Predicting device components for repair and/or replacement using artificial intelligence techniques
Methods, apparatus, and processor-readable storage media for predicting device components for repair and/or replacement using artificial intelligence techniques are provided herein. An example computer-implemented method includes obtaining information pertaining to at least one device defect; defining multiple device component categories related to the device defect(s); determining one or more of the device component categories as associated with the device defect(s) by processing at least a first portion of the information using one or more artificial intelligence techniques; identifying one or more device components associated with at least a second portion of the information; predicting at least one of the identified device component(s), based on comparing the identified device component(s) and the one or more determined device component categories, as needing to be repaired and/or replaced in connection with at least a portion of the device defect(s); and performing one or more automated actions based on the predicting.
1 . A computer-implemented method comprising:
obtaining information pertaining to at least one device defect;
defining multiple device component categories related to the at least one device defect;
determining one or more of the multiple device component categories as associated with the at least one device defect by processing at least a first portion of the obtained information in a processor-based machine learning system comprising at least one tokenization generator, at least one embedding layer having an input and an output, and at least one classification layer having an input coupled to the output of the embedding layer, wherein: (i) the at least one tokenization generator generates a tokenized representation of at least a first portion of the obtained information comprising at least one symptom description associated with the at least one device defect; (ii) the at least one embedding layer receives the tokenized representation at the input thereof and transforms the tokenized representation into a vector representation at the output of the at least one embedding layer; and (iii) the at least one classification layer receives the vector representation at the input of the at least one classification layer and generates, at an output of the at least one classification layer, one or more classification probability values associated with respective ones of the multiple device component categories;
identifying one or more device components associated with at least a second portion of the obtained information;
predicting at least one of the one or more device components associated with the at least a second portion of the obtained information, based at least in part on comparing (i) the one or more device components associated with the at least a second portion of the obtained information and (ii) the one or more device component categories associated with the at least one device defect, as needing to be at least one of repaired and replaced in connection with at least a portion of the at least one device defect, wherein predicting at least one of the one or more device components as needing to be at least one of repaired and replaced comprises:
determining one or more intersections between descriptions of device components associated with the one or more determined device component categories and descriptions of the one or more device components associated with the at least a second portion of the obtained information; and
ranking at least a portion of the one or more intersections based at least in part on a number of one or more terms shared across the at least a portion of the one or more intersections; and
performing one or more automated actions based at least in part on the predicting of the at least one of the one or more device components as needing to be at least one of repaired and replaced, wherein performing one or more automated actions comprises automatically initiating, using one or more automated systems in conjunction with the ranking of the at least a portion of the one or more intersections, at least one of one or more device component repair operations and one or more device component replacement operations;
wherein the method is performed by at least one processing device comprising a processor coupled to a memory.
2 . The computer-implemented method of claim 1 , wherein obtaining information pertaining to at least one device defect comprises obtaining at least one symptom description associated with the at least one device defect and at least one device identifier associated with at least a portion of the at least one device defect.
3 . The computer-implemented method of claim 2 , wherein determining one or more of the multiple device component categories as associated with the at least one device defect comprises processing at least a portion of the at least one symptom description associated with the at least one device defect in the processor-based machine learning system.
4 . The computer-implemented method of claim 2 , wherein identifying one or more device components associated with at least a second portion of the obtained information comprises filtering at least one dataset pertaining to multiple device components related to the at least one device defect using the at least one device identifier associated with the at least a portion of the at least one device defect.
5 . The computer-implemented method of claim 1 , wherein predicting at least one of the one or more device components as needing to be at least one of repaired and replaced comprises ranking at least a portion of the one or more intersections based at least in part on information pertaining to historical frequency of at least one of device component repairs and device component replacements components across the one or more device components identified as associated with the at least a second portion of the obtained information.
6 . The computer-implemented method of claim 1 , wherein determining one or more of the multiple device component categories as associated with the at least one device defect comprises processing the at least a first portion of the obtained information using at least one long short-term memory (LSTM) model.
7 . The computer-implemented method of claim 1 , wherein determining one or more of the multiple device component categories as associated with the at least one device defect comprises processing the at least a first portion of the obtained information using at least one transformers-based model.
8 . The computer-implemented method of claim 1 , wherein determining one or more of the multiple device component categories as associated with the at least one device defect comprises processing the at least a first portion of the obtained information using at least one clustering algorithm in conjunction with one or more deep learning techniques.
9 . The computer-implemented method of claim 1 , wherein performing one or more automated actions comprises automatically training at least a portion of the processor-based machine learning system using feedback related to the predicting of the at least one of the one or more device components as needing to be at least one of repaired and replaced.
10 . A non-transitory processor-readable storage medium having stored therein program code of one or more software programs, wherein the program code when executed by at least one processing device causes the at least one processing device:
to obtain information pertaining to at least one device defect;
to define multiple device component categories related to the at least one device defect;
to determine one or more of the multiple device component categories as associated with the at least one device defect by processing at least a first portion of the obtained information in a processor-based machine learning system comprising at least one tokenization generator, at least one embedding layer having an input and an output, and at least one classification layer having an input coupled to the output of the embedding layer, wherein: (i) the at least one tokenization generator generates a tokenized representation of at least a first portion of the obtained information comprising at least one symptom description associated with the at least one device defect; (ii) the at least one embedding layer receives the tokenized representation at the input thereof and transforms the tokenized representation into a vector representation at the output of the at least one embedding layer; and (iii) the at least one classification layer receives the vector representation at the input of the at least one classification layer and generates, at an output of the at least one classification layer, one or more classification probability values associated with respective ones of the multiple device component categories;
to identify one or more device components associated with at least a second portion of the obtained information;
to predict at least one of the one or more device components associated with the at least a second portion of the obtained information, based at least in part on comparing (i) the one or more device components associated with the at least a second portion of the obtained information and (ii) the one or more device component categories associated with the at least one device defect, as needing to be at least one of repaired and replaced in connection with at least a portion of the at least one device defect, wherein predicting at least one of the one or more device components as needing to be at least one of repaired and replaced comprises:
determining one or more intersections between descriptions of device components associated with the one or more determined device component categories and descriptions of the one or more device components associated with the at least a second portion of the obtained information; and
ranking at least a portion of the one or more intersections based at least in part on a number of one or more terms shared across the at least a portion of the one or more intersections; and
to perform one or more automated actions based at least in part on the predicting of the at least one of the one or more device components as needing to be at least one of repaired and replaced, wherein performing one or more automated actions comprises automatically initiating, using one or more automated systems in conjunction with the ranking of the at least a portion of the one or more intersections, at least one of one or more device component repair operations and one or more device component replacement operations.
11 . The non-transitory processor-readable storage medium of claim 10 , wherein obtaining information pertaining to at least one device defect comprises obtaining at least one symptom description associated with the at least one device defect and at least one device identifier associated with at least a portion of the at least one device defect.
12 . The non-transitory processor-readable storage medium of claim 11 , wherein determining one or more of the multiple device component categories as associated with the at least one device defect comprises processing at least a portion of the at least one symptom description associated with the at least one device defect in the processor-based machine learning system.
13 . The non-transitory processor-readable storage medium of claim 11 , wherein identifying one or more device components associated with at least a second portion of the obtained information comprises filtering at least one dataset pertaining to multiple device components related to the at least one device defect using the at least one device identifier associated with the at least a portion of the at least one device defect.
14 . An apparatus comprising:
at least one processing device comprising a processor coupled to a memory;
the at least one processing device being configured:
to obtain information pertaining to at least one device defect;
to define multiple device component categories related to the at least one device defect;
to determine one or more of the multiple device component categories as associated with the at least one device defect by processing at least a first portion of the obtained information in a processor-based machine learning system comprising at least one tokenization generator, at least one embedding layer having an input and an output, and at least one classification layer having an input coupled to the output of the embedding layer, wherein: (i) the at least one tokenization generator generates a tokenized representation of at least a first portion of the obtained information comprising at least one symptom description associated with the at least one device defect; (ii) the at least one embedding layer receives the tokenized representation at the input thereof and transforms the tokenized representation into a vector representation at the output of the at least one embedding layer; and (iii) the at least one classification layer receives the vector representation at the input of the at least one classification layer and generates, at an output of the at least one classification layer, one or more classification probability values associated with respective ones of the multiple device component categories;
to identify one or more device components associated with at least a second portion of the obtained information;
to predict at least one of the one or more device components associated with the at least a second portion of the obtained information, based at least in part on comparing (i) the one or more device components associated with the at least a second portion of the obtained information and (ii) the one or more device component categories associated with the at least one device defect, as needing to be at least one of repaired and replaced in connection with at least a portion of the at least one device defect, wherein predicting at least one of the one or more device components as needing to be at least one of repaired and replaced comprises:
determining one or more intersections between descriptions of device components associated with the one or more determined device component categories and descriptions of the one or more device components associated with the at least a second portion of the obtained information; and
ranking at least a portion of the one or more intersections based at least in part on a number of one or more terms shared across the at least a portion of the one or more intersections; and
to perform one or more automated actions based at least in part on the predicting of the at least one of the one or more device components as needing to be at least one of repaired and replaced, wherein performing one or more automated actions comprises automatically initiating, using one or more automated systems in conjunction with the ranking of the at least a portion of the one or more intersections, at least one of one or more device component repair operations and one or more device component replacement operations.
15 . The apparatus of claim 14 , wherein obtaining information pertaining to at least one device defect comprises obtaining at least one symptom description associated with the at least one device defect and at least one device identifier associated with at least a portion of the at least one device defect.
16 . The apparatus of claim 15 , wherein determining one or more of the multiple device component categories as associated with the at least one device defect comprises processing at least a portion of the at least one symptom description associated with the at least one device defect in the processor-based machine learning system.
17 . The apparatus of claim 15 , wherein identifying one or more device components associated with at least a second portion of the obtained information comprises filtering at least one dataset pertaining to multiple device components related to the at least one device defect using the at least one device identifier associated with the at least a portion of the at least one device defect.
18 . The apparatus of claim 14 , wherein predicting at least one of the one or more device components as needing to be at least one of repaired and replaced comprises ranking at least a portion of the one or more intersections based at least in part on information pertaining to historical frequency of at least one of device component repairs and device component replacements components across the one or more device components identified as associated with the at least a second portion of the obtained information.
19 . The apparatus of claim 14 , wherein determining one or more of the multiple device component categories as associated with the at least one device defect comprises processing the at least a first portion of the obtained information using at least one long short-term memory (LSTM) model.
20 . The apparatus of claim 14 , wherein determining one or more of the multiple device component categories as associated with the at least one device defect comprises processing the at least a first portion of the obtained information using at least one transformers-based model.