# US 2018/0322309 A1 · Perturbation-based techniques for anonymizing datasets > Agents: this data is available as JSON from the patentgrep API and MCP server — see https://patentgrep.com/llms.txt - Number: US 2018/0322309 A1 - Document ID: US20180322309A1 - Status: Published application - Kind: A1 - Application number: 15/972,085 - Filed: 2018-05-04 - Published: 2018-11-08 - Earliest claimed: 2017-05-08 - Assignees: Autodesk, Inc. - Applicants: Autodesk, Inc. - Inventors: Justin Frank MATEJKA; George FITZMAURICE - Web page: https://patentgrep.com/patent/US20180322309A1 ## CPC - G06F 21/6254: {by anonymising data, e.g. decorrelating personal data from the owner's identification} - G06F 16/2462: {Approximate or statistical queries} - G06F 17/18: for evaluating statistical data {, e.g. average values, frequency distributions, probability functions, regression analysis (forecasting specially adapted for a specific administrative, business or logistic context G06Q10/04)} - G06F 18/22: Matching criteria, e.g. proximity measures - G09C 5/00: Ciphering apparatus or methods not provided for in the preceding groups, e.g. involving the concealment or deformation of graphic data such as designs, written or printed messages - H04L 2209/88: Medical equipments - G06F: ELECTRIC DIGITAL DATA PROCESSING (computer systems based on specific computational models G06N) - G09C: CIPHERING OR DECIPHERING APPARATUS FOR CRYPTOGRAPHIC OR OTHER PURPOSES INVOLVING THE NEED FOR SECRECY - H04L: TRANSMISSION OF DIGITAL INFORMATION, e.g. TELEGRAPHIC COMMUNICATION (arrangements common to telegraphic and telephonic communication H04M) ## Abstract In various embodiments, a dataset generation application generates a new dataset based on an original dataset. The dataset generation engine perturbs a first data item included in the original dataset to generate a second data item. The dataset generation application then generates a test dataset based on the original dataset and the second data item. The test dataset includes the second data item instead of the first data item. Subsequently, the dataset generation application determines that the test dataset is characterized by a first property value that is substantially similar to a second property value that characterizes the original dataset. The first property value and the second property value are associated with the same property. Finally, the dataset generation application generates a new dataset based on the test dataset. The new dataset conveys aspect(s) of the original dataset without revealing the first data item. ## Claims ### Claim 1 A computer-implemented method for generating a new dataset based on an original dataset, the method comprising: perturbing a first data item included in the original dataset to generate a second data item; generating a test dataset based on the original dataset and the second data item, wherein the test dataset includes the second data item instead of the first data item; determining, via a processor, that the test dataset is characterized by a first property value that is substantially similar to a second property value that characterizes the original dataset, wherein both the first property value and the second property value are associated with a first property; and generating the new dataset based on the test dataset, wherein the new dataset conveys at least one aspect of the original dataset that is associated with the first property without revealing the first data item. ### Claim 2 (depends on 1) The computer-implemented method of claim 1, wherein generating the test dataset comprises: replacing the first data item included in the original dataset with the second data item to generate a potential dataset; computing a first average distance between the potential dataset and a target shape; computing a second average distance between the original dataset and the target shape; determining that the first average distance is less than the second average distance; and setting the test dataset equal to the potential dataset. ### Claim 3 (depends on 1) The computer-implemented method of claim 1, wherein generating the test dataset comprises: performing one or more simulated annealing operations that indicate that the second data item is to be included in the test dataset; and replacing the first data item included in the original dataset with the second data item to generate the test dataset. ### Claim 4 (depends on 1) The computer-implemented method of claim 1, wherein perturbing the first data item comprises: randomly selecting the first data item from a plurality of data items included in the original dataset; and modifying the first data item based on a randomly generated value. ### Claim 5 (depends on 1) The computer-implemented method of claim 1, further comprising, prior to perturbing the first data item: perturbing a third data item included in the original dataset to generate a fourth data item; generating an initial test dataset based on the original dataset and the fourth data item, wherein the initial test dataset includes the fourth data item instead of the third data item; determining that a third property value associated with both the first property and the initial test dataset is not substantially similar to the second property value; and discarding the initial test dataset. ### Claim 6 (depends on 1) The computer-implemented method of claim 1, wherein generating the test dataset comprises: replacing the first data item included in the original dataset with the second data item to generate a modified test dataset; determining that the modified test dataset is characterized by a third property value that is associated with the first property and is substantially similar to the second property value; perturbing a third data item included in the modified test dataset to generate a fourth data item; and replacing the third data item included in the modified test dataset with the fourth data item to generate the test dataset. ### Claim 7 (depends on 1) The computer-implemented method of claim 1, wherein generating the new dataset comprises: performing a plurality of replacement operations on the test dataset to generate a modified test dataset, wherein each replacement operation replaces a given data item included in the test dataset with a new data item that is generated based on the given data item; determining that the modified test dataset is characterized by a third property value that is associated with the first property and is substantially similar to the second property value; and setting the new dataset equal to the modified test dataset. ### Claim 8 (depends on 1) The computer-implemented method of claim 1, wherein the original dataset comprises a mufti-dimensional dataset. ### Claim 9 (depends on 1) The computer-implemented method of claim 1, wherein the original dataset comprises a design file or model geometry. ### Claim 10 (depends on 1) The computer-implemented method of claim 1, wherein the first property comprises a mathematical property, a statistical property, a visual property, a physical property, or a result of an application-specific action. ### Claim 11 A computer-readable storage medium including instructions that, when executed by a processor, cause the processor to generate a new dataset based on an original dataset by performing the steps of: perturbing a first data item included in the original dataset to generate a second data item; generating a test dataset based on the original dataset and the second data item, wherein the test dataset includes the second data item instead of the first data item; determining, via a processor, that the test dataset is characterized by a first property value that is substantially similar to a second property value that characterizes the original dataset, wherein both the first property value and the second property value are associated with a first property; and generating the new dataset based on the test dataset, wherein the new dataset conveys at least one aspect of the original dataset that is associated with the first property without revealing the first data item. ### Claim 12 (depends on 11) The computer-readable storage medium of claim 11, wherein generating the test dataset comprises: replacing the first data item included in the original dataset with the second data item to generate a potential dataset; computing a first average distance between the potential dataset and a target shape; computing a second average distance between the original dataset and the target shape; determining that the first average distance is less than the second average distance; and setting the test dataset equal to the potential dataset. ### Claim 13 (depends on 11) The computer-readable storage medium of claim 11, wherein generating the test dataset comprises: performing one or more simulated annealing operations that indicate that the second data item is to be included in the test dataset; and replacing the first data item included in the original dataset with the second data item to generate the test dataset. ### Claim 14 (depends on 11) The computer-readable storage medium of claim 11, wherein perturbing the first data item comprises: randomly selecting the first data item from a plurality of data items included in the original dataset; and modifying the first data item based on a randomly generated value. ### Claim 15 (depends on 11) The computer-readable storage medium of claim 11, further comprising, prior to perturbing the first data item: perturbing a third data item included in the original dataset to generate a fourth data item; generating an initial test dataset based on the original dataset and the fourth data item, wherein the initial test dataset includes the fourth data item instead of the third data item; determining that a third property value associated with both the first property and the initial test dataset is not substantially similar to the second property value; and discarding the initial test dataset. ### Claim 16 (depends on 11) The computer-readable storage medium of claim 11, wherein generating the new dataset comprises: perturbing the second data item included in the test dataset to generate a third data item; generating a modified test dataset based on the test dataset and the third data item, wherein the modified test dataset includes the third data item instead of the second data item; determining that the modified test dataset is characterized by a third property value that is associated with the first property and is substantially similar to the second property value; and setting the new dataset equal to the modified test dataset. ### Claim 17 (depends on 11) The computer-readable storage medium of claim 11, wherein generating the new dataset comprises: performing a plurality of replacement operations on the test dataset to generate a modified test dataset, wherein each replacement operation replaces a given data item included in the test dataset with a new data item that is generated based on the given data item; determining that the modified test dataset is characterized by a third property value that is associated with the first property and is substantially similar to the second property value; determining that the modified test dataset satisfies one or more completion criteria; and setting the new dataset equal to the modified test dataset. ### Claim 18 (depends on 11) The computer-readable storage medium of claim 11, wherein the first data item comprises a hierarchical data item. ### Claim 19 (depends on 11) The computer-readable storage medium of claim 11, wherein the first property comprises a mathematical property, a statistical property, a visual property, a physical property, or a result of an application-specific action. ### Claim 20 A system, comprising: a memory storing instructions; and a processor that is coupled to the memory and, when executing the instructions, is configured to: perturb a first data item included in the original dataset to generate a second data item; generate a test dataset based on the original dataset and the second data item, wherein the test dataset includes the second data item instead of the first data item; determine that the test dataset is characterized by a first property value that is substantially similar to a second property value that characterizes the original dataset, wherein both the first property value and the second property value are associated with a first property; and generate the new dataset based on the test dataset, wherein the new dataset conveys at least one aspect of the original dataset that is associated with the first property without revealing the first data item.