CPC G06F 11/0793 (2013.01) [G06F 18/2155 (2023.01); G06F 18/22 (2023.01); G06F 18/2415 (2023.01); G06N 5/022 (2013.01); G06N 5/04 (2013.01)] | 20 Claims |
1. A method executed by one or more processors, comprising:
receiving, by a support module, a support case indicating a problem associated with a networked storage environment, the support module implemented as a processor executable application programming interface (API) that is presented to one or more computing devices and a storage system of the networked storage environment, the one or more computing devices configured to access one or more storage devices of the storage system via a network connection;
extracting, by the support module, a feature vector from information included in the support case; predicting, by the support module using a training model, a problem category for the support case based on the feature vector; assigning, by the support module, a virtual problem space for the predicted problem category;
identifying, by the support module, from the training model, a plurality of previously processed proximate support cases based on a comparison of a distance between the support case and the proximate support cases within the virtual problem space to a configurable threshold distance value, the configurable threshold distance value varies based on a density of previously processed support cases around the support case and a total number of the previously processed support cases; determining, by the support module, from the training model, relevance of each identified proximate support case to the support case, based on P(E1∩E2)=P(E1)·P(E2), where E1 is a most recently used resolution code event of each identified proximate case, E2 is a most frequently used resolution code event for each identified proximate case, P(E1) is equal to a number of resolution codes that are less recent than a given resolution code divided by a total number of resolution codes within the assigned virtual problem space and P(E2) is equal to a number of resolution codes less frequent than a given resolution code divided by the total number of resolution codes within the assigned virtual problem space; outputting, by the support module, a resolution code for the support case based on the determined relevance of each proximate support case;
using the resolution code to resolve the support case by automatically modifying a configuration associated with a networked storage environment entity impacted by the problem;
and iteratively updating, by the support module, the training model with the resolution code and resolution codes of new support cases received and resolved after the support case using the training model.
|