# US 12,225,011 B2 · Data protection in network environments > Agents: this data is available as JSON from the patentgrep API and MCP server — see https://patentgrep.com/llms.txt - Number: US 12,225,011 B2 - Document ID: US12225011B2 - Status: Granted - Kind: B2 - Application number: 17/809,563 - Filed: 2022-06-29 - Published: 2024-01-04 - Granted: 2025-02-11 - Earliest claimed: 2022-06-29 - Assignees: International Business Machines Corporation - Applicants: International Business Machines Corporation - Inventors: Jin Wang; Lei Gao; A Peng Zhang; Dan Sun; Jing Zhang; Na Liu; Xun Pan; Zi Yun Kang - Web page: https://patentgrep.com/patent/US12225011B2 ## CPC - H04L 63/10: {for controlling access to devices or network resources} - G06N 5/01: Dynamic search techniques; Heuristics; Dynamic trees; Branch-and-bound - G06N 20/00: Machine learning - G06N 20/20: Ensemble learning - H04L 63/20: {for managing network security; network security policies in general (filtering policies H04L63/0227)} - H04L: TRANSMISSION OF DIGITAL INFORMATION, e.g. TELEGRAPHIC COMMUNICATION (arrangements common to telegraphic and telephonic communication H04M) - G06N: COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS ## Abstract Computer technology for protecting data security in a computerized system for recommending content to users where, a processing unit generates an identifier for a first data record relating to a user device based on a first machine learning model. Then, the processing unit sends the identifier to a service provider, and the service provider uses the identifier to determine one or more contents to be sent to the user device. Creating and using a decision tree machine learning (ML) model and a cluster ML model with training records and a transformed records. ## Claims ### Claim 1 A computer-implemented method, comprising: generating, by one or more processing units, an identifier for a first data record relating to a user device, wherein the identifier is generated based on a value output by a first machine learning model receiving the first data record as input; and sending, by one or more processing units, the identifier to a service provider, wherein the identifier is used by the service provider to determine one or more contents to be sent to the user device. ### Claim 2 (depends on 1) The computer-implemented method of claim 1, wherein generating the identifier for the first data record further comprises: applying, by one or more processing units, the first data record as an input to the first machine learning model; obtaining, by one or more processing units, the value outputted by the first machine learning model corresponding to the first data record; and generating, by one or more processing units, the identifier based on the value. ### Claim 3 (depends on 2) The computer-implemented method of claim 2, wherein the first machine learning model is created by: obtaining, by one or more processing units, training records; assigning, by one or more processing units, values respectively to the training records; and creating, by one or more processing units, the first machine learning model based on the training records using one or more classification and regression tree algorithms, wherein each leaf node of the first machine learning model corresponds to one of the assigned values, and the first machine learning model is capable to map an input to a leaf node of the first machine learning model and output a value of the leaf node. ### Claim 4 (depends on 3) The computer-implemented method of claim 3, wherein the training records include data records collected previously from one or more user devices and/or records generated based on the collected data records. ### Claim 5 (depends on 1) The computer-implemented method of claim 1, wherein the first data record includes an original data record received from the user device comprising user data to be protected. ### Claim 6 (depends on 1) The computer-implemented method of claim 1, wherein the first data record includes a transformed record corresponding to an original data record received from the user device comprising user data to be protected, and the transformed record is obtained by: clustering, by one or more processing units, the original data record to a first cluster based on a second machine learning model; and obtaining, by one or more processing units, a record of a center point of the first cluster as the transformed record. ### Claim 7 (depends on 6) The computer-implemented method of claim 6, wherein the second machine learning model is created by: obtaining, by one or more processing units, training records; setting, by one or more processing units, a number of clusters to be applied to the training records; and creating, by one or more processing units, the second machine learning model based on the training records using one or more clustering algorithms, wherein the second machine learning model is capable to cluster an input to a cluster. ### Claim 8 (depends on 7) The computer-implemented method of claim 7, wherein the one or more clustering algorithms comprise k-means algorithms. ### Claim 9 (depends on 1) The computer-implemented method of claim 1, wherein the service provider maintains mapping relationship between identifiers and contents, and the one or more contents are determined by the service provider based on the identifier using the mapping relationship. ### Claim 10 (depends on 7) The computer-implemented method of claim 7, wherein the first machine learning model or the second machine learning model is further updated using subsequent records after obtaining of the training records. ### Claim 11 A computer program product (CPP) comprising: a computer-readable memory coupled to one or more processors; and computer readable instructions stored in the computer-readable memory that, when executed by the one or more processors, performs operations comprising: generating an identifier for a first data record relating to a user device, wherein the identifier is generated based on a value output by a first machine learning model receiving the first data record as input; and sending the identifier to a service provider, wherein the identifier is used by the service provider to determine one or more contents to be sent to the user device. ### Claim 12 (depends on 11) The computer program product of claim 11, wherein generating the identifier for the first data record further comprises: applying the first data record as an input to the first machine learning model; obtaining a value outputted by the first machine learning model corresponding to the first data record; and generating the identifier based on the value. ### Claim 13 (depends on 12) The computer program product of claim 12, wherein the first machine learning model is created by: obtaining training records; assigning values respectively to the training records; and creating the first machine learning model based on the training records using one or more classification and regression tree algorithms, wherein each leaf node of the first machine learning model corresponds to one of the assigned values, and the first machine learning model is capable to map an input to a leaf node of the first machine learning model and output a value of the leaf node. ### Claim 14 (depends on 13) The computer program product of claim 13, wherein the training records include data records collected previously from one or more user devices and/or records generated based on the collected data records. ### Claim 15 (depends on 11) The computer program product of claim 11, wherein the first data record includes an original data record received from the user device comprising user data to be protected. ### Claim 16 (depends on 11) The computer program product of claim 11, wherein the first data record includes a transformed record corresponding to an original data record received from the user device comprising user data to be protected, and the transformed record is obtained by: clustering the original data record to a first cluster based on a second machine learning model; and obtaining a record of a center point of the first cluster as the transformed record. ### Claim 17 (depends on 16) The computer program product of claim 16, wherein the second machine learning model is created by: obtaining training records; setting a number of clusters to be applied to the training records; and creating the second machine learning model based on the training records using one or more clustering algorithms, wherein the second machine learning model is capable to cluster an input to a cluster. ### Claim 18 (depends on 11) The computer program product of claim 11, wherein the computer program product is in a computer system and the computer system includes the one or more processors. ### Claim 19 A computer implemented method (CIM) comprising: applying a set of classification and regression tree algorithms to a plurality of training records to obtain a decision tree machine learning (ML) model including a plurality of leaf nodes, with the plurality of leaf nodes respectively corresponding to the plurality of training records, and with a given first leaf node of the plurality of leaf nodes including a first assigned value that is assigned to a training record corresponding to the first leaf node; and applying a set of clustering algorithms to the plurality of training records to obtain a cluster ML model including a plurality of clusters and a plurality of center points; and transforming a new data record to obtain a transformed record. ### Claim 20 (depends on 19) The CIM of claim 19 wherein the transformation of the new data record includes: clustering the new data record to a first cluster of the plurality of clusters; and obtaining a record of a center point of the first cluster as the transformed record.