Machine-trainable automated-script customization

Patent metadata

Field	Value
Publication number	US-11086711-B2
Application number	US-201816139480-A
Country	US
Kind code	B2
Filing date	Sep 24, 2018
Priority date	Sep 24, 2018
Publication date	Aug 10, 2021
Grant date	Aug 10, 2021

Abstract

A cognitive automation engine receives notice that an unexpected event has occurred in a computing environment. The engine tries to address any resulting problems by running a previously generated automation script, customizing the script as required through cognitive means. If this fails, the engine forwards the script to a human expert for customization. In either case, the engine records any customization activities, extracts parameters from the recording that identify each customization step, cognitively assigns a level of risk to each step based on historical precedent, and determines whether running the customized script presents an unacceptable risk of adverse results. The system adds the revisions, other script-related information, and any results of running the revised script Loin a training corpus. The corpus is then incorporated into a machine-learning procedure that teaches the automation engine how to more intelligently customize a script the next time a similar event occurs.

First claim

What is claimed is:

1. A cognitive automation-engine system comprising a processor, a memory coupled to the processor, and a computer-readable hardware storage device coupled to the processor, the storage device containing program code configured to be run by the processor via the memory to implement a method for machine-trainable automated-script customization, the method comprising:
the processor initiating a customization procedure that comprises identifying customization steps capable of refining a selected automation script into a customized script that more specifically addresses a disruption to an operation of a computing environment,
where the selected automation script is selected from a library of existing scripts, and
where the identifying comprises:
the processor directing the customization-recording module to record actions, by a human expert, taken during the course of the customization procedure to address the disruption; and
the processor directing the customization component to intelligently infer the customization steps from the recording and from a comparison of the selected automation script to a candidate customized script that would be generated by applying, to the selected automation script, recorded actions performed to address the disruption;

the processor generating a customized script by applying the customization steps to the selected automation script;
the processor updating a machine-learning corpus with a characterization of the customized script; and
the processor, by submitting the updated corpus to a machine-learning training component of the automation engine, training a customization component of the automation engine to intelligently customize automation scripts.

2. The system of claim 1 , further comprising:
the processor identifying a relative risk of each customization step, where the relative risk of a first customization step identifies a relative likelihood that running a candidate automation script comprising an instruction associated with the first customization step would adversely affect operation of the computing environment;
the processor deriving, as a function of the relative risks of all customization steps, an aggregate risk that running the customized script would adversely affect operation of the computing environment;
the processor determining whether the aggregate risk exceeds a threshold acceptable level of risk;
the processor responding to a determination that the aggregate risk does not exceed the threshold risk by running the customized script, adding the customized script to a script library, and directing a customization-recording module of the automation engine to record a result produced by running the customized script; and
the processor responding to a determination that the aggregate risk does exceed the threshold risk by:
requesting extrinsic authorization to run the customized script, and
if receiving the extrinsic authorization, running the customized script, adding the customized script to the script library, and directing the customization-recording module to record a result produced by running the customized script.

3. The system of claim 1 , where the initiating the customization procedure comprises:
the processor directing a human expert to perform actions that address the disruption.

4. The system of claim 1 , where the initiating the customization procedure comprises:
the processor directing the customization component to automatically identify the customization steps without human intervention; and
the processor, if determining that the customization component is unable to automatically identify the customization steps, directing a human expert to perform actions that address the disruption.

5. The system of claim 1 ,
where machine-learning training sessions have trained the customization component to intelligently identify the relative likelihood that running the candidate automation script would adversely affect operation of the computing environment,
where the intelligent identification is performed as a function of historical records comprised by one or more corpora submitted to the customization component during the machine-learning training sessions, and
where the historical records indicate how often running a script comprising the instruction has in the past produced a result that adversely affected operation of the computing environment.

6. The system of claim 1 , where values of the aggregate risk and of each relative risk are selected from the group consisting of: Unacceptably High Risk, High Risk, Moderate Risk, Low Risk, and No Risk.

7. The system of claim 1 , where the updating further comprises adding, to the corpus, identifications of items selected from the group consisting of:
an unexpected event that created the disruption, the selected automation script, the customized script, the customization steps, the aggregate risk, the relative risks, and any results produced by running the customized script.

8. A method for machine-trainable automated-script customization, the method comprising:
a processor of a cognitive automation engine initiating a customization procedure that comprises identifying customization steps capable of refining a selected automation script into a customized script that more specifically addresses a disruption to an operation of a computing environment,
where the selected automation script is selected from a library of existing scripts, and
where the identifying comprises:

the processor directing the customization-recording module to record actions, by a human expert, taken during the course of the customization procedure to address the disruption; and
the processor directing the customization component to intelligently infer the customization steps from the recording and from a comparison of the selected automation script to a candidate customized script that would be generated by applying, to the selected automation script, recorded actions performed to address the disruption;

the processor generating a customized script by applying the customization steps to the selected automation script;
the processor updating a machine-learning corpus with a characterization of the customized script; and
the processor, by submitting the updated corpus to a machine-learning training component of the automation engine, training a customization component of the automation engine to intelligently customize automation scripts.

9. The method of claim 8 , where the initiating the customization procedure comprises:
the processor directing a human expert to perform actions that address the disruption.

10. The method of claim 8 , where the initiating the customization procedure comprises:
the processor directing the customization component to automatically identify the customization steps without human intervention; and
the processor, if determining that the customization component is unable to automatically identify the customization steps, directing a human expert to perform actions that address the disruption.

11. The method of claim 8 ,
where machine-learning training sessions have trained the customization component to intelligently identify the relative likelihood that running the candidate automation script would adversely affect operation of the computing environment,
where the intelligent identification is performed as a function of historical records comprised by one or more corpora submitted to the customization component during the machine-learning training sessions, and
where the historical records indicate how often running a script comprising the instruction has in the past produced a result that adversely affected operation of the com

Assignees

• IBM

Inventors

• Shwartz Larisa
• Herrero Victor R
• Guerra Victoria Lee
• Coronel Jose Rizaldy

Classifications

• G06F8/70

  Software maintenance or management · CPC title

• G06F11/0793 Primary

  Remedial or corrective actions (recovery from an exception in an instruction pipeline G06F9/3861; by retry G06F11/1402; for recovering from a failure of a protocol instance or entity H04L69/40) · CPC title

• G06F2201/86

  Event-based monitoring · CPC title

• G06F11/302

  where the computing system component is a software system · CPC title

• G06F11/3058

  Monitoring arrangements for monitoring environmental properties or parameters of the computing system or of the computing system component, e.g. monitoring of power, currents, temperature, humidity, position, vibrations (thermal management in cooling arrangements of a computing system G06F1/206) · CPC title