US-12627572-B2 - Recommendation engine with machine learning for guided service management, such as for use with events related to telecommunications subscribers

Abstract

A system can obtain a service request message indicating an event that affects a user device of a telecommunications network. The system can store an indication of the event at a memory device that stores historical information of events and can process the event with a machine learning (ML) model that is trained based on the historical information to simulate the effect of the event on the telecommunications network. The system generates an output for display on a computing device. The output includes a recommendation and identifies an attribute of the event and a predicted value configured to reduce a workload or increase operational efficiency. The recommendation is configured to make the event actionable. The system receives feedback that a reviewer has acted on the recommendation. In response to the feedback, the system can re-train the ML model and configure the process for handling the service request.

Inventors

• Vijay VEGGALAM
• Prabha Jayaram

Assignees

• T-MOBILE USA, INC.

Dates

Publication Date

20260512

Application Date

20230816

Claims (20)

1. 1 . A non-transitory, computer-readable storage medium comprising instructions recorded thereon that, when executed by at least one processor of a system, cause the system to: obtain a service message indicating an event that affects a subscriber of a telecommunications network supported by the system; store an indication of the event at a memory device configured to store historical information; simulate an effect of the event on the telecommunications network based on a machine learning (ML) model, wherein the ML model is trained based on the historical information; generate, based on the ML model, an output for display on a computing device, wherein the output includes one or more data pairs, each data pair including an attribute of the event and a predicted value configured to reduce noise and increase efficiency of the system, wherein the attribute of the event and the predicted value are stored in a dictionary, and wherein each data pair includes a control enabling an alternative selection to accept or reject each data pair; receive the alternative selection from a reviewer to accept or reject a data pair using the control located in the displayed output; adjust a weight associated with the attribute of the event in response to receiving the alternative selection; retrain the ML model using the adjusted weight based on the alternative selection received from the reviewer to generate a subsequent predicted value configured to reduce the noise and increase the efficiency of the system; and update the dictionary, by the re-trained ML model, with the subsequent predicted value based on the feedback from the reviewer.
2. 2 . The non-transitory, computer-readable storage medium of claim 1 , wherein the data pair of the one or more data pairs has been accepted, and wherein the data pair that has been accepted includes an indication that the event is redundant or unimportant.
3. 3 . The non-transitory, computer-readable storage medium of claim 1 , wherein the ML model is configured to: classify the event in accordance with a classification framework for different types of events including alarms, incidents, and problems, wherein an alarm indicates occurrence of an event and is detected by the system, wherein an incident indicates an unplanned interruption or reduction in quality to a service, and wherein a problem indicates a cause of an incident including a particular device or software.
4. 4 . The non-transitory, computer-readable storage medium of claim 1 , wherein the system is further caused to: train the ML model to conform with an existing policy for prioritizing and escalating events that satisfy a criterion.
5. 5 . The non-transitory, computer-readable storage medium of claim 1 : wherein the system is an Information Technology Service Management (ITSM) of a Network Operations Center (NOC), and wherein the service message includes a request to add, delete, or change a software component or hardware component.
6. 6 . The non-transitory, computer-readable storage medium of claim 1 : wherein the telecommunications network comprises a 5G network, and wherein the ML model routes the service message to an assignment group based on having previously resolved an interruption or degradation of the telecommunications network more efficiently than another assignment group of the system.
7. 7 . The non-transitory, computer-readable storage medium of claim 1 , wherein to obtain the service message includes causing the system to: monitor the telecommunications network; detect that the event is anomalous and affects the telecommunications network as experienced by the subscriber; and in response to detecting that the event is anomalous, trigger an alarm for the anomalous event, wherein the service message indicates the alarm, and wherein an attribute for the alarm include a time delay, a view type, or a behavior.
8. 8 . The non-transitory, computer-readable storage medium of claim 1 , wherein the ML model is caused to: analyze a behavior of the event to determine that the event is anomalous and determine whether the anomalous event is malicious or non-malicious; and generate the data pair indicative of a recommendation to ignore the event as an anomalous event that is non-malicious based on the historical information stored at the memory device.
9. 9 . The non-transitory, computer-readable storage medium of claim 1 , wherein to obtain the service message includes causing the system to: detect an incident that affects the subscriber, wherein an attribute for the incident includes a cause, an assignment group, or a resolution to the service message.
10. 10 . The non-transitory, computer-readable storage medium of claim 1 , wherein to obtain the service message includes causing the system to: receive the service message indicating a problem that affects the subscriber, wherein an attribute of the problem includes a time delay after which the event is expected to resolve.
11. 11 . The non-transitory, computer-readable storage medium of claim 1 , wherein the ML model is caused to: continuously adapt to ongoing training based on new events; or replace at least a portion of process after being sufficiently trained based on events.
12. 12 . The non-transitory, computer-readable storage medium of claim 1 , wherein the ML model is caused to: generate the data pair indicative of a recommendation that indicates a particular cause of the event, wherein the particular cause is determined based on prior events having a characteristic in common with the event; or generate the data pair indicative of a recommendation that identifies whether the event warrants action by an assignment group or should be ignored or discarded.
13. 13 . The non-transitory, computer-readable storage medium of claim 1 , wherein the ML model is caused to: generate the data pair indicative of a recommendation to route the service message to an assignment group selected by the ML model from among multiple assignment groups, wherein the assignment group is selected based on prior events having a characteristic in common with the event and being more efficiently processed by the assignment group compared to another assignment group of the multiple assignment groups.
14. 14 . The non-transitory, computer-readable storage medium of claim 1 , wherein the system is further caused to receive a feedback message, and wherein prior to feedback message being received: cause the computing device to display, on an interface, statistical information of the event, wherein the statistical information indicates an effectiveness of resolving the event, a significance of the event, knowledge of the event, or an indication whether the event needs correlation with another event.
15. 15 . The non-transitory, computer-readable storage medium of claim 14 , wherein the system is further caused to, prior to the feedback message being received: cause the computing device to display, on an interface, supporting information for the event, wherein the supporting information includes links to additional information about the event which can be alternatively accepted or rejected.
16. 16 . The non-transitory, computer-readable storage medium of claim 14 , wherein the system is further caused to, prior to the feedback message being received: cause the computing device to display, on a user interface, graphical cue cards including data pairs indicative of recommendations for processing the event and a control element that enables accepting or rejecting all the recommendations.
17. 17 . A method performed by a system, the method comprising: obtaining a message indicating an event that affects a telecommunications network, wherein the message indicates a reduction in a quality of the telecommunications network; storing an indication of the event at a device; simulating the event based on a machine learning (ML) model, wherein the ML model is trained based on indications stored at the device; generating, based on the ML model, an output for display, wherein the output includes data pairs that include attributes of the event and values configured to reduce noise and increase efficiency of the system, wherein the attribute of the event and the predicted value are stored in a dictionary, and wherein the data pairs include controls enabling a selection to accept or reject the data pairs; receiving the selection from a reviewer to accept or reject a data pair using the controls located in the displayed output; adjusting one or more weights associated with the attributes of the event in response to receiving the selection; retraining the ML model using the adjusted one or more weights based on the selection received from the reviewer to generate subsequent predicted values configured to reduce the noise and increase the efficiency of the system; and updating the dictionary, by the re-trained ML model, with the subsequent predicted values based on the selection from the reviewer.
18. 18 . The method of claim 17 , wherein the data pair of the data pairs has been accepted, and wherein the data pair that has been accepted includes an indication that the event is unimportant.
19. 19 . The method of claim 17 further comprising: classifying, by the ML model, the event in accordance with a classification framework for different types of events including alarms, incidents, and problems, wherein an alarm indicates occurrence of an event and is detected by the system, wherein an incident indicates an unplanned interruption or reduction in quality to a service, and wherein a problem indicates a cause of an incident including a particular device or software.
20. 20 . The method of claim 17 further comprising: training the ML model to conform with an existing policy for prioritizing and escalating events that satisfy a criterion.

Description

CROSS-REFERENCE TO RELATED APPLICATIONS This application is a continuation of U.S. patent application Ser. No. 17/515,053, filed on Oct. 29, 2021, entitled RECOMMENDATION ENGINE WITH MACHINE LEARNING FOR GUIDED SERVICE MANAGEMENT, SUCH AS FOR USE WITH EVENTS RELATED TO TELECOMMUNICATIONS SUBSCRIBERS, which is hereby incorporated by reference in its entirety. BACKGROUND Information Technology Service Management (ITSM) relates to activities that are performed by an entity (e.g., telecommunications operator) to design, build, deliver, operate, and control information technology (IT) services. ITSM is characterized by adopting a process approach towards management, focusing on customer needs and IT services for customers rather than IT systems. ITSM processes, especially workflow driven processes, can benefit significantly from being supported with specialized software tools. Core to the tools is a workflow management system for handling events/alarms/faults, incidents, service requests, problems, changes, etc. The ability of these tools to enable easy linking between event, incident, service request, problem, and/or change records with each other and with records of configuration items in a database can be advantageous. A service desk of a telecommunications operator is an example of an ITSM function. The service desk is considered a central point of contact between service providers and users/customers on a day-to-day basis. It is also a focal point for reporting incidents (e.g., disruptions or potential disruptions in service availability or quality) and for users making service requests (e.g., routine requests). A call center or help desk is a type of service desk that provides only a portion of what a service desk offers. Specifically, a service desk has a broader and user-centered approach which is designed to provide a user with an informed single point of contact for all IT requests. A service desk seeks to facilitate the integration of business processes into the service management infrastructure. In addition to actively monitoring and owning incidents and user questions, and providing the communications channel for other service management disciplines with the user community, a service desk also provides an interface for other activities such as customer change requests, third-party requests (e.g., maintenance contracts), and software licensing. BRIEF DESCRIPTION OF THE DRAWINGS Detailed descriptions of implementations of the present invention will be described and explained through the use of the accompanying drawings. FIG. 1 is a block diagram that illustrates a wireless communications system in which at least some operations described herein can be implemented. FIG. 2 is a block diagram of a system that implements service management workflows. FIG. 3 is a block diagram that illustrates a service management system that includes a recommendation engine. FIG. 4 illustrates a service management interface including recommendations to guide users through a workflow that reduces noise and improves operational efficiencies. FIG. 5 illustrates a process performed by a service management system including a recommendation engine with machine learning capabilities to guide users through a workflow. FIG. 6 is a block diagram that illustrates an example of a computer system in which at least some operations described herein can be implemented. The technologies described herein will become more apparent to those skilled in the art from studying the Detailed Description in conjunction with the drawings. Embodiments or implementations describing aspects of the invention are illustrated by way of example, and the same references can indicate similar elements. While the drawings depict various implementations for the purpose of illustration, those skilled in the art will recognize that alternative implementations can be employed without departing from the principles of the present technologies. Accordingly, while specific implementations are shown in the drawings, the technology is amenable to various modifications. DETAILED DESCRIPTION The disclosed technology relates to a recommendation engine that guides service management to reduce noise and improve operational efficiencies. For example, the recommendation engine can guide Information Technology Service Management (ITSM) to reduce noise and improve operational efficiencies of a Network Operations Center (NOC). The recommendation engine leverages machine learning (ML) algorithms to make predictions based on prior service events (“events”) such as alarms, incidents, tickets, problems, changes, etc. related to subscribers of telecommunications services and recommends the future behavior for an alarm, incident, problem or change record. In one example, a telecommunications system implements a service management system that includes a recommendation engine with ML. The service management system can effectively monitor and manage the telecommunications network, host, a