# US 2022/0044138 A1 · Game service execution method and apparatus based on artificial intelligence, device, and medium > Agents: this data is available as JSON from the patentgrep API and MCP server — see https://patentgrep.com/llms.txt - Number: US 2022/0044138 A1 - Document ID: US20220044138A1 - Status: Published application - Kind: A1 - Application number: 17/506,460 - Filed: 2021-10-20 - Published: 2022-02-10 - Earliest claimed: 2019-11-12 - Assignees: Tencent Technology (Shenzhen) Company Limited - Applicants: Tencent Technology (Shenzhen) Company Limited - Inventors: Jieqiong LIN - Web page: https://patentgrep.com/patent/US20220044138A1 ## CPC - G06N 5/04: Inference or reasoning models - G06N 5/01: Dynamic search techniques; Heuristics; Dynamic trees; Branch-and-bound - G06N 7/01: Probabilistic graphical models, e.g. probabilistic networks - G06N 20/00: Machine learning - G06N: COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS ## Abstract This application discloses a game service execution method performed at a computer device for a game service. The method includes: constructing, in an ith round of game in the game service including a static game, a Monte-Carlo tree corresponding to the ith round of game, where the Monte-Carlo tree includes: an initial node, at least two participant nodes, at least two logic distribution nodes, and at least two logic gate nodes that correspond to the static game; simulating a game result on each path in the Monte-Carlo tree; backpropagating benefits of the game results corresponding to multiple paths in the Monte-Carlo tree to a highest-layer AI participant node of the AI participant in the Monte-Carlo tree; and selecting an execution action of the highest-layer AI participant node on a path that has the highest benefit, and outputting the execution action as an AI execution action of the AI participant. ## Claims ### Claim 1 A game service execution method performed at a computer device for a game service in which n participants perform a game according to a game rule, n≥2 and n being an integer, the n participants comprising an artificial intelligence AI participant, and the method comprising: constructing, in an ith round of game in the game service comprising a static game, a Monte-Carlo tree corresponding to the ith round of game, the Monte-Carlo tree comprising: an initial node, at least two participant nodes, at least two logic distribution nodes, and at least two logic gate nodes that correspond to the static game, the logic distribution node being configured to forward, according to an arrangement combination, candidate execution actions corresponding to the at least two participant nodes in the static game to different logic gate nodes, each logic gate node being configured to output a target candidate execution action with a highest priority of at least two candidate execution actions in a same arrangement combination to a lower-layer node according to the game rule, and i being a positive integer; simulating a game result on each path in the Monte-Carlo tree, the game result including a benefit corresponding to the path in the Monte-Carlo tree; backpropagating benefits of the game results corresponding to multiple paths in the Monte-Carlo tree to a highest-layer AI participant node of the AI participant in the Monte-Carlo tree; and selecting an execution action of the highest-layer AI participant node on a path that has the highest benefit, and outputting the execution action as an AI execution action of the AI participant in the ith round of game. ### Claim 2 (depends on 1) The method according to claim 1, wherein the constructing, in an ith round of game that comprises a static game in the game service, a Monte-Carlo tree corresponding to the ith round of game comprises: constructing each layer of nodes in the Monte-Carlo tree from a root node in the ith round of game in the game service; performing expansion to the Monte-Carlo tree by using the at least two participant nodes participating in the static game as (j+1)th-layer nodes in the Monte-Carlo tree when a jth-layer node comprises an initial node used for triggering the static game, wherein j is an integer greater than 1; and predicting m candidate execution actions corresponding to the at least two participant nodes in the static game, and performing expansion the Monte-Carlo tree by using m logic distribution nodes in a one-to-one correspondence with the m candidate execution actions as (j+2)th-layer nodes in the Monte-Carlo tree, where m is a positive integer; and predicting P arrangement combinations corresponding to the m candidate execution actions, and performing expansion the Monte-Carlo tree by using P logic gate nodes in a one-to-one correspondence with the P arrangement combinations as (j+3)th-layer nodes in the Monte-Carlo tree, wherein each logic gate node is configured to output a target candidate execution action with a highest priority of at least two candidate execution actions in a same arrangement combination to the lower-layer node, and P is a positive integer. ### Claim 3 (depends on 1) The method according to claim 1, wherein the backpropagating benefits of the game results corresponding to multiple paths in the Monte-Carlo tree to a highest-layer AI participant node of the AI participant in the Monte-Carlo tree comprises: backpropagating the benefits corresponding to the game results from a leaf node in the Monte-Carlo tree to the at least two logic gate nodes; backpropagating the benefits corresponding to the game results from the logic gate nodes to the initial node; calculating average benefits corresponding to the static game by using the initial node; and backpropagating the average benefits from the initial node to the highest-layer AI participant node of the AI participant in the Monte-Carlo tree. ### Claim 4 (depends on 3) The method according to claim 3, wherein the logic gate nodes are set as (j+3)th-layer nodes in the Monte-Carlo tree; and the backpropagating the benefits corresponding to the game results from the logic gate nodes to the initial node comprises: backpropagating the benefit of the game result corresponding to each path from P logic gate nodes to m logic distribution nodes that are of (j+2)th-layer nodes and that correspond to the P logic gate nodes, wherein P and m are both positive integers, and j is an integer greater than 1; and backpropagating the benefit of the game result corresponding to each path from the m logic distribution nodes to at least two participant nodes that are of (j+1)th-layer nodes and that correspond to the m logic distribution nodes. ### Claim 5 (depends on 2) The method according to claim 2, wherein the constructing each layer of nodes in the Monte-Carlo tree from a root node further comprises: setting, when an execution action corresponding to the participant node at a (j−1)th layer is an execution action used for triggering the static game, an action distribution node corresponding to the execution action at a ith layer as the initial node used for triggering the static game, wherein j is an integer greater than 1; wherein the action distribution node is configured to forward the execution action corresponding to the participant node to the participant nodes corresponding to the at least two participants participating in the static game. ### Claim 6 (depends on 1) The method according to claim 1, wherein the execution action corresponding to the participant node comprises at least two of the following execution actions in response to that the game service comprises a roshambo service: executing a hand outstretching action corresponding to rock, executing a hand outstretching action corresponding to scissors, and executing a hand outstretching action corresponding to paper; the execution action corresponding to the participant node comprises at least two of the following execution actions in response to that the game service comprises a Mahjong service: taking a tile, discarding a tile, chow, pung, kong, and win; or the execution action corresponding to the participant node comprises the following two execution actions in response to that the game service comprises a sealed-bid auction service: a bidding action and a bidding dropping action. ### Claim 7 A computer device comprising a processor and a memory, the memory storing at least one computer program for a game service in which n participants perform a game according to a game rule, n≥2 and n being an integer, the n participants comprising an artificial intelligence AI participant, the at least one computer program being loaded and executed by the processor to perform a plurality of operations including: constructing, in an ith round of game in the game service comprising a static game, a Monte-Carlo tree corresponding to the ith round of game, the Monte-Carlo tree comprising: an initial node, at least two participant nodes, at least two logic distribution nodes, and at least two logic gate nodes that correspond to the static game, the logic distribution node being configured to forward, according to an arrangement combination, candidate execution actions corresponding to the at least two participant nodes in the static game to different logic gate nodes, each logic gate node being configured to output a target candidate execution action with a highest priority of at least two candidate execution actions in a same arrangement combination to a lower-layer node according to the game rule, and i being a positive integer; simulating a game result on each path in the Monte-Carlo tree, the game result including a benefit corresponding to the path in the Monte-Carlo tree; backpropagating benefits of the game results corresponding to multiple paths in the Monte-Carlo tree to a highest-layer AI participant node of the AI participant in the Monte-Carlo tree; and selecting an execution action of the highest-layer AI participant node on a path that has the highest benefit, and outputting the execution action as an AI execution action of the AI participant in the ith round of game. ### Claim 8 (depends on 7) The computer device according to claim 7, wherein the constructing, in an ith round of game that comprises a static game in the game service, a Monte-Carlo tree corresponding to the ith round of game comprises: constructing each layer of nodes in the Monte-Carlo tree from a root node in the ith round of game in the game service; performing expansion to the Monte-Carlo tree by using the at least two participant nodes participating in the static game as (j+1)th-layer nodes in the Monte-Carlo tree when a jth-layer node comprises an initial node used for triggering the static game, wherein j is an integer greater than 1; and predicting m candidate execution actions corresponding to the at least two participant nodes in the static game, and performing expansion the Monte-Carlo tree by using m logic distribution nodes in a one-to-one correspondence with the m candidate execution actions as (j+2)th-layer nodes in the Monte-Carlo tree, where m is a positive integer; and predicting P arrangement combinations corresponding to the m candidate execution actions, and performing expansion the Monte-Carlo tree by using P logic gate nodes in a one-to-one correspondence with the P arrangement combinations as (j+3)th-layer nodes in the Monte-Carlo tree, wherein each logic gate node is configured to output a target candidate execution action with a highest priority of at least two candidate execution actions in a same arrangement combination to the lower-layer node, and P is a positive integer. ### Claim 9 (depends on 7) The computer device according to claim 7, wherein the backpropagating benefits of the game results corresponding to multiple paths in the Monte-Carlo tree to a highest-layer AI participant node of the AI participant in the Monte-Carlo tree comprises: backpropagating the benefits corresponding to the game results from a leaf node in the Monte-Carlo tree to the at least two logic gate nodes; backpropagating the benefits corresponding to the game results from the logic gate nodes to the initial node; calculating average benefits corresponding to the static game by using the initial node; and backpropagating the average benefits from the initial node to the highest-layer AI participant node of the AI participant in the Monte-Carlo tree. ### Claim 10 (depends on 9) The computer device according to claim 9, wherein the logic gate nodes are set as (j+3)th-layer nodes in the Monte-Carlo tree; and the backpropagating the benefits corresponding to the game results from the logic gate nodes to the initial node comprises: backpropagating the benefit of the game result corresponding to each path from P logic gate nodes to m logic distribution nodes that are of (j+2)th-layer nodes and that correspond to the P logic gate nodes, wherein P and m are both positive integers, and j is an integer greater than 1; and backpropagating the benefit of the game result corresponding to each path from the m logic distribution nodes to at least two participant nodes that are of (j+1)th-layer nodes and that correspond to the m logic distribution nodes. ### Claim 11 (depends on 8) The computer device according to claim 8, wherein the constructing each layer of nodes in the Monte-Carlo tree from a root node further comprises: setting, when an execution action corresponding to the participant node at a (j−1)th layer is an execution action used for triggering the static game, an action distribution node corresponding to the execution action at a jth layer as the initial node used for triggering the static game, wherein j is an integer greater than 1; wherein the action distribution node is configured to forward the execution action corresponding to the participant node to the participant nodes corresponding to the at least two participants participating in the static game. ### Claim 12 (depends on 7) The computer device according to claim 7, wherein the execution action corresponding to the participant node comprises at least two of the following execution actions in response to that the game service comprises a roshambo service: executing a hand outstretching action corresponding to rock, executing a hand outstretching action corresponding to scissors, and executing a hand outstretching action corresponding to paper; the execution action corresponding to the participant node comprises at least two of the following execution actions in response to that the game service comprises a Mahjong service: taking a tile, discarding a tile, chow, pung, kong, and win; or the execution action corresponding to the participant node comprises the following two execution actions in response to that the game service comprises a sealed-bid auction service: a bidding action and a bidding dropping action. ### Claim 13 A non-transitory computer-readable storage medium, the computer-readable storage medium storing at least one computer program for a game service in which n participants perform a game according to a game rule, n≥2 and n being an integer, the n participants comprising an artificial intelligence AI participant, the at least one computer program being loaded and executed by a processor of a computer device to perform a plurality of operations including: constructing, in an ith round of game in the game service comprising a static game, a Monte-Carlo tree corresponding to the ith round of game, the Monte-Carlo tree comprising: an initial node, at least two participant nodes, at least two logic distribution nodes, and at least two logic gate nodes that correspond to the static game, the logic distribution node being configured to forward, according to an arrangement combination, candidate execution actions corresponding to the at least two participant nodes in the static game to different logic gate nodes, each logic gate node being configured to output a target candidate execution action with a highest priority of at least two candidate execution actions in a same arrangement combination to a lower-layer node according to the game rule, and i being a positive integer; simulating a game result on each path in the Monte-Carlo tree, the game result including a benefit corresponding to the path in the Monte-Carlo tree; backpropagating benefits of the game results corresponding to multiple paths in the Monte-Carlo tree to a highest-layer AI participant node of the AI participant in the Monte-Carlo tree; and selecting an execution action of the highest-layer AI participant node on a path that has the highest benefit, and outputting the execution action as an AI execution action of the AI participant in the ith round of game. ### Claim 14 (depends on 13) The non-transitory computer-readable storage medium according to claim 13, wherein the constructing, in an ith round of game that comprises a static game in the game service, a Monte-Carlo tree corresponding to the ith round of game comprises: constructing each layer of nodes in the Monte-Carlo tree from a root node in the ith round of game in the game service; performing expansion to the Monte-Carlo tree by using the at least two participant nodes participating in the static game as (j+1)th-layer nodes in the Monte-Carlo tree when a jth-layer node comprises an initial node used for triggering the static game, wherein j is an integer greater than 1; and predicting m candidate execution actions corresponding to the at least two participant nodes in the static game, and performing expansion the Monte-Carlo tree by using m logic distribution nodes in a one-to-one correspondence with the m candidate execution actions as (j+2)th-layer nodes in the Monte-Carlo tree, where m is a positive integer; and predicting P arrangement combinations corresponding to the m candidate execution actions, and performing expansion the Monte-Carlo tree by using P logic gate nodes in a one-to-one correspondence with the P arrangement combinations as (j+3)th-layer nodes in the Monte-Carlo tree, wherein each logic gate node is configured to output a target candidate execution action with a highest priority of at least two candidate execution actions in a same arrangement combination to the lower-layer node, and P is a positive integer. ### Claim 15 (depends on 13) The non-transitory computer-readable storage medium according to claim 13, wherein the backpropagating benefits of the game results corresponding to multiple paths in the Monte-Carlo tree to a highest-layer AI participant node of the AI participant in the Monte-Carlo tree comprises: backpropagating the benefits corresponding to the game results from a leaf node in the Monte-Carlo tree to the at least two logic gate nodes; backpropagating the benefits corresponding to the game results from the logic gate nodes to the initial node; calculating average benefits corresponding to the static game by using the initial node; and backpropagating the average benefits from the initial node to the highest-layer AI participant node of the AI participant in the Monte-Carlo tree. ### Claim 16 (depends on 15) The non-transitory computer-readable storage medium according to claim 15, wherein the logic gate nodes are set as (j+3)th-layer nodes in the Monte-Carlo tree; and the backpropagating the benefits corresponding to the game results from the logic gate nodes to the initial node comprises: backpropagating the benefit of the game result corresponding to each path from P logic gate nodes to m logic distribution nodes that are of (j+2)th-layer nodes and that correspond to the P logic gate nodes, wherein P and m are both positive integers, and j is an integer greater than 1; and backpropagating the benefit of the game result corresponding to each path from the m logic distribution nodes to at least two participant nodes that are of (j+1)th-layer nodes and that correspond to the m logic distribution nodes. ### Claim 17 (depends on 14) The non-transitory computer-readable storage medium according to claim 14, wherein the constructing each layer of nodes in the Monte-Carlo tree from a root node further comprises: setting, when an execution action corresponding to the participant node at a (j−1)th layer is an execution action used for triggering the static game, an action distribution node corresponding to the execution action at a jth layer as the initial node used for triggering the static game, wherein j is an integer greater than 1; wherein the action distribution node is configured to forward the execution action corresponding to the participant node to the participant nodes corresponding to the at least two participants participating in the static game. ### Claim 18 (depends on 13) The non-transitory computer-readable storage medium according to claim 13, wherein the execution action corresponding to the participant node comprises at least two of the following execution actions in response to that the game service comprises a roshambo service: executing a hand outstretching action corresponding to rock, executing a hand outstretching action corresponding to scissors, and executing a hand outstretching action corresponding to paper; the execution action corresponding to the participant node comprises at least two of the following execution actions in response to that the game service comprises a Mahjong service: taking a tile, discarding a tile, chow, pung, kong, and win; or the execution action corresponding to the participant node comprises the following two execution actions in response to that the game service comprises a sealed-bid auction service: a bidding action and a bidding dropping action.