# US 12,184,525 B2 · Methods and systems for determining severity of disruptions in communication networks in non-homogenous environments > Agents: this data is available as JSON from the patentgrep API and MCP server — see https://patentgrep.com/llms.txt - Number: US 12,184,525 B2 - Document ID: US12184525B2 - Status: Granted - Kind: B2 - Application number: 18/489,548 - Filed: 2023-10-18 - Published: 2024-02-08 - Granted: 2024-12-31 - Earliest claimed: 2021-10-18 - Assignees: Capital One Services, LLC - Applicants: Capital One Services, LLC - Inventors: Matthew Louis Nowak; Michael Anthony Young, Jr.; Christopher McDaniel - Web page: https://patentgrep.com/patent/US12184525B2 ## CPC - H04L 41/16: using machine learning or artificial intelligence - H04L 43/0876: Network utilisation, e.g. volume of load or congestion level - G06F 40/205: Parsing - G06N 3/084: Backpropagation, e.g. using gradient descent - G06N 20/00: Machine learning - G06N 20/20: Ensemble learning - H04L 41/064: {involving time analysis} - H04L 41/0681: Configuration of triggering conditions - H04L 41/142: using statistical or mathematical methods - H04L 43/16: Threshold monitoring - H04L 67/10: in which an application is distributed across nodes in the network (software deployment G06F8/60; multiprogramming arrangements G06F9/46) - H04L 41/22: comprising specially adapted graphical user interfaces [GUI] - H04L: TRANSMISSION OF DIGITAL INFORMATION, e.g. TELEGRAPHIC COMMUNICATION (arrangements common to telegraphic and telephonic communication H04M) - G06F: ELECTRIC DIGITAL DATA PROCESSING (computer systems based on specific computational models G06N) - G06N: COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS ## Abstract Methods and systems that use a plurality of machine learning models to both monitor user-generated data entries corresponding to differences in network traffic that may be evidence of a disruption and determine severity levels based on: (i) current and historic differences in average network traffic over the plurality of communication networks; (ii) current and historic user-generated data entries; and (iii) labeled severity levels for historic differences in average network traffic over the plurality of communication networks. ## Claims ### Claim 1 A system for determining severity levels of disruptions in communication networks in non-homogenous environments using models, the system comprising: cloud-based storage circuitry configured to store: a first model, wherein the first model comprises a natural language processing model that parses text strings describing differences in network characteristics; and a second model, wherein the second model is trained to determine severity levels based on: network characteristics (i) current and historic user-generated data entries; and (ii) labeled severity levels for historic differences in average network characteristics over a plurality of communication networks; cloud-based control circuitry configured to: determine, using the first model, a user-generated data entry corresponding to a difference in average network characteristics over a first communication network of a plurality of communication networks; generate a feature input based on the user-generated data entry and the difference; input the feature input into the second model; receive an output from the second model; determine, using the second model, a severity level from a plurality of severity levels based on the user-generated data entry and the difference; and cloud-based input/output circuitry configured to generate for display, on a user interface, a recommendation based on the severity level, wherein the recommendation comprises a description of a disruption corresponding to the difference that is normalized across different communication networks. ### Claim 2 A method for determining severity levels of disruptions in communication networks in non-homogenous environments using models, the method comprising: determining a difference in average network characteristics over a first communication network of a plurality of communication networks; determining, using a first model, a user-generated data entry corresponding to the difference, wherein the first model comprises a natural language processing model; determining, using a second model, a severity level from a plurality of severity levels based on the user-generated data entry and the difference, wherein the second model is trained to determine severity levels based on: network characteristics (i) current and historic user-generated data entries; and (ii) labeled severity levels for historic differences in average network characteristics over the plurality of communication networks; and generating for display, on a user interface, a recommendation based on the severity level. ### Claim 3 (depends on 2) The method of claim 2, wherein determining the difference in average network characteristics over the first communication network of the plurality of communication networks comprises: determining a first amount of network characteristics over the first communication network during a first current time period; determining a first average amount of network characteristics over the first communication network during a first plurality of historic time periods; and determining the difference based on comparing the first amount to the first average amount. ### Claim 4 (depends on 2) The method of claim 2, wherein determining the user-generated data entry corresponding to the difference comprises: retrieving a data file created by a user at a date; parsing the data file for a plurality of text strings; and applying the natural language processing model to the plurality of text strings to determine user-generated data entries corresponding to the difference. ### Claim 5 (depends on 2) The method of claim 2, wherein determining the severity level from the plurality of severity levels based on the user-generated data entry comprises: comparing the user-generated data entry to a plurality of labeled user-generated data entries, wherein each labeled user-generated data entry of the plurality of labeled user-generated data entries corresponds to a respective severity level of the plurality of severity levels; and retrieving, based on comparing the user-generated data entry to the plurality of labeled user-generated data entries, the respective severity level for a labeled user-generated data entry that corresponds to the user-generated data entry. ### Claim 6 (depends on 2) The method of claim 2, wherein determining the severity level from the plurality of severity levels based on the user-generated data entry comprises: determining a frequency of the user-generated data entry in data files; comparing the frequency to a plurality of labeled frequencies, wherein each labeled frequency of the plurality of labeled frequencies corresponds to a respective severity level of the plurality of severity levels; and retrieving, based on comparing the frequency to the plurality of labeled frequencies, the respective severity level for a labeled frequency that corresponds to the frequency. ### Claim 7 (depends on 2) The method of claim 2, wherein determining the severity level from the plurality of severity levels based on the user-generated data entry comprises: determining a number of instances of the user-generated data entry in data files created at a date; comparing the number of instances to a plurality of labeled number of instances, wherein each labeled number of instances of the plurality of labeled number of instances corresponds to a respective severity level of the plurality of severity levels; and retrieving, based on comparing the number of instances to the plurality of labeled number of instances, the respective severity level for a labeled number of instances that corresponds to the number of instances. ### Claim 8 (depends on 2) The method of claim 2, wherein determining the severity level from the plurality of severity levels based on the user-generated data entry comprises: determining a number of data files containing the user-generated data entry created at a date; comparing the number of data files to a plurality of labeled number of data files, wherein each labeled number of data files of the plurality of labeled number of data files corresponds to a respective severity level of the plurality of severity levels; and retrieving, based on comparing the number of data files to the plurality of labeled number of data files, the respective severity level for a labeled number of data files that corresponds to the number of data files. ### Claim 9 (depends on 2) The method of claim 2, wherein determining the severity level from the plurality of severity levels based on the user-generated data entry comprises: determining a number of different users that created data files containing the user-generated data entry at a date; comparing the number of different users to a plurality of labeled number of different users, wherein each labeled number of different users of the plurality of labeled number of different users corresponds to a respective severity level of the plurality of severity levels; and retrieving, based on comparing the number of different users to the plurality of labeled number of different users, the respective severity level for a labeled number of different users that corresponds to the number of different users. ### Claim 10 (depends on 2) The method of claim 2, further comprising: comparing the difference to a threshold difference; and in response to determining that the difference equals or exceeds the threshold difference, retrieving a data file created by a user at a date. ### Claim 11 (depends on 2) The method of claim 2, wherein using the second model comprises: generating a feature input based on the user-generated data entry and the difference; inputting the feature input into the second model; and receiving an output from the second model. ### Claim 12 A non-transitory, computer-readable medium for determining severity levels of disruptions in communication networks in non-homogenous environments using models, comprising instructions that when executed by one or more processors, causes operations comprising: determining a difference in average network characteristics over a first communication network of a plurality of communication networks during a first time interval; determining, using a first model, a user-generated data entry corresponding to the difference, wherein the first model comprises a natural language processing model; determining, using a second model, a severity level from a plurality of severity levels based on the user-generated data entry and the difference, wherein the second model is trained to determine severity levels based on user-generated data entries and labeled severity levels for historic differences in average network characteristics over the plurality of communication networks; and generating for display a recommendation based on the severity level. ### Claim 13 (depends on 12) The non-transitory, computer-readable medium of claim 12, wherein determining the difference in average network characteristics over the first communication network of the plurality of communication networks during the first time interval comprises: determining a first amount of network characteristics over the first communication network during a first current time period; determining a first average amount of network characteristics over the first communication network during a first plurality of historic time periods; and determining the difference based on comparing the first amount to the first average amount. ### Claim 14 (depends on 12) The non-transitory, computer-readable medium of claim 12, wherein determining the user-generated data entry corresponding to the difference comprises: retrieving a data file created by a user at a date; parsing the data file for a plurality of text strings; and applying the natural language processing model to the plurality of text strings to determine user-generated data entries corresponding to the difference. ### Claim 15 (depends on 12) The non-transitory, computer-readable medium of claim 12, wherein determining the severity level from the plurality of severity levels based on the user-generated data entry comprises: comparing the user-generated data entry to a plurality of labeled user-generated data entries, wherein each labeled user-generated data entry of the plurality of labeled user-generated data entries corresponds to a respective severity level of the plurality of severity levels; and retrieving, based on comparing the user-generated data entry to the plurality of labeled user-generated data entries, the respective severity level for a labeled user-generated data entry that corresponds to the user-generated data entry. ### Claim 16 (depends on 12) The non-transitory, computer-readable medium of claim 12, wherein determining the severity level from the plurality of severity levels based on the user-generated data entry comprises: determining a frequency of the user-generated data entry in data files; comparing the frequency to a plurality of labeled frequencies, wherein each labeled frequency of the plurality of labeled frequencies corresponds to a respective severity level of the plurality of severity levels; and retrieving, based on comparing the frequency to the plurality of labeled frequencies, the respective severity level for a labeled frequency that corresponds to the frequency. ### Claim 17 (depends on 12) The non-transitory, computer-readable medium of claim 12, wherein determining the severity level from the plurality of severity levels based on the user-generated data entry comprises: determining a number of instances of the user-generated data entry in data files created at a date; comparing the number of instances to a plurality of labeled number of instances, wherein each labeled number of instances of the plurality of labeled number of instances corresponds to a respective severity level of the plurality of severity levels; and retrieving, based on comparing the number of instances to the plurality of labeled number of instances, the respective severity level for a labeled number of instances that corresponds to the number of instances. ### Claim 18 (depends on 12) The non-transitory, computer-readable medium of claim 12, wherein determining the severity level from the plurality of severity levels based on the user-generated data entry comprises: determining a number of data files containing the user-generated data entry created at a date; comparing the number of data files to a plurality of labeled number of data files, wherein each labeled number of data files of the plurality of labeled number of data files corresponds to a respective severity level of the plurality of severity levels; and retrieving, based on comparing the number of data files to the plurality of labeled number of data files, the respective severity level for a labeled number of data files that corresponds to the number of data files. ### Claim 19 (depends on 12) The non-transitory, computer-readable medium of claim 12, wherein determining the severity level from the plurality of severity levels based on the user-generated data entry comprises: determining a number of different users that created data files containing the user-generated data entry at a date; comparing the number of different users to a plurality of labeled number of different users, wherein each labeled number of different users of the plurality of labeled number of different users corresponds to a respective severity level of the plurality of severity levels; and retrieving, based on comparing the number of different users to the plurality of labeled number of different users, the respective severity level for a labeled number of different users that corresponds to the number of different users. ### Claim 20 (depends on 12) The non-transitory, computer-readable medium of claim 12, wherein the instructions cause further operations comprising: comparing the difference to a threshold difference; and in response to determining that the difference equals or exceeds the threshold difference, retrieving a data file created by a user at a date.