US 12,282,383 B2
Apparatuses, methods, and computer program products for ML assisted service risk analysis of unreleased software code
Karthik Muralidharan, Bangalore (IN); Shashank Prasad Rao, Noida (IN); and Sri Vardhamanan A, Chennai (IN)
Assigned to ATLASSIAN PTY LTD., Sydney (AU); and ATLASSIAN US, INC., San Francisco, CA (US)
Filed by ATLASSIAN PTY LTD., Sydney (AU); and ATLASSIAN, INC., San Francisco, CA (US)
Filed on Sep. 27, 2021, as Appl. No. 17/449,042.
Prior Publication US 2023/0095634 A1, Mar. 30, 2023
Int. Cl. G06F 11/07 (2006.01); G06F 8/71 (2018.01); G06F 11/00 (2006.01); G06F 11/30 (2006.01); G06N 20/00 (2019.01)
CPC G06F 11/0784 (2013.01) [G06F 8/71 (2013.01); G06F 11/004 (2013.01); G06F 11/301 (2013.01); G06N 20/00 (2019.01)] 19 Claims
OG exemplary drawing
 
1. A computer-implemented method, comprising:
receiving a service risk analysis request associated with an unreleased code object;
extracting, using a service risk analysis layer, one or more service risk analysis attributes based at least in part on the unreleased code object, wherein the one or more service risk analysis attributes are associated with a service risk analysis metadata, and the service risk analysis metadata describes at least a source repository identifier and a destination repository identifier;
generating, using a service risk analysis machine learning model, a service risk analysis score data object based at least in part on the one or more service risk analysis attributes, wherein the service risk analysis score data object comprises (i) at least one service risk analysis score indicative of a probability that the unreleased code object will cause one or more alerts in one or more associated microservices and (ii) one or more predicted alert types, wherein the one or more predicted alert types are indicative of one or more mitigating actions which, when executed, reduce a probability that the unreleased code object will cause one or more alerts in one or more associated microservices, and wherein the service risk analysis machine learning model is trained based at least in part on a plurality of service alert data objects comprising one or more alert types; and
outputting the service risk analysis score data object.