, the average distance between the labeled-incident data and the neighbor-labeled data by comparing the vectorized representation of the labeled-incident data with a vectorized representation of the neighbor-labeled data, wherein the comparing outputs a score corresponding to the average distance;
• and
• determining whether the score corresponding to the average distance is less than a threshold distance, thereby calculating the degree of proximity;
• determining, using the trained machine learning model, that the incident data is related to the migration incident based on the calculated degree of proximity;
• based on the determining that the incident data is related to the migration incident, extracting, from a plurality of notes previously received from the computing system and using a mapping identifier of the data object, a migration note associated with the migration incident, wherein the migration note comprises one or more actions to be taken towards a system failure experienced by the computing system;
• determining, by applying relevancy criteria to the migration note, that the one or more actions of the migration note provides the solution for the migration incident of the computing system;
• and
• continuing the system migration of the computer system by resolving the migration incident for the migration, wherein the resolving comprises causing execution of the one or more actions to address the system failure of the computing system.