# US 12,216,619 B2 · Automated generation of game tags > Agents: this data is available as JSON from the patentgrep API and MCP server — see https://patentgrep.com/llms.txt - Number: US 12,216,619 B2 - Document ID: US12216619B2 - Status: Granted - Kind: B2 - Application number: 18/376,021 - Filed: 2023-10-03 - Published: 2024-01-25 - Granted: 2025-02-04 - Earliest claimed: 2020-05-27 - Assignees: Roblox Corporation - Applicants: Roblox Corporation - Inventors: Eric Holmdahl; Nikolaus Sonntag; Aswath Manoharan - Web page: https://patentgrep.com/patent/US12216619B2 ## CPC - G06F 16/164: {File meta data generation} - G06N 20/00: Machine learning - A63F 13/216: using geographical information, e.g. location of the game device or player using GPS - A63F 13/323: between game devices with different hardware characteristics, e.g. hand-held game devices connectable to game consoles or arcade machines - A63F 13/48: Starting a game, e.g. activating a game device or waiting for other players to join a multiplayer session - A63F 13/75: Enforcing rules, e.g. detecting foul play or generating lists of cheating players - A63F 13/77: involving data related to game devices or game servers, e.g. configuration data, software version or amount of memory - G06N 3/045: Combinations of networks - G06N 3/0464: Convolutional networks [CNN, ConvNet] - G06N 3/08: Learning methods - G06N 3/09: Supervised learning - A63F 13/327: using wireless networks, e.g. Wi-Fi® or piconet - G06F: ELECTRIC DIGITAL DATA PROCESSING (computer systems based on specific computational models G06N) - G06N: COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS - A63F: CARD, BOARD, OR ROULETTE GAMES; INDOOR GAMES USING SMALL MOVING PLAYING BODIES; VIDEO GAMES; GAMES NOT OTHERWISE PROVIDED FOR ## Abstract Implementations relate to systems, methods, and computer-readable media to generate text tags for games. A computer-implemented method is provided to generate one or more text tags for a game using a trained machine learning model. Data that includes a game identifier of the game and a set of digital assets associated with the game are provided as input to the trained machine learning model. Predicted text tags are generated using the trained machine learning model based on the set of digital assets associated with the game. The text tags are associated with a respective prediction score. One or more text tags are selected from the plurality of predicted text tags based on the respective prediction score. ## Claims ### Claim 1 A computer-implemented method to validate one or more text tags for a game, the method comprising: providing, as input to a plurality of machine learning models, data that includes a game identifier of the game and a set of digital assets of different classes associated with the game; generating, based on the set of digital assets and by respective models of the plurality of machine learning models, a respective feature vector of a plurality of feature vectors; generating, based on the plurality of feature vectors, a plurality of predicted text tags for the game, each predicted text tag associated with a respective prediction score; selecting one or more text tags from the plurality of predicted text tags based on the respective prediction score; obtaining a developer provided text tag for the game; validating the developer provided text tag for the game by comparing the developer provided text tag with the one or more selected text tags; determining that the developer provided text tag is irrelevant based on the comparison; and removing the developer provided text tag from association with the game. ### Claim 2 (depends on 1) The computer-implemented method of claim 1, wherein generating the plurality of predicted text tags comprises selecting from a collection of words associated with a game platform that hosts the game, wherein the selecting is based on the plurality of prediction scores. ### Claim 3 (depends on 1) The computer-implemented method of claim 1, wherein generating the plurality of predicted text tags comprises generating predicted text tags based on words included in a language dictionary. ### Claim 4 (depends on 1) The computer-implemented method of claim 1, wherein at least one of the plurality of machine learning models includes a convolutional neural network (CNN) and wherein the feature vector generated by the at least one of the plurality of machine learning models is based on a set of image assets from the set of digital assets. ### Claim 5 (depends on 1) The computer-implemented method of claim 1, further comprising displaying a user interface that includes a list of games associated with at least one of the one or more selected text tags. ### Claim 6 (depends on 1) The computer-implemented method of claim 1, further comprising displaying a user interface that includes the one or more selected text tags along with a game icon associated with the game. ### Claim 7 (depends on 1) The computer-implemented method of claim 1, wherein the digital assets comprise one or more of game objects, meshes, avatars, game source code, game configuration parameters, game lighting, avatar movements, text content of the game, game sounds, game background music, code coverage, or frequency of use of digital assets during gameplay. ### Claim 8 (depends on 1) The computer-implemented method of claim 1, further comprising training the plurality of machine learning models, wherein the training comprises, for each model: providing, as input to the model, training data that includes a plurality of game identifiers and a respective set of digital assets of different classes associated with each game identified by the game identifiers; generating, based on the set of digital assets and by the model, a respective feature vector; generating, based on the feature vector, a plurality of predicted text tags; comparing the plurality of predicted tags with respective tags associated with each game identified by the game identifiers; and adjusting one or more parameters of the model based on the comparison of the plurality of predicted tags with respective tags associated with each game identified by the game identifiers. ### Claim 9 (depends on 1) The computer-implemented method of claim 1, wherein each machine learning model utilizes a respective subset of the set of digital assets that includes assets of a particular class. ### Claim 10 A non-transitory computer-readable medium comprising instructions that, responsive to execution by a processing device, causes the processing device to perform operations to validate one or more text tags for a game comprising: providing, as input to a plurality of machine learning models, data that includes a game identifier of the game and a set of digital assets of different classes associated with the game; generating, based on the set of digital assets and by respective models of the plurality of machine learning models, a respective feature vector of a plurality of feature vectors; generating, based on the plurality of feature vectors, a plurality of predicted text tags for the game, each predicted text tag associated with a respective prediction score; selecting one or more text tags from the plurality of predicted text tags based on the respective prediction score; obtaining a developer provided text tag for the game; validating the developer provided text tag for the game by comparing the developer provided text tag with the one or more selected text tags; determining that the developer provided text tag is irrelevant based on the comparison; and removing the developer provided text tag from association with the game. ### Claim 11 (depends on 10) The non-transitory computer-readable medium of claim 10, wherein generating the plurality of predicted text tags comprises selecting from a collection of words associated with a game platform that hosts the game, wherein the selecting is based on the plurality of prediction scores. ### Claim 12 (depends on 10) The non-transitory computer-readable medium of claim 10, wherein at least one of the plurality of machine learning models includes a convolutional neural network (CNN) and wherein the feature vector generated by the at least one of the plurality of machine learning models is based on a set of image assets from the set of digital assets. ### Claim 13 (depends on 10) The non-transitory computer-readable medium of claim 10, wherein the operations further comprise displaying a user interface that includes a list of games associated with at least one of the one or more selected text tags. ### Claim 14 (depends on 10) The non-transitory computer-readable medium of claim 10, wherein the digital assets comprise one or more of game objects, meshes, avatars, game source code, game configuration parameters, game lighting, avatar movements, text content of the game, game sounds, game background music, code coverage, or frequency of use of digital assets during gameplay. ### Claim 15 (depends on 10) The non-transitory computer-readable medium of claim 10, wherein the operations further comprise displaying a user interface that includes the one or more selected text tags along with a game icon associated with the game. ### Claim 16 A system comprising: a memory with instructions stored thereon; and a processing device, coupled to the memory, the processing device configured to access the memory and execute the instructions, wherein the instructions cause the processing device to perform operations to validate one or more text tags for a game comprising: providing, as input to a plurality of machine learning models, data that includes a game identifier of the game and a set of digital assets of different classes associated with the game; generating, based on the set of digital assets and by respective models of the plurality of machine learning models, a respective feature vector of a plurality of feature vectors; generating, based on the plurality of feature vectors, a plurality of predicted text tags for the game, each predicted text tag associated with a respective prediction score; selecting one or more text tags from the plurality of predicted text tags based on the respective prediction score; obtaining a developer provided text tag for the game; validating the developer provided text tag for the game by comparing the developer provided text tag with the one or more selected text tags; determining that the developer provided text tag is irrelevant based on the comparison; and removing the developer provided text tag from association with the game. ### Claim 17 (depends on 16) The system of claim 16, wherein generating the plurality of predicted text tags comprises selecting from a collection of words associated with a game platform that hosts the game, wherein the selecting is based on the plurality of prediction scores. ### Claim 18 (depends on 16) The system of claim 16, wherein at least one of the plurality of machine learning models includes a convolutional neural network (CNN) and wherein the feature vector generated by the at least one of the plurality of machine learning models is based on a set of image assets from the set of digital assets. ### Claim 19 (depends on 16) The system of claim 16, wherein the digital assets comprise one or more of game objects, meshes, avatars, game source code, game configuration parameters, game lighting, avatar movements, text content of the game, game sounds, game background music, code coverage, or frequency of use of digital assets during gameplay. ### Claim 20 (depends on 16) The system of claim 16, wherein the operations further comprise training the plurality of machine learning models, wherein the training comprises, for each model: providing, as input to the model, training data that includes a plurality of game identifiers and a respective set of digital assets of different classes associated with each game identified by the game identifiers; generating, based on the set of digital assets and by the model, a respective feature vector; generating, based on the feature vector, a plurality of predicted text tags; comparing the plurality of predicted tags with respective tags associated with each game identified by the game identifiers; and adjusting one or more parameters of the model based on the comparison of the plurality of predicted tags with respective tags associated with each game identified by the game identifiers.