112b. As additional simulated scenarios are generated and executed, additional training data may be generated and used to update, validate, re-train, or the like, the machine learning model. Accordingly, the abnormality analyzer 112d may detect or predict abnormalities in a deployed application, as well as other applications that might be impacted (e.g., upstream and/or downstream applications).

For instance, while it might not be possible (e.g., given resource constraints) to constantly monitor every interaction between deployed applications, the abnormality analyzer 112d may rely on patterns of abnormalities to predict issues and impacted applications. In one example, if a certificate issue led to a blank page in Application A but not a system failure, the abnormality analyzer 112d (e.g., via the machine learning engine) may predict a likelihood that the abnormality happened in a particular application and may identify other applications that may be impacted.

If abnormalities are identified in deployed applications, the simulation generation and abnormality detection computing platform 110 may attempt to resolve the abnormality using the abnormality solver 112e. In some examples, machine learning may be used to analyze the deployed applications to systems to identify any abnormalities. For instance, machine learning engine 112f may execute a machine learning model configured to receive, as inputs, identified abnormalities (e.g., from the simulations and stored in the database 112c), as well as the deployed applications or systems, parameters of the deployed applications or systems or the like. Execution of the model may then output one or more abnormalities detected in the deployed applications or systems. In some examples, the machine learning model may be trained using historical abnormality data.

Abnormality solver 112e may store instructions and/or data that may cause or enable the simulation generation and abnormality detection computing platform 110 to retrieve, from the database 112c, one or more resolutions to identified abnormalities and automatically execute those resolution(s) to address any abnormalities detected in the deployed version of the application or system. For instance, the abnormality solver may update fonts, modify color schemes, reduce or compress image or video size, remove videos from a page, or the like, based on resolutions stored in the database 112c. In some examples, machine learning may be used to identify a potential resolution to a detected abnormality. For instance, machine learning engine 112f may include a machine learning model configured to receive, as inputs, the detected abnormality and, upon execution of the model, output one or more potential resolutions. In some examples, the machine learning model may be trained using historical resolution data.

In some examples, the abnormality solver 112e may include a use case prioritization engine. The use case prioritization engine may evaluate usage of an application or system in which an abnormality is detected and prioritize resolution (e.g., by the abnormality solver 112e) of applications or systems having greater impact (e.g., if 80% of users of Application A are getting a blank page when they try to load the application, and 40% of users have an illegible font on Application B, the use case prioritization engine may prioritize the abnormality in Application A for resolution by the abnormality solver 112e).

Database 112c may store data related to simulated scenarios, connectivity (e.g., between applications, within applications, or the like), certificates for various sites, detected abnormalities, successful resolutions, and/or other data that enables performance of aspects described herein by the simulation generation and abnormality detection computing platform 110.

FIGS. 2A-2D depict one example illustrative event sequence for implementing simulation generation and abnormality detection functions in accordance with one or more aspects described herein. The events shown in the illustrative event sequence are merely one example sequence and additional events may be added, or events may be omitted, without departing from the invention. Further, one or more processes discussed with respect to FIGS. 2A-2D may be performed in real-time or near real-time.

With reference to FIG. 2A, at step 201, simulation generation and abnormality detection computing platform 110 may establish a connection with entity computing system 120. For instance, simulation generation and abnormality detection computing platform 110 may establish a first wireless connection with entity computing system 120. Upon establishing the first wireless connection, a communication session may be initiated between simulation generation and abnormality detection computing platform 110 and entity computing system 120.

Although only one entity computing system 120 is shown in FIGS. 2A-2D, additional entity computing systems, such as entity computing system 125, may be used and additional connections may be established, applications identified, and the like, from additional entity computing systems without departing from the invention.

At step 202, simulation generation and abnormality detection computing platform 110 may identify one or more applications or systems for evaluation. For instance, in some examples, all applications hosted by or executing on entity computing system 120 may be identified for evaluation. Additionally or alternatively, applications having at least a threshold number of users, amount of traffic, or the like, may be identified for evaluation. Additional criteria may be used to identify applications without departing from the invention.

At step 203, simulation generation and abnormality detection computing platform 110 may generate one or more simulations or simulated scenarios. For instance, simulation generation and abnormality detection computing platform 110 may identify one or more simulations of communications within an application, communications between applications, API calls, user interface displays for an application, or the like.

At step 204, simulation generation and abnormality detection computing platform 110 may execute the identified simulation scenarios. For instance, simulation generation and abnormality detection computing platform 110 may execute each identified simulation scenario on a version of the application or system in a secure testing environment. For instance, simulation generation and abnormality detection computing platform 110 may simulate each identified scenario in a secure computing environment to detect any abnormalities (e.g., inaccurate display of text, inaccurate date, slow loading data, issues with display of images or video, or the like).

In some examples, upon detecting an abnormality, simulation generation and abnormality detection computing platform 110 may execute one or more commands or instructions including a potential resolution to the abnormality detected in the simulated scenario. For instance, simulation generation and abnormality detection computing platform 110 may execute one or more commands or instructions attempting to automatically resolve the abnormality detected in the simulated scenario (e.g., replace font with standard font, compress image size, or the like). If the attempted resolution is successful, the resolution may be stored.

At step 205, based on the executed simulated scenarios, simulation generation and abnormality detection computing platform 110 may capture any detected abnormalities, as well as any successful resolutions to the abnormalities detected in the simulated scenarios. In some examples, capturing detected abnormalities may include capturing abnormalities detected upstream and/or downstream of the application being evaluated. For instance, if an issue with Application A causes issues with downstream Application C, the abnormality may be captured for both Application A and Application C. Accordingly, by testing all links, communications, and the like, associated with an application, the interactions between a particular application and other applications that feed data into the application, or receive outputs from the application, may be evaluated for potential abnormalities.

With reference to FIG. 2B, at step 206, simulation generation and abnormality detection computing platform 110 may store the detected abnormalities in, for instance, database 112c. In some examples, the scenario that prompted the abnormality may also be stored (e.g., in association with the abnormality), as well as any resolutions that were successfully executed on the abnormalities detected in the simulated scenarios.

At step 207, simulation generation and abnormality detection computing platform 110 may receive deployed application data. For instance, simulation generation and abnormality detection computing platform 110 may receive or evaluate applications deployed in a production environment that correspond to the applications evaluated using the simulated scenarios.

At step 208, the deployed application data may be analyzed to determine whether abnormalities detected in the simulated scenarios exist in the deployed applications. In some examples, a machine learning model may be used to evaluate the deployed application data. For instance, the machine learning model may receive, as inputs, the deployed application parameters, data, APIs, or the like, as well as the abnormalities detected in the simulated scenarios, and may identify patterns or sequences in the data to identify corresponding abnormalities in the deployed application. In some examples, the machine learning model may be trained using historical abnormality data (e.g., issues that were previously detected, applications impacted, and the like). Accordingly, the machine learning model may learn to recognize patterns or sequences in applications, communications, abnormalities, and the like.

In some examples, the machine learning model may be or include one or more supervised learning models (e.g., decision trees, bagging, boosting, random forest, neural networks, linear regression, artificial neural networks, logical regression, support vector machines, and/or other models), unsupervised learning models (e.g., clustering, anomaly detection, artificial neural networks, and/or other models), knowledge graphs, simulated annealing algorithms, hybrid quantum computing models, and/or other models. In some examples, training the machine learning model may include training the model using labeled data (e.g., labeled data identifying detected abnormalities, associated applications, and the like) and/or unlabeled data.

At step 209, based on the evaluation at step 208, simulation generation and abnormality detection computing platform 110 may identify one or more abnormalities in the deployed application being evaluated (or in one or more applications if a plurality of applications are being evaluated in parallel). In some examples, the identified one or more abnormalities may be output by the machine learning model.

At step 210, the identified abnormalities may be evaluated to determine whether it may be automatically resolved by the simulation generation and abnormality detection computing platform 110. For instance, simulation generation and abnormality detection computing platform 110 may evaluate the identified abnormality and determine whether a successful resolution to the abnormality was performed in the simulated scenario. In some examples, the machine learning model may be used to determine whether an automatic resolution is available for the identified abnormality. In some examples, the machine learning model may receive, as inputs, the identified abnormality and application information, and may identify patterns or sequences in the data to output a resolution to the abnormality. In some examples, the machine learning model may be further trained using historical resolution data (e.g., successful resolution of previous abnormalities, successful resolution of abnormalities detected in the simulated scenarios, and the like). In some examples, the machine learning model may be or include one or more supervised learning models (e.g., decision trees, bagging, boosting, random forest, neural networks, linear regression, artificial neural networks, logical regression, support vector machines, and/or other models), unsupervised learning models (e.g., clustering, anomaly detection, artificial neural networks, and/or other models), knowledge graphs, simulated annealing algorithms, hybrid quantum computing models, and/or other models. In some examples, training the machine learning model may include training the model using labeled data (e.g., labeled data identifying detected abnormalities and associated successful resolutions) and/or unlabeled data.

If, at step 210, the evaluation of the detected abnormality in the deployed application does not have a resolution that may be automatically executed, at step 211 in FIG. 2C, simulation generation and abnormality detection computing platform 110 may generate a notification indicating that an abnormality was identified in a particular application and that an automatic resolution is not available. For instance, FIG. 4 illustrates one example notification 400 that may be generated. As shown in FIG. 4, user interface 400 may include identification of detection of an abnormality in a particular application (e.g., “Application A”). The notification 400 may further include identification of other applications that may be impacted by the abnormality (e.g., “Application D,” “Application E”). In some examples, the additional applications impacted may be identified from the simulated scenario associated with the identified abnormality. For instance, if an abnormality is detected in Application A, the simulated scenario associated with detection of the particular abnormality may be used to identify other applications impacted by the abnormality (e.g., as determined or identified in the simulated scenario).

At step 212, simulation generation and abnormality detection computing platform 110 may establish a connection with entity user computing device 140. For instance, simulation generation and abnormality detection computing platform 110 may establish a second wireless connection with entity user computing device 140. Upon establishing the second wireless connection, a communication session may be initiated between simulation generation and abnormality detection computing platform 110 and entity user computing device 140.

At step 213, simulation generation and abnormality detection computing platform 110 may transmit or send the generated notification to the entity user computing device 140. For instance, simulation generation and abnormality detection computing platform 110 may transmit or send the notification via the communication session initiated upon establishing the second wireless connection. In some examples, transmitting or sending the notification may cause the notification to be displayed by a display of the entity user computing device 140.

At step 214, entity user computing device 140 may receive and display the notification.

Returning to step 210, the evaluation of the detected abnormality in the deployed application does have a resolution that may be automatically executed, at step 215, simulation generation and abnormality detection computing platform 110 may identify the available automatic resolution for the identified abnormality. For instance, if the abnormality includes incorrect display of a custom font, the simulation generation and abnormality detection computing platform 110 may identify a standard font that may be used to replace the custom font via an automated process (e.g., without user interaction).

With reference to FIG. 2D, at step 216, simulation generation and abnormality detection computing platform 110 may automatically execute the identified resolution to resolve the abnormality in the deployed application.

At step 217, simulation generation and abnormality detection computing platform 110 may update or validate the machine learning model. For instance, an outcome of evaluation of the abnormality for automatic resolution, execution of automatic resolution, or the like, may be provided via a dynamic feedback loop to continuously update or validate the machine learning model, thereby continuously improving accuracy of model outputs.

In some instances, simulation generation and abnormality detection computing platform 110 may continuously update, validate, refine, or the like, the machine learning model. In some examples, the simulation generation and abnormality detection computing platform 110 may maintain an accuracy threshold for the machine learning model and may pause refinement (through the dynamic feedback loop) of the model if the corresponding accuracy is identified as greater than the accuracy threshold. Further, if the accuracy is at or below the accuracy threshold, the simulation generation and abnormality detection computing platform 110 may resume refinement of the model through the corresponding dynamic feedback loop.

Although aspects described with respect to FIGS. 2A-2D include a simulator proxy generating and executing simulations and an analyzer then analyzing the deployed applications to determine whether abnormalities exist, in some examples, these processes may run in parallel with continuous or near continuous monitoring or deployed applications or systems to detect, in real-time, abnormalities and quickly resolve any detected abnormalities.

FIG. 3 is a flow chart illustrating one example method of simulation generation and abnormality detection in accordance with one or more aspects described herein. The processes illustrated in FIG. 3 are merely some example processes and functions. The steps shown may be performed in the order shown, in a different order, more steps may be added, or one or more steps may be omitted, without departing from the invention. In some examples, one or more steps may be performed simultaneously with other steps shown and described. One of more steps shown in FIG. 3 may be performed in real-time or near real-time.

At step 300, a computing platform, such as simulation generation and abnormality detection computing platform 110, may identify a plurality of applications for evaluation or analysis. For instance, the computing platform may identify a plurality of applications associated with an enterprise organization to be analyzed or evaluated for abnormalities using the arrangements described herein. In some examples, one or more applications of the plurality of applications may be customer-facing applications associated with the enterprise organization. Additionally or alternatively, one or more applications of the plurality of applications may be internal applications used by one or more employees or other associates of the enterprise organization.

At step 302, an abnormality simulator proxy of the computing platform may execute a plurality of simulated scenarios for each application of the plurality of applications identified for analysis. For instance, the abnormality simulator proxy may identify communications within each application, between applications, API calls associated with each application, text, images or videos displayed by each application, and the like, and may generate and execute simulations of each scenario (e.g., each communication, API call, UI with text, images or videos displayed, or the like). The simulations may be executed in a secure environment to identify any abnormalities in each application of the plurality of applications.

In some examples, executing the simulated scenarios for each application may include identifying and/or executing one or more potential resolutions to any identified or detected abnormalities. For instance, the abnormality simulator proxy may attempt to resolve any identified or detected abnormalities by executing one or more potential resolutions.

At step 304, the computing platform may capture abnormality results generated based on execution of the simulated scenarios. For instance, any detected or identified abnormalities that occurred during execution of the simulated scenarios may be captured in the abnormality results.

In some examples, capturing the abnormality results may further include capturing a resolution (or a plurality of resolutions), of the one or more potential resolutions, that resolved one or more of the abnormalities detected in the simulated scenarios (e.g., a successful resolution to an abnormality may be captured).

At step 306, the computing platform may store, in a database, the abnormality results and simulated scenarios for each application. In some examples, the computing platform may further store the resolution that successfully resolved the one or more abnormalities detected in the simulated scenarios.

At step 308, the computing platform may receive deployed application data corresponding to the identified plurality of applications and may compare the deployed versions of each application to the plurality of simulated scenarios and the captured abnormality results for a corresponding application to identify an abnormality in a deployed version of an application. In some examples, abnormalities may include at least one of: an incorrectly displayed font, an image or video that does not load properly, an image or video that has a load time greater than a threshold time or an expected time, or the like.

At step 310, the computing platform may evaluate, based on the identified abnormality in the deployed version of the application and via an abnormality solver of the computing platform, the identified abnormality to determine whether it can be resolved automatically (e.g., by the computing platform and without user input or interaction). In some examples, evaluating whether the identified abnormality can be resolved automatically may include determining whether a stored resolution to the identified abnormality exists for the identified abnormality (e.g., a resolution captured and stored during execution of the simulated scenarios).

If, at step 310, the identified abnormality cannot be resolved automatically, at step 312, a notification may be generated by the computing platform. In some examples, the notification may include identification of the abnormality, an associated application, other impacted applications (e.g., upstream or downstream applications), and the like. At step 314, the notification may be transmitted to a computing device, such as an administrator computing device, for display by the administrator computing device.

If, at step 310, the identified abnormality can be resolved automatically, at step 316, the computing platform may execute one or more commands to modify the deployed application to resolve the identified abnormality. In some arrangements, executing the one or more commands to modify the deployed application may include identifying the stored resolution that exists to resolve the identified abnormality and executing the stored resolution. For instance, the identified abnormality may include a custom font that does not display correctly. A successful resolution identified via the simulated scenarios may include replacing the custom font with a standard font that displays correctly. Accordingly, the computing platform may retrieve the stored resolution and execute one or more commands to modify the deployed application to replace the custom font with the standard font.

As discussed herein, aspects described provide for secure identification of abnormalities in applications, systems, and the like, using, for instance, system simulations, and, in some examples, automatic resolution of detected abnormalities. Accordingly, the arrangements described herein enable error detection in multiple, large distributed systems and enable resolution on the fly. As discussed, the arrangement provided use system simulation to identify errors, determine a cause of the error and resolve the error.

Some example use cases of the simulation generation and abnormality detection computing platform 110 and arrangements described herein are provided below. The use cases described are merely some example user cases and aspects described herein may be used in other use cases without departing from the invention.

In some examples, customer facing applications may include various fonts, images, and the like. During a testing phase of an application, it may be difficult for developers to test all use cases or scenarios to determine whether fonts, images, and the like, are displaying correctly and/or are legible to users. Accordingly, the simulation generation and abnormality detection computing platform 110 may simulate the customer facing application and determine that it does not support certain fonts, different languages, or the like. In some examples, the simulated scenarios may include use of different web browsers and/or devices to attempt to load an application or page in order to determine whether abnormalities exist for some browsers and not others or some devices and not others. The simulated scenarios may determine whether fonts and/or images are properly rendered. If they do not render properly, this may be captured by the abnormality simulator proxy as an abnormality and stored.

In another example, in some arrangements, fonts might not be legible due to a background color being incompatible with a font color. This abnormality may be captured by the abnormality simulator proxy and stored.

The detected and/or stored abnormalities (e.g., fonts or images not rendering properly, fonts being illegible, or the like) may be analyzed by the abnormality analyzer to determine whether the abnormalities detected in the simulated scenarios are present in real-time in the deployed application. If so, the abnormality solver may be used to identify and attempt an automatic resolution of the font or images issues identified in the deployed application. For instance, the abnormality solver may replace custom fonts with standard fonts that may render properly and be legible (e.g., in all browsers, for all devices, or the like). In another example, if an image is nor rendering properly, the solver may compress the image or use a smaller image to improve loading speed or rendering quality. These modifications of the deployed application may be performed automatically without user interaction.

In another example use case, in some examples, videos provided via an application may include audio transcripts, sign language interpretation, or the like, that may be provided for hard of hearing viewers. In some examples, a video may load but the transcript and/or sign language interpretation might not display or might not display properly. This abnormality may be detected by the abnormality simulator proxy and stored. The abnormality analyzer may analyze the abnormality and identify that the transcript or interpretation is not properly visible and the abnormality solver may automatically generate a transcript of the content of the video and update the deployed application to include the generated transcript.

In yet another example use case, in a dynamic API system, API errors may be detected in real-time. For instance, if an actual API response does not match an expected response, the solver may generate a response (e.g., using machine learning or artificial intelligence). For instance, a since API call may include a plurality of internal micro-app calls. If the API call fails, the arrangement described herein can quickly identify a cause of the failure (e.g., if one or more micro-app calls failed) and resolve the issue to ensure continuous access.

In still another example use case, the arrangements described herein may be used to identify and implement contingency plans. For instance, an application with a dependency on a first data center may go through a first service to access the first data center. If the first data center or first service has an issue (e.g., data is not accessible), the simulator proxy and analyzer may detect the issue and the abnormality solver may automatically write code to access a second data center or use a second service to access the first data center.

In still another example, some applications may use text, video and an animated guide. However, the amount of media may slow the app or make is cumbersome to use. Accordingly, users might only view the text and might not view videos or the animated guide (e.g., because of the application is not responsive enough). Accordingly, the proxy and analyzer may detect an issue and identify that the page or application is slow to load. The solver may then recommend (e.g., based on machine learning analysis) recommendations to make the application or page faster, may remove code automatically to make the page more user friendly, or the like. In some examples, the machine learning model may be used to analyze user interactions with a page or application to output recommended resolutions. For instance, user impressions (e.g., number of clicks of a video or the like) may be captured and analyzed to determine that large multimedia might not be frequently accessed by users and the presence might cause the entire page to load slowly. Accordingly, the solver may remove some video not be accessed by the users to improve the responsiveness of the page.

In another example, if the simulation generation and abnormality detection computing platform 110 detects that users refresh a page at least a threshold number of times without playing a particular video, the computing platform may determine that the video is causing issues and may remove the video.

In some examples, aspects described herein may be used to customize applications, web pages, or the like for particular business units, geographic regions, or even for particular users. For instance, user interactions with a page or application (e.g., number of links clicked, videos played, page refresh, or the like) may be captured and analyzed, using a machine learning model associated with the abnormality solver, to identify modifications to an application or page that would make the application or page more accommodating for the particular user, geographic region, business unit, or the like. Accordingly, the arrangements described herein may be used to not only resolve issues with proper functioning of an application or system, but also to customize a user experience by making information more accessible to a user.