Scheme evaluation device, situation evaluation method, and program
By storing and analyzing log data from past game play, the current game situation is evaluated, solving the application problem of shogi situation evaluation technology in e-sports and realizing effective situation evaluation in complex games.
Patent Information
- Application Number
- CN202380033170.7
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Priority Date
- 2022-04-11
- Filing Date
- 2023-04-05
- Publication Date
- 2024-11-22
AI Technical Summary
Existing technologies make it difficult to directly apply shogi position evaluation techniques to esports, especially due to the frequent updates to esports rules and game elements, which makes it difficult to effectively evaluate game positions.
By storing reference data generated based on past game play logs, the system obtains current game status information, retrieves similar reference status information, evaluates the game situation based on its game results, and outputs the evaluation results.
It enables reliable evaluation of the game situation and outputs easy-to-understand situation evaluation results in games with frequently updated rules and game elements, and is suitable for complex games such as e-sports.
Smart Images

Figure CN119013086A_ABST
Abstract
Description
Technical Field
[0001] The present invention relates to a situation evaluation device, a situation evaluation method, and a program. Background Art
[0002] In recent years, online broadcasts of shogi have become popular. In such broadcasts, since the situation evaluation (which side is advantageous) by a shogi AI (Artificial Intelligence) is displayed in real time, beginners can easily understand the complicated shogi situation. A technique for performing shogi situation evaluation is disclosed in Non-Patent Document 1.
[0003] Prior Art Documents
[0004] Non-Patent Documents
[0005] Non-Patent Document 1: Yoko Nishihara, Reona Takayama, Kensuke Hishida, and Ryosuke Yamanishi. 2018. Enjoy Watching Japanese Chess Games like Football: an Evaluation Method of Game Positions for Beginners. In Proceedings of the 2018 Annual Symposium on Computer-Human Interaction in Play Companion Extended Abstracts (CHIPLAY’18 Extended Abstracts). Association for Computing Machinery, New York, NY, USA, 569-575, [online], October 23, 2018, [retrieved March 24, 2022], Internet <URL: https: / / dl.acm.org / doi / 10.1145 / 3270316.<3271508> Summary of the Invention
[0006] Problems to be Solved by the Invention
[0007] The inventors of the present invention are interested in introducing situation evaluation AI into electronic sports (e-sports) so that even novices can intuitively understand the beauty of attack and defense in electronic sports where tactics are rapidly becoming complicated. However, unlike shogi, where the rules are generally fixed, in e-sports, AI that can cope with frequently updated rules and game elements is sought. The technology disclosed in non-patent document 1 is a technology for evaluating the situation of shogi, which is difficult to be directly used in the situation evaluation of e-sports due to the above reasons.
[0008] The present invention aims to provide a new game situation evaluation technology.
[0009] Solutions for solving problems
[0010] According to one embodiment of the present invention, there is provided a situation assessment device comprising:
[0011] a reference data storage unit storing reference data, the reference data being data generated based on log data of a plurality of game plays executed in the past, and the reference data including a plurality of combinations of reference state information indicating a game state in the game play and a game result determined after the game state;
[0012] an acquisition unit that acquires object state information, the object state information indicating a game state of a game play that is an evaluation object at an evaluation time point;
[0013] a retrieval unit configured to extract the reference state information similar to the object state information from the reference data;
[0014] an evaluation unit that evaluates the game situation of the game play as the evaluation object at the evaluation time point based on the game result associated with the extracted reference state information; and
[0015] An output unit outputs the evaluation result.
[0016] According to one embodiment of the present invention, there is provided a situation assessment method, wherein:
[0017] The computer stores reference data, the reference data being data generated based on log data of a plurality of game plays executed in the past, the reference data including a plurality of combinations of reference state information representing a game state in the game play and a game result determined after the game state,
[0018] In the situation assessment method, the following steps are performed by a computer:
[0019] An acquisition step of acquiring object state information, wherein the object state information indicates a game state of a game play as an evaluation object at an evaluation time point;
[0020] A retrieval step of extracting the reference state information similar to the object state information from the reference data;
[0021] an evaluation step of evaluating a game situation of the game play as the evaluation object at the evaluation time point based on the game result associated with the extracted reference state information; and
[0022] The output step outputs the evaluation result.
[0023] According to one embodiment of the present invention, there is provided a program for causing a computer to function as the following means:
[0024] A reference data storage unit storing reference data, the reference data being data generated based on log data of a plurality of game plays executed in the past, the reference data including a plurality of combinations of reference state information indicating a game state in the game play and a game result determined after the game state;
[0025] an acquisition unit, which acquires object state information, wherein the object state information represents a game state of a game play as an evaluation object at an evaluation time point;
[0026] a retrieval unit, which extracts the reference state information similar to the object state information from the reference data;
[0027] an evaluation unit that evaluates the game situation of the game play as the evaluation object at the evaluation time point based on the game result associated with the extracted reference state information; and
[0028] An output unit outputs the evaluation result.
[0029] Effects of the Invention
[0030] According to one embodiment of the present invention, a new game situation evaluation technology is realized. BRIEF DESCRIPTION OF THE DRAWINGS
[0031] The above-mentioned objects, other objects, features and advantages will be further clarified through the common embodiments described below and the accompanying drawings below.
[0032] Figure 1 This is a diagram showing an example of the hardware configuration of the situation evaluation device.
[0033] Figure 2 This is a diagram showing an example of a functional block diagram of a situation assessment device.
[0034] Figure 3This is a diagram schematically showing an example of information realized by the situation assessment device.
[0035] Figure 4 This is a flowchart showing an example of the flow of processing of the situation evaluation device.
[0036] Figure 5 This is a diagram showing another example of a functional block diagram of the situation assessment device.
[0037] Figure 6 This is a diagram schematically showing an example of a game board according to the embodiment.
[0038] Figure 7 This is a diagram schematically showing an example of the architecture of the embodiment.
[0039] Figure 8 It is a diagram schematically showing the structure of log data and reference data according to the embodiment.
[0040] Fig. 9 This is a diagram schematically showing an example of a vector representing a game state according to the embodiment. DETAILED DESCRIPTION
[0041] Hereinafter, the embodiments of the present invention will be described using the drawings. In addition, in all the drawings, the same components are denoted by the same reference numerals, and the description thereof will be appropriately omitted.
[0042] <First Embodiment>
[0043] "summary"
[0044] The situation evaluation device of the present embodiment evaluates the game situation of the game play as the evaluation target at the evaluation time point based on the log data of a plurality of game plays executed in the past.
[0045] More specifically, the situation evaluation device stores reference data, which is data generated based on log data of game play executed multiple times in the past, and the reference data includes a plurality of combinations of reference state information indicating a game state in the game play and a game result determined after the game state. That is, the reference data is data obtained by associating the game state actually generated in the game play executed multiple times in the past with the actual game result after such game state is reached.
[0046] Then, when the situation evaluation device obtains the object state information indicating the game state of the game play as the evaluation object at the evaluation time point, it extracts the reference state information similar to the object state information from the reference data. Next, the situation evaluation device evaluates the game situation of the game play as the evaluation object at the evaluation time point based on the game result associated with the extracted reference state information.
[0047] "Hardware Structure"
[0048] Next, an example of the hardware structure of the situation evaluation device is described. The functional parts of the situation evaluation device are realized by any combination of hardware and software, centering on the CPU (Central Processing Unit) of any computer, memory, program loaded into the memory, storage unit such as a hard disk storing the program (in addition to being able to store programs pre-stored since the device was shipped from the factory, it is also possible to store programs downloaded from storage media such as CDs (Compact Discs), servers on the Internet, etc.), and a network connection interface. Moreover, those skilled in the art should understand that there are various variations of the implementation method and device.
[0049] Figure 1 is a block diagram illustrating the hardware structure of the situation assessment device. Figure 1 As shown, the situation assessment device includes a processor 1A, a memory 2A, an input / output interface 3A, a peripheral circuit 4A, and a bus 5A. The peripheral circuit 4A includes various modules. The situation assessment device may not include the peripheral circuit 4A. In addition, the situation assessment device may also be composed of a plurality of devices separated physically and / or logically. In this case, the plurality of devices may each have the above-mentioned hardware structure.
[0050] The bus 5A is a data transmission path for the processor 1A, the memory 2A, the peripheral circuit 4A, and the input / output interface 3A to send and receive data to and from each other. The processor 1A is, for example, a CPU, a GPU (Graphics Processing Unit), and other computing devices. The memory 2A is, for example, a RAM (Random Access Memory), a ROM (Read Only Memory), and other memories. The input / output interface 3A includes an interface for obtaining information from an input device, an external device, an external server, an external sensor, a camera, and the like, and an interface for outputting information to an output device, an external device, an external server, and the like. The input device is, for example, a keyboard, a mouse, a microphone, a physical button, a touch panel, and the like. The output device is, for example, a display, a speaker, a printer, an email sender, and the like. The processor 1A is able to issue instructions to each module and perform operations based on the operation results of each module.
[0051] "Functional Structure"
[0052] Next, the functional structure of the situation assessment device of this embodiment is described in detail. Figure 2 An example of a functional block diagram of the situation assessment device 10 of this embodiment is shown in FIG. Figure 2As shown, the situation assessment device 10 includes a reference data storage unit 11 , an acquisition unit 12 , a search unit 13 , an assessment unit 14 , and an output unit 15 .
[0053] The reference data storage unit 11 stores reference data generated based on log data of multiple past game plays. The reference data is data that associates actual game states that occurred in multiple past game plays with actual game results that were determined after the game states occurred.
[0054] "Log data" is data that can identify the game state during game play, the actions (operations selected by the player) selected by the player in the game, and the game results that are subsequently determined. For example, log data is a sequence of the initial game state and the initial action, then the game state as the result obtained by the action and the next action, and finally ends with the final state that determines the winner.
[0055] The "game state" can be obtained from the game and is determined by the value of at least one parameter that can affect the game result. The parameter used to determine the game state is referred to as the game state determination parameter. The value of the game state determination parameter can change as the game progresses. There are various parameters that can be set as the game state determination parameter, which can be determined according to the type and content of the game.
[0056] For example, in the case of a battle-type card game, the card information on the board, the number of hand cards (Japanese: handbook), the number of discarded cards, the physical strength of the player, various points (the points consumed to put cards on the field, the points consumed to make the cards evolve, etc.), the number of rounds, the captain role information, etc. are illustrated, but not limited to these. In addition, the battle-type card game is used as an example to illustrate this, but the types of games that can be used in this embodiment are not limited to this.
[0057] In the game, the value of the game state determination parameter can be a numerical value, a variable name, etc. In the reference data, the value of each game state determination parameter can also directly use the numerical value and variable name used in the game. However, considering the ease of retrieval by the retrieval unit 13 described below, it is preferred to set the value obtained by standardizing the numerical value and variable name used in the game according to the prescribed rules (for example: 0 to 1) and encoding it into a vector as the value of the game state determination parameter in the reference data. In this way, the similarity between certain specific game states can be quantitatively measured as the similarity between vectors.
[0058] "Actions selected by the player in the game (operations selected by the player)" are actions selected by the player's input in the game. For example, in the case of a battle card game, the use of cards (which card is used) is exemplified.
[0059] The “game result” is defined according to the type and content of the game. For example, in the case of a battle card game, the game results may include “the first player wins” and “the second player wins”. In addition, in the case of a game that can produce a result such as “draw”, the game result may also include “draw”. In addition, in the case of a game in which players compete with each other based on scores, the game results may also include “the first player wins with a difference of P points or more”, “the first player wins with a difference of P points or less”, “the second player wins with a difference of P points or more”, and “the second player wins with a difference of P points or less”.
[0060] The acquisition unit 12 acquires object state information indicating the game state of the game play to be evaluated at the evaluation time point. In the object state information, the game state is indicated by the value of the game state identification parameter similarly to the reference data.
[0061] The acquisition unit 12 can acquire various information such as object status information from the game server of the online game and the player terminal (personal computer, smartphone, tablet terminal, mobile phone, game terminal, etc.) The acquisition unit 12 can repeatedly acquire the object status information at regular intervals during the game execution.
[0062] The search unit 13 extracts reference state information similar to the object state information acquired by the acquisition unit 12 from the reference data stored in the reference data storage unit 11. The search unit 13 can perform the extraction by nearest neighbor search or approximate nearest neighbor search.
[0063] For example, reference data is grouped based on the characteristics of game play. That is, reference data with the same characteristics of game play, or reference data similar to each other at a predetermined level or above are grouped together. Moreover, the reference data storage unit 11 stores reference data for each group. The characteristics of game play used for grouping are various, and can be determined according to the type and content of the game. For example, in the case of a battle-type card game, a character selected by the player can also be used. In addition, the values of game parameters such as the number of rounds can also be used as the characteristics of the above-mentioned game play. In addition, player characteristics (game proficiency (advanced, intermediate, beginner), gender, nationality, age group, number of years of playing games, etc.) can also be used as the characteristics of the above-mentioned game play. Moreover, the retrieval unit 13 determines the group to which the game play as the evaluation object belongs based on the characteristics of the game play as the evaluation object (or the characteristics of the game play as the evaluation object at the evaluation time point), and extracts reference state information similar to the object state information from the reference data belonging to the determined group. In addition, a specific example of this structure is described in the following embodiment.
[0064] The evaluation unit 14 evaluates the game situation of the game play as the evaluation object at the evaluation time point based on the game result associated with the reference state information extracted by the search unit 13. The evaluation unit 14 calculates the probability that the evaluation result of the game play as the evaluation object will become each of the plurality of game results defined as above ("first player wins", "second player wins", etc.) based on the game result associated with the reference state information extracted by the search unit 13.
[0065] Specifically, the evaluation unit 14 calculates the ratio of each of the plurality of game results included in the plurality of game results associated with the plurality of reference state information extracted by the search unit 13 as the above-mentioned probability of each of the plurality of game results. For example, when M1 reference state information is extracted by the search unit 13, and M2 of the reference state information is associated with the game result of "the first player wins", the evaluation unit 14 calculates M2 / M1 as the probability of the first player winning.
[0066] In addition, the evaluation unit 14 may also include a unit for correcting the above-mentioned calculated probabilities. As an example, the evaluation unit 14 may also correct the above-mentioned calculated probabilities based on the similarity between the set of multiple reference state information extracted by the retrieval unit 13 and the object state information. The evaluation unit 14 performs correction in such a manner that the higher the similarity, the higher the above-mentioned probability after the correction processing. For example, the correction processing may also be performed according to a rule such as "no correction is performed when the similarity is above the reference value, and the above-mentioned calculated probability is reduced when the similarity is less than the reference value." The reduction range may be set to correspond to the size of the deviation between the similarity and the reference value. In addition, the correction rule is only an example and is not limited to this.
[0067] "The similarity between the set of multiple reference state information extracted by the retrieval unit 13 and the object state information" can be set to, for example, a statistical value (average value, maximum value, minimum value, mode value, median value, etc.) of the similarity between each reference state information among the multiple reference state information extracted by the retrieval unit 13 and the object state information, or it can be calculated by other units.
[0068] As another example of a unit for correcting the above-calculated probability, the evaluation unit 14 may correct the above-calculated probability based on the deviation between the moment when the game play based on each of the multiple reference state information extracted by the retrieval unit 13 is executed and the current time point. The evaluation unit 14 performs correction in such a manner that the smaller the above-mentioned deviation is, the higher the above-mentioned probability after the correction processing is. For example, the correction processing may also be performed according to the rule that "no correction is performed when the above-mentioned deviation is less than the reference value, and the above-mentioned calculated probability is reduced when the above-mentioned deviation is greater than the reference value". The reduction range may be set to correspond to the size of the difference between the above-mentioned deviation and the reference value. In addition, the correction rule is only an example and is not limited to this.
[0069] "The moment when the game play based on each of the multiple reference state information extracted by the retrieval unit 13 is executed" can be set to a statistical value (earliest moment, latest moment, mode value, median value, etc.) of the moment (date and time) when the game play based on each of the multiple reference state information extracted by the retrieval unit 13 is executed.
[0070] In this way, the evaluation unit 14 can evaluate the game situation by measuring (counting, correcting, etc.) the game results associated with the reference state information extracted by the search unit 13 .
[0071] The output unit 15 outputs the result of the evaluation of the game situation by the evaluation unit 14. For example, the result of the evaluation of the game situation output from the output unit 15 may be input to the game server. Furthermore, the result of the evaluation of the game situation may be displayed on the game screen of the game being evaluated. In addition, the result of the evaluation of the game situation output from the output unit 15 may be input to a system for generating a screen for game viewers. Furthermore, the result of the evaluation of the game situation may be displayed on the screen for game viewers. Figure 3 An example of the result of the evaluation of the game situation displayed on the game screen or on the screen for game viewers is shown in FIG. Figure 3 In the example shown, the game result is "the first player wins" or "the second player wins", and the probability of each game result is shown as a result of the evaluation of the game situation.
[0072] In addition, the situation evaluation device 10 having the above functions may also be implemented on the game server. That is, the game server and the situation evaluation device 10 may also be physically and / or logically integrated. In addition, the situation evaluation device 10 may also be implemented on the player terminal. That is, the player terminal and the situation evaluation device 10 may also be physically and / or logically integrated. In addition, the situation evaluation device 10 may also be an independent device physically and / or logically separated from the game server and the player terminal.
[0073] Next, use Figure 4 The following process is a process for evaluating the game situation of the game play as the evaluation object at the evaluation time point. Prior to this process, reference data is pre-stored in the reference data storage unit 11. The reference data is data generated based on log data of multiple game plays executed in the past, and the reference data includes a plurality of combinations of reference state information indicating the game state in the game play and the game result determined after the game state.
[0074] First, the acquisition unit 12 acquires object state information indicating the game state of the game play as the evaluation object at the evaluation time point (S10). Next, the search unit 13 extracts reference state information similar to the object state information acquired in S10 from the reference data stored in the reference data storage unit 11 (S11). Next, the evaluation unit 14 evaluates the game situation of the game play as the evaluation object at the evaluation time point based on the game results associated with the reference state information extracted in S11 (S12). Then, the output unit 15 outputs the evaluation result calculated in S12, that is, the information indicating the game situation of the game play as the evaluation object at the evaluation time point (S13).
[0075] The situation evaluation device 10 can repeatedly perform the above-mentioned processing while the game to be evaluated is being played.
[0076] "Effects"
[0077] According to the situation evaluation device 10 of this embodiment, the game situation of the game play as the evaluation object at the evaluation time point can be evaluated based on the game state that can be generated during the game play and the actual performance of the game result after the game state is reached. Therefore, the game situation can be evaluated without considering the rules and game elements of the game. As a result, the game situation can be evaluated using this technology even in games where the rules and game elements are frequently updated.
[0078] Furthermore, it is known that the situation evaluation device 10 calculates the probability of each game result in such a state based on past log data and outputs the probability as the result of the evaluation of the game situation. Therefore, it is possible to output the evaluation result that is easy for the viewer to understand and accept.
[0079] <Second Embodiment>
[0080] In the second embodiment, the structure of the situation assessment device 10 is concretely described. Specifically, the structure of the database of reference data and the process of searching the database are concretely described.
[0081] Even if the game states are the same, the game results may differ depending on the timing of the game states, that is, whether it is the beginning, middle, or end of the game, etc. The situation evaluation device 10 of the second embodiment has a configuration capable of evaluating the game situation in consideration of this.
[0082] The game of the second embodiment is a battle-type game in which the first player and the second player take actions alternately. For example, a battle-type card game or the like is suitable for this game.
[0083] In the second embodiment, the reference data is grouped based on the number of actions counted from the first hand (sometimes also referred to as the "number of rounds"). That is, the reference data storage unit 11 groups the plurality of reference data based on the number of actions counted from the first hand, and stores the reference data separately for each group.
[0084] For example, groups may be created by the number of actions. In this case, a group corresponding to the number of actions 1, a group corresponding to the number of actions 2, and the like are created. In addition, groups may be created by multiple numbers of actions. For example, a predetermined number of consecutive numbers of actions may be grouped together. In this case, for example, groups corresponding to the number of actions 1 to 3, a group corresponding to the number of actions 4 to 6, and the like are created.
[0085] The acquisition unit 12 also acquires the number of actions of the game play as the evaluation target at the evaluation time point. Then, the search unit 13 extracts reference state information similar to the target state information from the reference data belonging to the group corresponding to the number of actions of the game play as the evaluation target at the evaluation time point.
[0086] In addition, the retrieval unit 13 can also extract reference state information similar to the object state information from reference data belonging to one or more groups corresponding to each number of actions included in the following range: from (the number of actions of the game as the evaluation object at the evaluation time point - α) to (the number of actions of the game as the evaluation object at the evaluation time point + α). α is a positive integer, and the specific value is a design matter, but it can be set to about 3 to 5, for example. Even if the number of actions is slightly different, as long as the game states are the same state, the same game results are often obtained. Not only the reference data belonging to the group corresponding to the number of actions of the game as the evaluation object at the evaluation time point is set as the retrieval object, but also the reference data belonging to the multiple groups as described above is set as the retrieval object, thereby increasing the number of reference data extracted. As a result, a more reliable evaluation result of the game situation can be obtained.
[0087] The other structures of the situation assessment device 10 of the second embodiment are the same as those of the situation assessment device 10 of the first embodiment.
[0088] According to the situation evaluation device 10 of the second embodiment, the same effects as those of the situation evaluation device 10 of the first embodiment are achieved. In addition, according to the situation evaluation device 10 of the second embodiment, the game situation of the game to be evaluated at the evaluation time point can be evaluated based on the past performance of the game results after the same game state as the game to be evaluated is reached at the time point of the same number of actions as the game to be evaluated at the evaluation time point.
[0089] Even if the game states are the same, the game results may be different depending on the timing of the game states, that is, whether it is the beginning, middle, or end of the game. According to the situation evaluation device 10 of the second embodiment as described above, the game situation can be evaluated by considering the timing of the game play as the evaluation object at the evaluation time point. Therefore, a more reliable evaluation can be performed. In addition, the reference data set as the search object can be locked based on the number of actions, so the processing burden of the computer is reduced.
[0090] <Third Embodiment>
[0091] In the third embodiment, the structure of the situation assessment device 10 is concretely described. Specifically, the structure of the database of reference data and the process of searching the database are concretely described.
[0092] Even if the game states are the same, the game results may be different depending on the characters selected by the players. The situation evaluation device 10 of the third embodiment has a configuration capable of evaluating the game situation in consideration of this point.
[0093] The game of the third embodiment is a game in which the first player and the second player each select a role to play. For example, a card game of a battle type fits this game. In addition, the name of "role" can be different for each game. The "role" of the third embodiment includes concepts such as role, occupation (class), attribute, etc.
[0094] Furthermore, in the third embodiment, the reference data is grouped based on the combination of the character selected by the first player and the character selected by the second player. That is, the reference data storage unit 11 groups a plurality of reference data according to the combination of the character selected by the first player and the character selected by the second player, and stores the reference data separately for each group. For example, in the reference data storage unit 11, reference data is stored for each combination of the characters selected by the first player and the second player, such as "reference data when the first player selects character A and the second player selects character B" and "reference data when the first player selects character A and the second player selects character C".
[0095] The acquisition unit 12 further acquires information indicating the character selected by the first player and the character selected by the second player in the game play to be evaluated. Then, the search unit 13 extracts reference state information similar to the target state information from the reference data belonging to the group corresponding to the following combination: the combination of the character selected by the first player and the character selected by the second player in the game play to be evaluated.
[0096] In addition, the situation assessment device 10 of the third embodiment may also have the structure of the situation assessment device 10 of the second embodiment. In this case, the reference data is grouped based on the number of actions counted from the first player and the combination of the character selected by the first player and the character selected by the second player. That is, the reference data storage unit 11 groups a plurality of reference data based on the number of actions counted from the first player and the combination of the character selected by the first player and the character selected by the second player, and stores the reference data separately for each group.
[0097] Then, the search unit 13 extracts reference state information similar to the target state information from reference data belonging to a group corresponding to the following combination: a combination of a character selected by the first player and a character selected by the second player in the game play being the evaluation target and corresponding to the number of actions at the evaluation time point.
[0098] The other structures of the situation assessment device 10 of the third embodiment are the same as those of the situation assessment device 10 of the first and second embodiments.
[0099] According to the situation evaluation device 10 of the third embodiment, the same effects as those of the situation evaluation device 10 of the first and second embodiments are achieved. In addition, according to the situation evaluation device 10 of the third embodiment, in a situation where the first player and the second player have selected the same character as the game to be evaluated, the game situation of the game to be evaluated at the evaluation time point can be evaluated based on the past performance of the game results after the game state becomes the same as the game to be evaluated.
[0100] Even if the game states become the same, the game results may be different depending on the characters selected by the first and second players. According to the situation evaluation device 10 of the third embodiment as described above, the game situation can be evaluated by considering the characters selected by the first and second players in the game play as the evaluation object. Therefore, a more reliable evaluation can be performed. In addition, the reference data set as the search object can be locked based on the characters selected by the first and second players, so the processing burden of the computer is reduced. In addition, by also adopting the structure of the situation evaluation device 10 of the second embodiment, a more reliable evaluation can be performed and the processing burden of the computer can be further reduced.
[0101] <Fourth Embodiment>
[0102] In the fourth embodiment, the structure of the situation assessment device 10 is concretely described. Specifically, the structure of the database of reference data and the process of searching the database are concretely described.
[0103] Even if the game states are the same, the game results may be different depending on the characteristics of the players. The situation evaluation device 10 of the fourth embodiment has a configuration capable of evaluating the game situation in consideration of this point.
[0104] In the fourth embodiment, the reference data is grouped based on the combination of the characteristics of the first player and the characteristics of the second player. That is, the reference data storage unit 11 groups a plurality of reference data according to the combination of the characteristics of the first player and the characteristics of the second player, and stores the reference data separately for each group. The player characteristics are the proficiency of the game (advanced, intermediate, beginner), gender, nationality, age group, number of years of playing the game, etc. For example, in the reference data storage unit 11, reference data is stored for each combination of the characteristics of the first player and the second player, such as "reference data when the first player and the second player are advanced players" and "reference data when the first player is an advanced player and the second player is an intermediate player".
[0105] The acquisition unit 12 acquires the characteristics of the first player and the second player in the game play to be evaluated. For example, the characteristics of each of a plurality of players may be pre-registered in the game server. Furthermore, the acquisition unit 12 may acquire the characteristics of the first player and the second player in the game play to be evaluated from the game server.
[0106] Then, the search unit 13 extracts the reference state information similar to the target state information from the reference data belonging to the group corresponding to the combination of the characteristics of the first player and the second player in the game play to be evaluated.
[0107] In addition, the situation evaluation device 10 of the fourth embodiment may also have the structure of the situation evaluation device 10 of the second embodiment. In this case, the reference data is grouped based on the number of actions counted from the first hand and the combination of the characteristics of the first player and the characteristics of the second player. That is, the reference data storage unit 11 groups a plurality of reference data based on the number of actions counted from the first hand and the combination of the characteristics of the first player and the characteristics of the second player, and stores the reference data separately for each group.
[0108] Then, the search unit 13 extracts reference state information similar to the target state information from reference data belonging to a group corresponding to the following combination: a combination of characteristics of the first player and the second player in the game play to be evaluated and corresponding to the number of actions at the evaluation time point.
[0109] In addition, the situation evaluation device 10 of the fourth embodiment may also have the structure of the situation evaluation device 10 of the third embodiment. In this case, the reference data is grouped based on the combination of the character selected by the first player and the character selected by the second player, and the combination of the player characteristics of the first player and the player characteristics of the second player. That is, the reference data storage unit 11 groups a plurality of reference data based on the combination of the character selected by the first player and the character selected by the second player, and the combination of the player characteristics of the first player and the player characteristics of the second player, and stores the reference data separately for each group.
[0110] Then, the retrieval unit 13 extracts reference state information similar to the object state information from reference data belonging to a group corresponding to the following combination of characters and corresponding to the following combination of player characteristics: a combination of a character selected by the first player and a character selected by the second player in the game play that is the evaluation object, and a combination of the player characteristics of the first player and the player characteristics of the second player in the game play that is the evaluation object.
[0111] In addition, the situation evaluation device 10 of the fourth embodiment may also have the structure of the situation evaluation device 10 of the second embodiment and the third embodiment. In this case, the reference data is grouped based on the number of actions counted from the first hand, the combination of the character selected by the first hand and the character selected by the second hand, and the combination of the player characteristics of the first hand and the player characteristics of the second hand. That is, the reference data storage unit 11 groups a plurality of reference data based on the number of actions counted from the first hand, the combination of the character selected by the first hand and the character selected by the second hand, and the combination of the player characteristics of the first hand and the player characteristics of the second hand, and stores the reference data separately for each group.
[0112] Then, the retrieval unit 13 extracts reference state information similar to the object state information from reference data belonging to a group corresponding to the following combination of characters and the following combination of player characteristics in the game play that is the evaluation object, and corresponding to the number of actions at the evaluation time point: a combination of a character selected by the first player and a character selected by the second player in the game play that is the evaluation object, and a combination of the player characteristics of the first player and the player characteristics of the second player in the game play that is the evaluation object.
[0113] The other structures of the situation assessment device 10 of the fourth embodiment are the same as the structures of the situation assessment device 10 of the first to third embodiments.
[0114] According to the situation evaluation device 10 of the fourth embodiment, the same effects as those of the situation evaluation device 10 of the first to third embodiments are achieved. In addition, according to the situation evaluation device 10 of the fourth embodiment, in a game played by a first player and a second player having the same player characteristics as the game played as the evaluation object, the game situation of the game played as the evaluation object at the evaluation time point can be evaluated based on the past performance of the game results after the game state becomes the same as that of the game played as the evaluation object.
[0115] Even if the game states become the same, the game results may be different depending on the characteristics of the first and second players. According to the situation evaluation device 10 of the fourth embodiment as described above, the game situation can be evaluated by considering the characteristics of the first and second players in the game play as the evaluation object. Therefore, a more reliable evaluation can be performed. In addition, the reference data set as the search object can be locked based on the characteristics of the first and second players, so the processing burden of the computer is reduced. In addition, by also adopting the structure of at least one of the situation evaluation devices 10 of the second and third embodiments, a more reliable evaluation can be performed and the processing burden of the computer can be further reduced.
[0116] <Fifth Embodiment>
[0117] The situation assessment device 10 of this embodiment also has the function of creating the above-mentioned reference data. Figure 5 An example of a functional block diagram of the situation assessment device 10 according to the fifth embodiment is shown in FIG. Figure 5 As shown in FIG. 1 , the situation assessment device 10 further includes a log data storage unit 16 and a reference data creation unit 17 , which is different from the first to fourth embodiments.
[0118] The log data storage unit 16 stores log data of a plurality of game plays executed in the past. The log data is stored in association with various attribute information. Examples of the attribute information include the above-mentioned player characteristics, the date and time when the game play was executed, and the like.
[0119] The reference data creating unit 17 creates the reference data described in the first to fourth embodiments based on the log data stored in the log data storage unit 16 , and stores the reference data in the reference data storage unit 11 .
[0120] Furthermore, the reference data generating unit 17 may extract predetermined log data from the log data stored in the log data storing unit 16 and generate reference data based on the extracted log data. In this case, the unextracted log data is not used for generating the reference data.
[0121] For example, the reference data generating unit 17 may extract log data of game play performed in the past by a player whose player characteristics satisfy a predetermined condition. The predetermined condition may be, for example, a proficiency level of a predetermined level or higher (eg, advanced).
[0122] In addition to this, the reference data creating unit 17 may extract log data for which the game play was executed on a date and time before a predetermined date and time (that is, relatively new).
[0123] The other structures of the situation assessment device 10 of the fifth embodiment are the same as the structures of the situation assessment device 10 of the first to fourth embodiments.
[0124] The situation assessment device 10 of the fifth embodiment achieves the same effects as the situation assessment devices 10 of the first to fourth embodiments. In addition, the situation assessment device 10 of the fifth embodiment can extract only predetermined log data from log data and create reference data based on the extracted log data.
[0125] For example, according to the situation evaluation device 10 of the fifth embodiment, it is possible to extract only the log data of players with high proficiency and create reference data based on the extracted log data. In the case where the player's proficiency is low, the actions taken by the player after entering a certain game state may be greatly deviated. As a result, the game results after entering such a game state may also be greatly deviated. If the log data of such low-proficiency players and the log data of high-proficiency players are combined to create reference data and evaluate the game situation, the reliability of the evaluation result becomes low. In addition, it is believed that the audience of the game wants the evaluation results of the game situation from the perspective of advanced players.
[0126] By extracting only the log data of highly skilled players and creating reference data based on the extracted log data, it is possible to create an evaluation result that is highly reliable and more in line with the needs of viewers (the evaluation result of the game situation from the perspective of an advanced player). In addition, the following actions can be achieved naturally and reasonably: "For a board that seems to have no difference in the eyes of ordinary people but is very different for a highly skilled player, a high probability of 90% or so is used to indicate the outcome, and for a board that is difficult for a highly skilled player to predict the outcome, such as the early to mid-game, a low probability of about 50% is used to indicate the outcome."
[0127] In addition, according to the situation evaluation device 10 of the fifth embodiment, only the log data of the date and time of executing the game play before the specified date and time (i.e., relatively new) can be extracted and reference data can be created based on the extracted log data. As time passes, new game strategies are developed and the way of playing the game also changes. Therefore, the following situation may occur: in the past, when a certain game state was reached, the probability of a certain game result was high, but it may not be the case recently.
[0128] By extracting only relatively new log data and creating reference data based on the extracted log data, a highly reliable evaluation result can be generated.
[0129] <Example>
[0130] "summary"
[0131] In the present embodiment, AI (situation evaluation device 10) evaluates the game situation of PvP game by establishing an approximate nearest neighbor search that associates the changing process of the board (game state) with the victory and defeat result (game result) in the state of treating the game environment as a black box. The technical feature of the present embodiment is the following aspects: when predicting the victory and defeat from a certain board, the situation similar to the board is searched according to the past data classified by each combination of the characters who have fought and each action number when counting from the first hand, and the probability of victory and defeat is calculated according to the victory and defeat result of the similar situation. Thus, the system does not need to learn the rules and environment of the game, that is, it can learn only by the pair of the board and the victory and defeat result in the game play in the state of black boxing the game system, so it can be widely used in PvP games.
[0132] Specifically, in this embodiment, for all log data of game play that is the learning object, each board consisting of them is transformed into a feature vector of several thousand dimensions, and a database for approximate nearest neighbor search is constructed using these data for each combination of characters that have fought and each number of actions when counting from the first hand. For example, in the case of a game with 8 characters and the battle ends after an average of 30 actions, 8×8×30=720 databases are constructed. This is equivalent to creating 720 groups described in the above embodiment.
[0133] When the game is executed, that is, in a situation where the situation is evaluated, when the cards and parameter information existing on the screen in the situation are set as feature vectors q, in this embodiment, an approximate nearest neighbor search (ANN) of the feature vectors q of the current situation is performed with a database of the number of actions of the combination of the same characters as the situation and included in the range from (the same number of actions as the situation + N) to (the same number of actions as the situation - N) as the object, to obtain a set of similar situation data. The ratio of wins and losses in the set is set as the probability of wins and losses in the situation as the inference object. Thus, the evaluation result can be calculated based on the basis that "M out of N people can win from this type of situation".
[0134] “Implementation Methods”
[0135] This embodiment is obviously "predicting the outcome of the current game based on the outcome of past games similar to the current game". Specifically, Figure 6 A system that encodes the game board status as shown in the figure into a multidimensional vector obtained by normalizing the values of each dimension with 0 to 1 and quickly searches for past board situations similar to the multidimensional vector. Figure 6 In FIG. 1 , the character selected by each player, the values of various parameters, the number of actions, the cards C1 and C3 held by each player, and the card C2 placed on the field are shown.
[0136] exist Figure 7 The architecture for implementing the system is shown in . In this embodiment, there is no need to use complex machine learning or deep learning to predict the outcome, but only the similarity of the board surface is used to predict the outcome, so a vector database can be constructed in real time based on the log data accumulated in normal game operations.
[0137] The core function of this embodiment is the cooperation between module M1 and module M2. [M2] Slicer will gradually change the board in a PvP battle (refer to Figure 6 ) are extracted as independent instantaneous playback information, and [M1] the encoder (Encoder) encodes the information disclosed to both players, such as the card information of the board, the number of cards in hand, the number of discarded cards, the player's stamina, various points (the points consumed to put cards on the field, the points consumed to evolve cards, etc.), the number of rounds, and the captain's role information, as data representing the status of the game, into a multidimensional vector obtained by normalizing the value of each dimension with 0 to 1. In this way, the similarity between certain specific board surfaces can be quantitatively measured as the similarity between vectors.
[0138] "Data Structure"
[0139] The basic data structure of this embodiment is both the log data generated in the game system and the vector data stored in the database for approximate nearest neighbor search. Figure 8 As shown, log data can be defined as a tree-structured text (e.g., a data format called JSON) representing the state on the board and a column of actions performed by the player in the board state. That is, log data is a sequence of pairs of an initial state and an initial action, then a state as a result of the action and the next action, and finally ends with a final state that determines the winner. An example of the sequence of log data is shown in Formula 1.
[0140] [Number 1]
[0141] Replaylog n :=[State0, Action o , State1, Action1..., Win_Lose] ···Formula (1)
[0143] Here, State i Indicates the i-th game state, Action i State represents the i-th action executed, and Win_Lose represents the result of the game (win or lose, draw, invalid game status, etc.). iThe simplest is the set of cards on the board and in the deck, which can be defined as in formula 2.
[0144] [Number 2]
[0145]
[0146] Here, card sp1 0 is the 0th card on the side of player 1 (first to attack) that is put on the field. sp2 0 is the 0th card placed on the field on the side of player 2 (back attack). dp1 0 is the 0th card that entered the hand of player 1 (who attacks first). dp2 0 is the 0th card that enters the hand of player 2 (who strikes first).
[0147] Next, the database for approximate nearest neighbor search can be defined as in Equation 3.
[0148] [Number 3]
[0149] DB:=[db1, db2, db3,...dbx]...Equation (3)
[0150] Here, each element db1, db2, db3, ..., dbx represents a sub-database generated for each combination of the first-hand occupation, the second-hand occupation, and the number of actions counted from the first hand. For example, a sub-database is constructed in units such as "first-hand ○ ○·second-hand ××·8th action", "first-hand ○ ○·second-hand ΔΔ·1st action". That is, in the case of a game with 8 types of characters and the battle ends after an average of 30 actions, 8×8×30=720 sub-databases are constructed. Each sub-database is independent, and the update operation on a specific sub-database will not have any effect on other sub-databases.
[0151] "Original function"
[0152] [M1] The encoder is a module that converts the game board at a specific moment into a vector of about 4000 dimensions. For example, in the case of a battle-type card game, the module encodes information disclosed to both players, such as the card information on the board, the number of cards in hand, the number of discarded cards, the player's stamina, various points (points consumed to put cards on the field, points consumed to evolve cards, etc.), the number of rounds, and the captain's role information, into a multidimensional vector obtained by standardizing the value of each dimension with 0 to 1. At this time, in order to distinguish between the first and second players for encoding, when the first and second players respectively put a card with ID: 1000030 on the field, it can be represented as 1000030first is 0.33, and 1000030second is 0.33. Fig. 9An example of this vector is shown in Fig. 9 The data shown is a multidimensional vector generated by the [M1] encoder. In the figure, as an example, vectors representing the cards of the first and second players are shown. The information of the first and second players can be identified by the text "F" corresponding to the first player and the text "S" corresponding to the second player.
[0153] In addition, as an extension of the [M1] encoder, interchangeable cards can also be associated with each other. When a specific card exists on the disk, a value is also set for the interchangeable card, so that a new card can be used in the query to retrieve past disks where the card did not exist.
[0154] For example, when a 4000-dimensional vector corresponding to 1000 existing cards is used in the system, when two new cards that are not included in these dimensions are added, the new cards can be added to the 4000-dimensional vector by using a transformation matrix that defines the relationship between these two cards and the existing 1000 cards. Specifically, in the case where the new card A is a card that simultaneously performs the functions of ID: 1000030 and ID: 1000050, the new card A can be mapped to the existing vector using the relationship information as shown in Table 1 that makes the dimensions of ID: 1000030 and ID: 1000050 have positive values such as 1.0 and 0.6. In addition, in the case where the new card B is a subspecies of ID: 1000040, only ID: 1000040 is assigned 1.0 as shown in Table 1. In the case of searching the past board from the board where the new card exists, by mapping the new card to the past card using this matrix (relationship information), a similar situation can be simulated.
[0155] [Table 1]
[0156] 1000030 1000040 1000050 New Card A 1.0 0 0.6 0 New Card B 0 1.0 0 0
[0157] That is, in situations such as "using a multidimensional vector to represent the game state in game play", "reference state information and object state information include a multidimensional vector", and "the multidimensional vector includes a vector associated with an object used in the game", when the multidimensional vector included in the object state information includes a vector associated with a new object (the above-mentioned "new card"), the retrieval unit 13 described in the above-mentioned embodiment can use prescribed transformation information (such as the relationship information in Table 1) to transform the vector associated with the new object in the multidimensional vector into a vector associated with other objects different from the new object, and use the transformed multidimensional vector to extract reference state information similar to the object state information.
[0158] [M2] Slicer is a module that transforms log data into a collection of information about a certain instantaneous board surface. Generally speaking, log data is implemented as a historical record of the actions taken by players, the game system as the manager of the game, and NPCs (also collectively referred to as participants) and the random numbers used at this time, and does not save "field information". This is because if the actions of the participants can be fully reproduced, the same field can be reproduced. However, log data in this state as a column of the participants' actions may not be suitable for machine learning. Therefore, in this embodiment, the log data is actually reproduced on the game system, and the information of each instantaneous board surface obtained therein is extracted. The board information extracted here is information inherent to the game system, for example, in a game implemented by the Unity game engine, it is expressed as an object of the C# language.
[0159] [M3]ANN Search Engine is a module that retrieves vectors similar to a specific board from a huge number of board vectors (reference data). In this embodiment, a very large number of vectors are generated. Specifically, assuming that a battle ends in an average of about 8 rounds, and 5 state changes are generated through the first-hand action and 5 actions of the same number are generated through the second-hand action in 1 round, 80 vectors are generated from a battle through the [M1] slicer. In addition, when 1 million battles are carried out in 1 day, 80 million vectors will be generated in 1 day. As a technology for quickly retrieving vectors similar to a specific vector from such a large number of vectors, in this embodiment, the approximate nearest neighbor search technology is used. As shown in Formula 4, when a vector q representing a specific situation and N sub-databases with the same first-hand occupation, second-hand occupation, and number of actions as q are given, and the difference is only a specified number of times from q, the ANN obtains k vectors similar to q from the N sub-databases.
[0160] [Number 4]
[0161] ann(q,dδ1,db2,…dbn)→[v1,v2,v3,…,v k ]…Formula (4)
[0162] By referring to the win / loss results of the log data corresponding to the k vectors, the proportion of the first-hand win in k is output as the result of the situation evaluation.
[0163] As described above, this embodiment can realize a situation evaluation that anyone can accept, such as “M out of N people can win in this type of situation”, based on a huge amount of log data.
[0164] "Effect"
[0165] The greatest advantage of this embodiment is that it realizes a situation evaluation that anyone can accept, such as "M out of N people can win from this type of situation." In addition to such a situation evaluation based on evidence, the present invention also has the following advantages.
[0166] High speed: Compared with the conventional strict nearest neighbor search, the approximate nearest neighbor search used in this embodiment operates at a very high speed. Specifically, similar vectors can be searched from billions of vectors with a waiting time of several milliseconds, and the situation evaluation can be performed following the real-time e-sports broadcast.
[0167] Incremental update: The approximate nearest neighbor search used in this embodiment is a search engine, so unlike machine learning, the addition of data will not have a significant negative impact on the inference results, such as deterioration of accuracy, and can be safely updated incrementally. Based on this feature, the situation assessment AI of this embodiment can easily incorporate the latest popular tactics as an update to the database used for approximate nearest search, assuming that meta information such as the best action always changes with the game operation and the circulation of strategy information.
[0168] Universality: The learning method of this embodiment can be widely applied to turn-based battles, and can expand AI that imitates human gaming tendencies to various themes.
[0169] "Variation"
[0170] As an application of this embodiment, by searching for the next action to be taken according to a certain situation, it is possible to make suggestions such as "An expert (advanced player) would use this move next" in real time in the game. For example, as an obstacle for novices, a new form of obstacle such as "You can accept assistance from AI up to N times" can be implemented. In this case, the player can accept the above suggestions up to N times.
[0171] The embodiments of the present invention have been described above with reference to the accompanying drawings, but these are illustrative of the present invention, and various structures other than the above can also be adopted. The structures of the above embodiments can also be combined with each other, or a part of the structure can be replaced with other structures. In addition, the structures of the above embodiments can also be subjected to various changes within the scope of not departing from the main purpose. In addition, the structures and processes disclosed in the above embodiments and modified examples can also be combined with each other.
[0172] In addition, in the multiple flow charts used in the above description, multiple processes (processes) are recorded in sequence, but the execution order of the processes executed in each embodiment is not limited to the order of the records. In each embodiment, the order of the processes shown in the diagram can be changed within the scope that does not hinder the content. In addition, the above-mentioned embodiments can be combined within the scope that the contents do not contradict each other.
[0173] A part or all of the above-mentioned embodiments can also be described as the following supplementary notes, but are not limited to the following supplementary notes.
[0174] 1. A situation assessment device, comprising:
[0175] a reference data storage unit storing reference data, the reference data being data generated based on log data of a plurality of game plays executed in the past, and the reference data including a plurality of combinations of reference state information indicating a game state in the game play and a game result determined after the game state;
[0176] an acquisition unit that acquires object state information, the object state information indicating a game state of a game play that is an evaluation object at an evaluation time point;
[0177] a retrieval unit configured to extract the reference state information similar to the object state information from the reference data;
[0178] an evaluation unit that evaluates the game situation of the game play as the evaluation object at the evaluation time point based on the game result associated with the extracted reference state information; and
[0179] An output unit outputs the evaluation result.
[0180] 2. A situation assessment device according to 1, wherein:
[0181] The evaluation unit measures the game result associated with the extracted reference state information,
[0182] The output unit outputs the measurement result as the evaluation result.
[0183] 3. A situation assessment device according to 1 or 2, wherein:
[0184] The search unit extracts the reference state information similar to the object state information from the reference data by a nearest neighbor search or an approximate nearest neighbor search.
[0185] 4. A situation assessment device according to any one of 1 to 3, wherein:
[0186] The reference data is grouped based on characteristics of the game play,
[0187] The reference data storage unit stores the reference data for each of the groups.
[0188] The search unit determines the group to which the evaluation target game belongs based on the characteristics of the evaluation target game, and extracts the reference state information similar to the target state information from the reference data belonging to the determined group.
[0189] 5. A situation assessment device according to claim 4, wherein:
[0190] The game is played in which the first player and the second player take actions alternately.
[0191] The reference data is grouped based on the number of actions counted from the first hand,
[0192] The acquisition unit further acquires the number of actions of the game play as the evaluation object at the evaluation time point,
[0193] The search unit extracts the reference state information similar to the object state information from the reference data belonging to a group corresponding to the number of actions at the evaluation time point.
[0194] 6. A situation assessment device according to 5, wherein:
[0195] The search unit extracts the reference state information similar to the object state information from the reference data belonging to the group corresponding to the number of actions included in the range from (the number of actions at the evaluation time point-α) to (the number of actions at the evaluation time point+α).
[0196] 7. A situation assessment device according to any one of 4 to 6, wherein:
[0197] The game is played by the first player and the second player respectively selecting a role to play.
[0198] The reference data is grouped based on the combination of the character selected by the first player and the character selected by the second player.
[0199] The acquisition unit further acquires information indicating a character selected by a first player and a character selected by a second player in the game play to be evaluated,
[0200] The search unit extracts the reference state information similar to the target state information from the reference data belonging to a group corresponding to a combination of a character selected by a first player and a character selected by a second player in the game play to be evaluated.
[0201] 8. A situation assessment device according to any one of 4 to 7, wherein:
[0202] The reference data is grouped based on a combination of the characteristics of the first player and the characteristics of the second player.
[0203] The acquisition unit further acquires the characteristics of the first player and the second player in the game play to be evaluated.
[0204] The search unit extracts the reference state information similar to the target state information from the reference data belonging to a group corresponding to a combination of characteristics of a first player and characteristics of a second player in the game play to be evaluated.
[0205] 9. A situation assessment device according to any one of 1 to 8, wherein:
[0206] There is also a reference data production unit, which extracts the log data of game plays performed in the past by the players whose player characteristics meet specified conditions from the log data of multiple game plays performed by multiple players in the past, and produces the reference data based on the extracted log data.
[0207] 10. A situation assessment device according to any one of 1 to 9, wherein:
[0208] Use a multi-dimensional vector to represent the game state in the game play.
[0209] The reference state information and the object state information include the multi-dimensional vector,
[0210] The multidimensional vector includes a vector associated with an object used in the game.
[0211] In a case where the multidimensional vector included in the object state information includes a vector associated with the new object, the retrieval unit uses specified transformation information to transform the vector associated with the new object in the multidimensional vector into a vector associated with other objects different from the new object, and uses the transformed multidimensional vector to extract the reference state information similar to the object state information.
[0212] 11. A situation assessment method, wherein:
[0213] The computer stores reference data, the reference data being data generated based on log data of a plurality of game plays executed in the past, the reference data including a plurality of combinations of reference state information representing a game state in the game play and a game result determined after the game state,
[0214] In the situation assessment method, the following steps are performed by a computer:
[0215] An acquisition step of acquiring object state information, wherein the object state information indicates a game state of a game play as an evaluation object at an evaluation time point;
[0216] A retrieval step of extracting the reference state information similar to the object state information from the reference data;
[0217] an evaluation step of evaluating a game situation of the game play as the evaluation object at the evaluation time point based on the game result associated with the extracted reference state information; and
[0218] The output step outputs the evaluation result.
[0219] 12. A program that causes a computer to function as:
[0220] A reference data storage unit storing reference data, the reference data being data generated based on log data of a plurality of game plays executed in the past, the reference data including a plurality of combinations of reference state information indicating a game state in the game play and a game result determined after the game state;
[0221] an acquisition unit, which acquires object state information, wherein the object state information represents a game state of a game play as an evaluation object at an evaluation time point;
[0222] a retrieval unit, which extracts the reference state information similar to the object state information from the reference data;
[0223] an evaluation unit that evaluates the game situation of the game play as the evaluation object at the evaluation time point based on the game result associated with the extracted reference state information; and
[0224] An output unit outputs the evaluation result.
[0225] This application claims the priority based on Japanese patent application No. 2022-065152 filed on April 11, 2022, the disclosure of which is incorporated herein in its entirety.
[0226] Description of Reference Numerals
[0227] 10: Situation evaluation device; 11: Reference data storage unit; 12: Acquisition unit; 13: Retrieval unit; 14: Evaluation unit; 15: Output unit; 16: Log data storage unit; 17: Reference data creation unit; 1A: Processor; 2A: Memory; 3A: Input / output I / F; 4A: Peripheral circuit; 5A: Bus.
Claims
1. A situation assessment device, comprising: a reference data storage unit storing reference data, the reference data being data generated based on log data of a plurality of game plays executed in the past, and the reference data including a plurality of combinations of reference state information indicating a game state in the game play and a game result determined after the game state; an acquisition unit that acquires object state information, the object state information indicating a game state of a game play that is an evaluation object at an evaluation time point; a retrieval unit configured to extract the reference state information similar to the object state information from the reference data; an evaluation unit that evaluates the game situation of the game play as the evaluation object at the evaluation time point based on the game result associated with the extracted reference state information; as well as An output unit outputs the evaluation result.
2. The situation assessment device according to claim 1, wherein: The evaluation unit measures the game result associated with the extracted reference state information, The output unit outputs the measurement result as the evaluation result.
3. The situation assessment device according to claim 1 or 2, wherein: The search unit extracts the reference state information similar to the object state information from the reference data by a nearest neighbor search or an approximate nearest neighbor search.
4. The situation assessment device according to claim 1 or 2, wherein: The reference data is grouped based on characteristics of the game play, The reference data storage unit stores the reference data for each of the groups. The search unit determines the group to which the evaluation target game belongs based on the characteristics of the evaluation target game, and extracts the reference state information similar to the target state information from the reference data belonging to the determined group.
5. The situation assessment device according to claim 3, wherein: The reference data is grouped based on characteristics of the game play, The reference data storage unit stores the reference data for each of the groups. The search unit determines the group to which the evaluation target game belongs based on the characteristics of the evaluation target game, and extracts the reference state information similar to the target state information from the reference data belonging to the determined group.
6. The situation assessment device according to claim 4, wherein: The game is played in which the first player and the second player take actions alternately. The reference data is grouped based on the number of actions counted from the first hand, The acquisition unit further acquires the number of actions of the game play as the evaluation object at the evaluation time point, The search unit extracts the reference state information similar to the object state information from the reference data belonging to a group corresponding to the number of actions at the evaluation time point.
7. The situation assessment device according to claim 5, wherein: The game is played in which the first player and the second player take actions alternately. The reference data is grouped based on the number of actions counted from the first hand, The acquisition unit further acquires the number of actions of the game play as the evaluation object at the evaluation time point, The search unit extracts the reference state information similar to the object state information from the reference data belonging to a group corresponding to the number of actions at the evaluation time point.
8. The situation assessment device according to claim 6, wherein: The search unit extracts the reference state information similar to the object state information from the reference data belonging to the group corresponding to the number of actions included in the range from (the number of actions at the evaluation time point-α) to (the number of actions at the evaluation time point+α).
9. The situation assessment device according to claim 7, wherein: The search unit extracts the reference state information similar to the object state information from the reference data belonging to the group corresponding to the number of actions included in the range from (the number of actions at the evaluation time point-α) to (the number of actions at the evaluation time point+α).
10. The situation assessment device according to claim 4, wherein: The game is played by the first player and the second player respectively selecting a role to play. The reference data is grouped based on the combination of the character selected by the first player and the character selected by the second player. The acquisition unit further acquires information indicating a character selected by a first player and a character selected by a second player in the game play to be evaluated, The search unit extracts the reference state information similar to the target state information from the reference data belonging to a group corresponding to a combination of a character selected by a first player and a character selected by a second player in the game play to be evaluated.
11. The situation assessment device according to claim 4, wherein: The reference data is grouped based on a combination of the characteristics of the first player and the characteristics of the second player. The acquisition unit further acquires the characteristics of the first player and the second player in the game play to be evaluated. The search unit extracts the reference state information similar to the target state information from the reference data belonging to a group corresponding to a combination of characteristics of a first player and characteristics of a second player in the game play to be evaluated.
12. The situation assessment device according to claim 1 or 2, wherein: There is also a reference data production unit, which extracts the log data of game plays performed in the past by the players whose player characteristics meet specified conditions from the log data of multiple game plays performed by multiple players in the past, and produces the reference data based on the extracted log data.
13. The situation assessment device according to claim 1 or 2, wherein: Use a multi-dimensional vector to represent the game state in the game play. The reference state information and the object state information include the multi-dimensional vector, The multidimensional vector includes a vector associated with an object used in the game. In a case where the multidimensional vector included in the object state information includes a vector associated with the new object, the retrieval unit uses specified transformation information to transform the vector associated with the new object in the multidimensional vector into a vector associated with other objects different from the new object, and uses the transformed multidimensional vector to extract the reference state information similar to the object state information.
14. A situation assessment method, wherein: The computer stores reference data, the reference data being data generated based on log data of a plurality of game plays executed in the past, the reference data including a plurality of combinations of reference state information representing a game state in the game play and a game result determined after the game state, In the situation assessment method, the following steps are performed by a computer: An acquisition step of acquiring object state information, wherein the object state information indicates a game state of a game play as an evaluation object at an evaluation time point; A retrieval step of extracting the reference state information similar to the object state information from the reference data; An evaluation step of evaluating a game situation of the game play as the evaluation object at the evaluation time point based on the game result associated with the extracted reference state information; as well as The output step outputs the evaluation result.
15. A program that causes a computer to function as: A reference data storage unit storing reference data, the reference data being data generated based on log data of a plurality of game plays executed in the past, the reference data including a plurality of combinations of reference state information indicating a game state in the game play and a game result determined after the game state; an acquisition unit, which acquires object state information, wherein the object state information represents a game state of a game play as an evaluation object at an evaluation time point; a retrieval unit, which extracts the reference state information similar to the object state information from the reference data; an evaluation unit that evaluates the game situation of the game play as the evaluation object at the evaluation time point based on the game result associated with the extracted reference state information; as well as An output unit outputs the evaluation result.
