Information processing system, information processing device, and program

Abstract

The present invention suppresses trouble that can develop from communication between users. The present invention is an information processing system that allows users that take part in a game to interact within the game by posting posted comments to each other. The information processing system has a game control unit that determines game results for each user that takes part in the game, an evaluation unit that evaluates whether posted comments by the users for whom the game results were obtained are malicious using a machine learning model that generates evaluation results for a posted comment as output data in response to input data that includes game results and the posted comment, and a trouble countermeasure unit that designates users that posted posted comments that have been evaluated as malicious as trouble countermeasure users.

Description

Information Processing System, Information Processing Apparatus, and Program

[0001] The present invention relates to an information processing system, an information processing apparatus, and a program.

[0002] In a game system that monitors whether words included in a posted comment posted by a user to a friend using chat match a preset prohibited word, when a user satisfies a release condition by posting a comment using a large number of prohibited words to a friend, it is known that the friendship is released (for example, Patent Document 1).

[0003] Japanese Patent No. 7232955

[0004] When a plurality of users participating in a game communicate by posting comments to each other, depending on the game situation in which the user who is the source of the post or the user who is the destination of the post is placed, the meaning conveyed to the other party from the content of the post may change.

[0005] For example, when posting a comment from the losing side to the winning side of a game, even if it is conveyed as a word of praise for the other party, if the same comment is posted from the winning side to the losing side of the game, it may be conveyed as a word that insults the other party or incites the other party.

[0006] Therefore, it is difficult to monitor the use of malicious posted comments including slander only by using preset prohibited words, and if such unpleasant expressions are used in the posted comments, there is a risk of developing into trouble in the communication between users.

[0007] The present invention has been made in view of such circumstances, and an object thereof is to suppress troubles that develop from communication between users.

[0008] The main invention of the present invention for solving the above problems is an information processing system configured to enable multiple users participating in a game to interact with each other by posting comments, comprising: a game control unit that determines the game results for each user participating in the game; an evaluation unit that evaluates whether a user's posted comment is malicious or not, using a machine learning model that generates an evaluation result of the posted comment as output data for input data including the game result and the posted comment; and a trouble prevention unit that, if a posted comment is evaluated as malicious, sets the user who posted the comment as a trouble prevention user. Other features of the present invention will be made clear by the description in this specification and the accompanying drawings.

[0009] This is a configuration diagram showing an example of the information processing system 1 according to this embodiment. This is a hardware configuration diagram showing an example of the computer 50 according to this embodiment. This is a functional block diagram showing an example of the server device 10 according to this embodiment. This is a configuration diagram showing an example of the training data according to this embodiment. This is a configuration diagram showing an example of the character information according to this embodiment. This is a configuration diagram showing an example of the user information according to this embodiment. This is a configuration diagram showing an example of the game play information according to this embodiment. This is a functional block diagram showing an example of the client terminal 20 according to this embodiment. This is a flowchart showing an example of the operation related to monitoring posted comments in this embodiment. This is a diagram showing an example of the configuration of the game screen 500 according to this embodiment. This is a flowchart showing an example of the operation related to the execution of troubleshooting processing to restrict game play by users in this embodiment. This is a flowchart showing an example of the operation related to the execution of troubleshooting processing to restrict the posting of comments by users in this embodiment.

[0010] The following matters will become clear from the description in this specification and the attached drawings: an information processing system configured to allow multiple users participating in a game to interact with each other by posting comments, comprising: a game control unit that determines the game results for each user participating in the game; an evaluation unit that uses a machine learning model to generate an evaluation result of the posted comments as output data for input data including game results and posted comments, to evaluate whether a user's posted comment is malicious or not; and a trouble prevention unit that, if a posted comment is evaluated as malicious, sets the user who posted the comment as a trouble prevention user. With such an information processing system, by using a machine learning model, it becomes possible to evaluate whether a user's posted comment is malicious or not according to the game situation in which the posting user or the receiving user is, and since users whose posted comments are evaluated as malicious are set as trouble prevention users, it becomes possible to suppress troubles that develop from communication between users.

[0011] Furthermore, in such an information processing system, the game control unit may determine whether each user wins or loses as a game result in a competitive game in which multiple users participate, and the evaluation unit may use a machine learning model that generates an evaluation result of the posted comments as output data for input data including the game result of win or loss and the posted comments to evaluate whether the posted comments of a user who achieved a win or loss are malicious. With such an information processing system, by using a machine learning model, it becomes possible to evaluate whether a user's posted comments are malicious or not according to the game situation of the user who achieved a win or loss, and users whose posted comments are evaluated as malicious will be set as trouble prevention users, thereby making it possible to suppress troubles that develop from communication between users participating in competitive games.

[0012] Furthermore, in such an information processing system, the trouble prevention unit may execute a process to restrict gameplay for users designated as trouble prevention users. With such an information processing system, since gameplay is restricted for users designated as trouble prevention users, it becomes possible to suppress troubles that may arise from communication between users.

[0013] Furthermore, in such an information processing system, the trouble prevention unit may execute a process to restrict the movement of the character used in the game for the user designated as a trouble prevention user. With such an information processing system, the movement of the character used in the game is restricted for the user designated as a trouble prevention user, making it possible to suppress troubles that may arise from communication between users.

[0014] Furthermore, in such an information processing system, the trouble prevention unit may execute a process to restrict the posting of comments by the user designated as a trouble prevention user. With such an information processing system, since the user designated as a trouble prevention user is restricted from posting comments, it is possible to suppress troubles that may arise from communication between users.

[0015] Furthermore, in such an information processing system, the trouble countermeasures unit may also perform a process to delete posted comments made by a user designated as a trouble countermeasures user. With such an information processing system, since the posted comments of a user designated as a trouble countermeasures user are deleted, it becomes possible to suppress troubles that may arise from communication between users.

[0016] Furthermore, in such an information processing system, the trouble countermeasures unit may also progressively increase the trouble level of a user designated as a trouble countermeasures user each time that user's posted comment is evaluated as malicious. With such an information processing system, since the trouble level of a user designated as a trouble countermeasures user will progressively increase each time that user's posted comment is evaluated as malicious, it becomes possible to suppress troubles that develop from communication between users.

[0017] Furthermore, the information processing device is configured to allow multiple users participating in the game to interact with each other by posting comments, and comprises: a game control unit that determines the game results for each user participating in the game; an evaluation unit that uses a machine learning model to generate an evaluation result of the posted comments as output data for input data including game results and posted comments, to evaluate whether the posted comments of users who have obtained game results are malicious or not; and a trouble prevention unit that, if a posted comment is evaluated as malicious, sets the user who posted the comment as a trouble prevention user. Such an information processing device makes it possible to suppress troubles that develop from communication between users.

[0018] Furthermore, the program enables an information processing device configured to allow multiple users participating in the game to interact with each other by posting comments, to function as a game control means for determining the game results of each user participating in the game; an evaluation means for evaluating whether a user's posted comment is malicious or not, using a machine learning model that generates an evaluation result of the posted comment as output data for input data including the game results and posted comments; and a trouble prevention means for setting the user who posted the comment as a trouble prevention user if the comment is evaluated as malicious. Such a program makes it possible to suppress troubles that develop from communication between users.

[0019] ===Embodiments=== The following describes in detail an information processing system, program, information processing device, and information processing method according to an embodiment of the present invention. The present invention can be broadly applied to information processing systems, programs, information processing devices, and information processing methods, etc., for multiple users participating in a game to communicate with each other by posting comments. In this embodiment, a game that uses a character as an example of content as digital content within the game will be specifically described.

[0020] <<<System Configuration>>> Figure 1 is a configuration diagram showing an example of the information processing system 1 according to this embodiment. As shown in Figure 1, in the information processing system 1 according to this embodiment, a server device 10 and one or more client terminals 20 are connected via a network N.

[0021] The server device 10 is an information processing device that provides various services to users participating in the game using client terminals 20. Client terminals 20 are information processing devices such as smartphones, tablets, PCs, and dedicated gaming devices for home or business use used by the user. The network N is the internet, and includes mobile wireless base stations, etc.

[0022] It goes without saying that the information processing system 1 shown in Figure 1 is just one example, and there are various system configurations depending on the application and purpose. For example, the server device 10 may be configured to distribute some of its functions across multiple computers. Also, for example, the server device 10 may perform some of the functions of the client terminal 20, or the client terminal 20 may perform some of the functions of the server device 10.

[0023] <<<Hardware Configuration>>> <Server Device and Client Terminal> Figure 2 is a hardware configuration diagram showing an example of a computer 50 according to this embodiment. The server device 10 and client terminal 20 according to this embodiment are realized by a computer 50 with the hardware configuration shown in Figure 2, for example. Note that the computer 50 is an example of an information processing device.

[0024] As shown in Figure 2, the computer 50 includes a CPU 51, RAM 52, ROM 53, communication interface 54, input device 55, display device 56, external interface 57, and HDD 58, and each is interconnected via bus line B. In the server device 10, the input device 55 and display device 56 may be connected and used only when necessary.

[0025] The CPU 51 is an arithmetic unit that controls and implements the functions of the entire computer by reading programs and data from storage devices such as ROM 53 and HDD 58 onto RAM 52 and executing various processes based on the read programs and data.

[0026] RAM 52 is an example of a volatile semiconductor memory (storage device) for temporarily holding programs and data, and is also used as a work area when the CPU 51 performs various processes.

[0027] ROM 53 is an example of a non-volatile semiconductor memory (storage device) that can retain programs and data even when the power is turned off. ROM 53 stores programs and data such as the BIOS, OS settings, and network settings that are executed when the computer 50 starts up.

[0028] The communication interface 54 is an interface for connecting the computer 50 to the network N. This allows the computer 50 to perform data communication via the communication interface 54.

[0029] The input device 55 is a device used by the user or game operator to input various data. In this embodiment, the input device 55 is, for example, an operating device such as a touch panel, operation keys or buttons, a keyboard or mouse.

[0030] The display device 56 is a device for displaying various information on a screen to the user or game operator. In this embodiment, the display device 56 is, for example, a display such as a liquid crystal or organic EL.

[0031] The external interface 57 is an interface for connecting to an external device in a data communication manner. This allows the computer 50 to read and / or write to a recording medium via the external interface 57. The external device is, for example, a recording medium such as a flexible disk, CD, DVD, SD memory card, or USB memory.

[0032] The HDD58 is an example of a non-volatile storage device that stores programs and data. The programs and data stored include the operating system (OS), which is the basic software that controls the entire computer, and applications that provide various functions on the OS.

[0033] Alternatively, instead of the HDD58, a drive device that uses flash memory as a storage medium (for example, a solid-state drive: SSD) may be used.

[0034] The server device 10 and client terminal 20 according to this embodiment can perform various processes as described later by executing a program on the computer 50 with the hardware configuration described above.

[0035] <<<Software Configuration>>> <Server Device> Figure 3 is a functional block diagram showing an example of a server device 10 according to this embodiment. The server device 10 according to this embodiment is realized by, for example, the functional blocks shown in Figure 3.

[0036] The server device 10 according to this embodiment realizes a server control unit 100, a server storage unit 120, and a server communication unit 140 by executing a program.

[0037] The server control unit 100 has the function of executing various processes in the server device 10. The server control unit 100 according to this embodiment includes at least a game control unit 101, a posted comment control unit 102, a troubleshooting unit 103, and a learning unit 104.

[0038] The game control unit 101 performs control related to various games. In this embodiment, the game control unit 101 matches two or more users from among multiple users participating in the game, and performs control related to a competitive game in which each user becomes an opponent of the other through matching and competes for victory or defeat.

[0039] Furthermore, it is possible to control not only competitive games like this, but also cooperative games where users work together through matchmaking to achieve a common objective.

[0040] The comment posting control unit 102 controls the exchange of comments posted between users participating in the game. The comment posting control unit 102 in this embodiment has an acquisition unit 102A and an evaluation unit 102B.

[0041] The acquisition unit 102A acquires posted comments submitted by users using the client terminal 20. In this embodiment, the acquisition unit 102A acquires the user's posted comments along with the game results obtained by the user who played the game.

[0042] The evaluation unit 102B evaluates whether the comments posted by users who have obtained game results are malicious or not. In this embodiment, the evaluation unit 102B evaluates whether the comments posted by users who have obtained game results are malicious or not by using a machine learning model that generates an evaluation result of posted comments as output data for input data including game results and posted comments.

[0043] The trouble prevention unit 103 executes trouble prevention processing for users who require measures to prevent potential problems between users. In this embodiment, if the trouble prevention unit 103 is evaluated by the evaluation unit 102B as having malicious posting comments, it sets the user who posted the comment as a trouble prevention user.

[0044] As for the trouble-shooting process for a user set as a trouble-shooting user, for example, a process of restricting game play, a process of restricting the movement of characters used in the game, a process of restricting the posting of posted comments, a process of deleting already posted posted comments, etc. are executed.

[0045] The learning unit 104 performs learning of the machine learning model so as to generate an evaluation result of the posted comment as output data for the input data including the game result and the posted comment. The learning unit 104 according to the present embodiment adjusts the parameters of the learned machine learning model constructed by pre-learning using a large-scale data set by fine-tuning using additional training data prepared in advance by the game operator. By adjusting the parameters according to the training data by this additional learning, the estimation accuracy of the evaluation result of the posted comment generated by the machine learning model can be improved.

[0046] The server storage unit 120 has a function of storing various information in the server device 10. The server storage unit 120 in the present embodiment includes, as shown in FIG. 3, a machine learning model storage unit 121, a training data storage unit 122, a character information storage unit 123, a user information storage unit 124, and a game play information storage unit 125.

[0047] The machine learning model storage unit 121 stores the program and parameters of the machine learning model. The machine learning model stored in the machine learning model storage unit 121 is a learned model whose parameters are adjusted so as to output an evaluation result of the posted comment when the game result and the posted comment of the user are input.

[0048] The machine learning model has a neural network structure, and has an input layer that receives input data, one or more hidden layers that perform arithmetic processing on the input data received in the input layer, and an output layer that generates output data based on the arithmetic result of the hidden layer. The machine learning model has been learned using the training data stored in the training data storage unit 122.

[0049] The training data storage unit 122 stores the training data used to train the machine learning model. The training data stored in the training data storage unit 122 consists of pairs of input and output data used to build the machine learning model, and is composed of the following types of data.

[0050] Figure 4 is a diagram showing an example of training data according to this embodiment. The training data shown in Figure 4 includes items such as training data ID, input data, and output data. In this embodiment, for example, the input data of the training data is the game result and posted comments (input text), and the output data of the training data is the evaluation result of the posted comments (output text).

[0051] Furthermore, it is possible to pre-vectorize the input data of the training data stored in the training data storage unit 122 using a machine learning model, and store the N-dimensional vector values ​​converted from text in association with that input data. This makes it possible to use the vectorized data when comparing them during the calculation of cosine similarity, as described later.

[0052] The character information storage unit 123 stores character information about a character as an example of content. The character information stored by the character information storage unit 123 consists of the following types of data.

[0053] Figure 5 is a diagram showing an example of character information according to this embodiment. The character information shown in Figure 5 includes items such as character ID, character name, rarity, level, and abilities.

[0054] The character ID is information used to uniquely identify a character. The character name is information indicating the character's name. Rarity is an example of an attribute, indicating the character's rarity. Here, characters are divided into five ranks, from rarity 1 to 5. However, the number of rarity ranks is not limited to five; for example, there may be two ranks, such as "high rarity" and "low rarity," or any other number of ranks.

[0055] Level is an example of a parameter, representing information that indicates the degree of growth a character undergoes through experience. Here, each character has an initial level value, and the system is designed to increase gradually. For example, the system is designed so that completing the game results in a larger level increase.

[0056] Abilities are just one example of parameters, representing information that indicates a character's strength. Here, each character has initial values ​​set for ability parameters, including attack, defense, and HP, and these are configured to increase gradually. For example, ability parameters increase as the character levels up.

[0057] The user information storage unit 124 stores user information about users participating in the game. The user information stored by the user information storage unit 124 consists of the following types of data.

[0058] Figure 6 is a diagram showing an example of user information according to this embodiment. The user information shown in Figure 6 includes items such as user ID, username, owned characters, troubleshooting, game history, and posting history.

[0059] The User ID is information used to uniquely identify a user participating in the game. The Username is information indicating the user's name. The Owned Characters section indicates the characters currently owned by the user from the characters set in the character information. The current level and abilities are also recorded for each owned character. The Troubleshooting section is flag information indicating whether or not the user is a user designated for troubleshooting.

[0060] Game history is historical information about games a user has played, and is updated each time the user plays a game. This game history includes, for example, the date and time the user played a game, the game result, and the opponent.

[0061] The posting history is historical information about the posts and comments a user has made so far, and is updated each time the user makes a post. This posting history includes, for example, the date and time the user posted a comment, the comment itself, the people the user interacted with, and their rating.

[0062] The gameplay information storage unit 125 stores gameplay information related to gameplay. The gameplay information stored by the gameplay information storage unit 125 is information that is updated according to the status of gameplay and consists of the following types of data.

[0063] Figure 7 is a diagram showing an example of gameplay information according to this embodiment. The gameplay information shown in Figure 7 includes items such as play ID, play date and time, matching result, game result, and posted comment information.

[0064] The Play ID is information used to uniquely identify gameplay. The Play Date and Time indicates the date and time the gameplay took place. The Matching Result identifies users who have been matched to play the game together. The Game Result is information about each user's game results. The Posted Comment Information is information about posted comments exchanged between users playing the game together, and is updated according to the posting status of comments between users.

[0065] Figure 8 is a diagram showing an example of post comment information according to this embodiment. The post comment information shown in Figure 8 includes items such as post ID, post date and time, poster, game result, post comment, and post rating.

[0066] The post ID is information used to uniquely identify a posted comment. The post date and time indicates the date and time the comment was posted. The poster is information that identifies the user who posted the comment. The game result is information that indicates the game result of the user who posted the comment. The post comment is information that indicates the content of the post. The post evaluation is information that indicates the evaluation result of the post comment generated by the machine learning model.

[0067] The server communication unit 140 has the function of communicating with each of the multiple client terminals 20 via the network N, which is a wired and wireless communication network.

[0068] As mentioned above, in the information processing system 1 according to this embodiment, the server device 10 performs control related to the game, but the client terminal 20 may perform some of the various control and processing. Specifically, some functions of the game control unit 101 may be configured not to be provided in the server control unit 100 of the server device 10, while they may be provided in the terminal control unit 200 of the client terminal 20.

[0069] 《Client Terminal》 Figure 9 is a functional block diagram showing an example of a client terminal 20 according to this embodiment. The client terminal 20 according to this embodiment is realized by, for example, the functional blocks shown in Figure 9.

[0070] In this embodiment, the client terminal 20 implements a terminal control unit 200, a terminal storage unit 220, a terminal communication unit 240, an operation input reception unit 250, and a screen display unit 260 by executing a program.

[0071] The terminal control unit 200 has the function of executing various processes on the client terminal 20. In this embodiment, the terminal control unit 200 includes at least a server access unit 201 and a display control unit 202.

[0072] The server access unit 201, in response to a user's game operation or comment posting operation, accesses the server device 10 via the network and requests the server device 10 to execute various processes in response to the user's game operation or comment posting operation.

[0073] Furthermore, the server access unit 201 accesses the server device 10 via the network, depending on the situation in the game, and obtains necessary information from the various types of information stored in the server storage unit 120 of the server device 10.

[0074] The display control unit 202 controls the screen display on the client terminal 20 by outputting various game screens and the like to the screen display unit 260 according to the situation in the game, or by outputting various game screens and the like to the screen display unit 260 in response to game operations or comment posting operations by the user.

[0075] The terminal storage unit 220 has the function of storing various information on the client terminal 20. In this embodiment, the terminal storage unit 220 stores various programs and various information (including information received from the server device 10, etc.) that are necessary for the client terminal 20.

[0076] The terminal communication unit 240 has the function of communicating with the server device 10 via the network N, which is a wired and wireless communication network.

[0077] The operation input receiving unit 250 receives user input when a user performs a game operation or comments posting operation on various game screens related to the game.

[0078] The screen display unit 260 displays information on the client terminal 20's screen by receiving control signals from the display control unit 202.

[0079] <<<System Overview>>> The overview of the information processing system 1 in this embodiment will be described below. In the information processing system 1 in this embodiment, users participating in the game can enjoy playing against other matched users as opponents, or playing cooperatively with other matched users as allies. Furthermore, each user who plays such competitive or cooperative games can communicate with each other by posting comments within the game.

[0080] For example, when users who have played a game against each other exchange messages, if the user who lost the game (the loser) posts comments such as "Good job" or "Nice game!" to the user who won the game (the winner), it will be conveyed to the other user as praise for the winner's play.

[0081] However, if a user who achieved a "victory" in the game (winner) posts the same comments such as "Good job" or "Nice game!" to a user who achieved a "loss" (loser), it can be interpreted by the loser as insulting or provoking them.

[0082] Thus, even when the same comment is used by different users, the meaning conveyed to the recipient may differ depending on the game situation in which the original user or the recipient user is in (i.e., the game results obtained by the original user or the recipient user).

[0083] Therefore, simply using a list of prohibited words prepared in advance by the game operators makes it difficult to monitor the use of malicious comments, including slander and defamation. If such offensive language is used in comments, it could lead to trouble in communication between users.

[0084] Therefore, in the information processing system 1 of this embodiment, the machine learning model constructed through pre-training is further trained (fine-tuned) with training data prepared in advance by the game operator, and the use of this machine learning model is used to monitor the use of malicious posted comments.

[0085] By doing this, in the example above, the machine learning model can receive user game results such as "win" or "lose" along with user comments such as "Nice game!" as input data. It can then estimate whether the comment was posted by a user who lost or won. If it is estimated to be the former, it outputs an evaluation result ("OK") indicating that it is not a malicious comment; if it is estimated to be the latter, it outputs an evaluation result ("NG") indicating that it is a malicious comment. As a result, users who achieved a "win" can be identified as users who require troubleshooting.

[0086] <<<System Operation>>> The operation of the information processing system 1 in this embodiment will be explained below. The following explanation will take the case where users who have played a competitive game against each other exchange posted comments as an example.

[0087] 《Monitoring of Posted Comments》 The operation related to monitoring posted comments in this embodiment will be explained using Figures 10 and 11. Figure 10 is a flowchart showing an example of the operation related to monitoring posted comments in this embodiment. Figure 11 is a diagram showing an example of the configuration of the game screen 500 in this embodiment.

[0088] As shown in Figure 10, the game control unit 101 of the server device 10 determines the game result for gameplay between users who have been matched as opponents (step S11).

[0089] In this embodiment, in a competitive game, a "victory" game result is determined for one user who meets the victory conditions, and a "defeat" game result is determined for the other user who does not meet the victory conditions. Victory conditions include, for example, achieving the objective faster than the opponent or scoring a higher score than the opponent.

[0090] When the game control unit 101 of the server device 10 determines the game result for a game played between matched users, it updates the game play information stored in the game play information storage unit 125 by registering the game result in association with the play ID of that game, as shown in Figure 7.

[0091] Next, the acquisition unit 102A of the server device 10 acquires posted comments from the client terminal 20 via the network, which were posted by users who obtained a "win" or "lose" in the game (step S12).

[0092] When the acquisition unit 102A of the server device 10 acquires a user's posted comment that obtained a game result in that game play, it refers to the game play information shown in Figure 7 and identifies the posted comment information associated with the play ID of that game play. Then, as shown in Figure 8, it updates the identified posted comment information by registering the necessary information, including the user's posted comment and game result, in association with the post ID assigned in the order of posting.

[0093] In this embodiment, a user who achieves a "win" or "loss" in gameplay can post a comment to their opponent when the game screen 500 shown in Figure 11 is displayed on the screen display unit 260 of the client terminal 20.

[0094] The game screen 500 in this embodiment is displayed on the screen when the game result is determined during gameplay, and as shown in Figure 11, it consists of a gameplay area 501 where a character that operates according to the user's input is displayed, a comment display area 502 where submitted comments are displayed in the order they were submitted, a comment input area 503 for entering text, and a submit button 504.

[0095] In this game screen 500, when a user who has achieved a "win" or "lose" in the game enters a comment in the comment input area 503 and then selects the submit button 504, the submitted comments will be displayed in the submitted comments display area 502 in the order they were submitted.

[0096] Returning to Figure 10, the evaluation unit 102B of the server device 10 uses a machine learning model that has been trained to output an evaluation result of a posted comment when a game result and a posted comment are input to evaluate whether or not the posted comment of the user who obtained the game result in that game play is a malicious posted comment (step S13).

[0097] Specifically, the evaluation unit 102B of the server device 10 inputs the game results and posted comments of the user who made the post during gameplay as input text to the machine learning model stored in the machine learning model storage unit 121, based on the posted comment information shown in Figure 8.

[0098] The machine learning model performs morphological analysis on the input text of the game results and posted comments, decomposing the input text string into individual words, then vectorizing each decomposed word into a vector. Based on the vectors of each word, the entire input text is then vectorized, converting the input text into a vector. Similarly, the input text of the training data stored in the training data storage unit 122 is converted into vectors. Note that methods such as Doc2Vec or Word2Vec may be used to convert the input text into vectors using machine learning.

[0099] Next, the machine learning model calculates the similarity between the input texts based on the vectors converted from the input texts of the users who posted the gameplay and the vectors converted from the input texts of the training data.

[0100] In this embodiment, cosine similarity is used. Cosine similarity can be calculated by the dot product of vectors. When the cosine similarity value is "-1" (when the angle between the vectors is 180 degrees), there is a complete mismatch; when the cosine similarity value is "0" (when the angle between the vectors is 90 degrees), there is no relationship; and when the cosine similarity value is "1" (when the angle between the vectors is 0 degrees), there is a complete match. In other words, with cosine similarity, the closer the input text is to "1" the more similar it is, and the closer it is to "-1" the less similar it is.

[0101] In this way, the machine learning model identifies the input text in the training data stored in the training data storage unit 122 that most closely resembles the input text of the user who made the post during that gameplay (i.e., the one with the cosine similarity value closest to "1"). Then, the machine learning model refers to the training data shown in Figure 4 and outputs output data corresponding to the identified input text in the training data as the evaluation result of the posted comment.

[0102] When the evaluation unit 102B of the server device 10 outputs an evaluation result of a posted comment by the machine learning model, it determines that if the evaluation result is "OK", the posted comment is not malicious (post evaluation "OK"), while if the evaluation result is "NG", it determines that the posted comment is malicious (post evaluation "NG").

[0103] Then, the evaluation unit 102B of the server device 10 refers to the game play information shown in Figure 7 to identify the posted comment information associated with the play ID of that game play, and updates the identified posted comment information by registering the post evaluation ("OK" / "NG") for the determined posted comment, associated with the post ID assigned in the order of posting, as shown in Figure 8.

[0104] Next, the troubleshooting unit 103 of the server device 10 determines whether the user who made the post during gameplay is a user who requires measures to prevent potential problems between users (step S14).

[0105] In this embodiment, the troubleshooting unit 103 determines that troubleshooting is unnecessary if the posting evaluation for the posted comment determined by the evaluation unit 102B in step S13 is "OK", while determining that troubleshooting is necessary if the posting evaluation for the posted comment is "NG".

[0106] Next, if the troubleshooting unit 103 of the server device 10 determines that troubleshooting is necessary for the user who made the post during gameplay (step S14: YES), it sets the user who made the post as a troubleshooting user (step S15).

[0107] In this embodiment, the troubleshooting unit 103 updates the user information shown in Figure 6 by registering a flag indicating that the user is a troubleshooting user, associating it with the user who made the post during gameplay.

[0108] Thus, according to the information processing system 1 of this embodiment, by utilizing a machine learning model trained to output an evaluation result of a posted comment when game results and posted comments are input, it becomes possible to evaluate whether a user's posted comment is malicious or not according to the game situation in which the posting user and the recipient user are, and users whose posted comments are evaluated as malicious will be set as trouble prevention users, thereby making it possible to suppress troubles that develop from communication between users.

[0109] 《Execution of Troubleshooting》 The operation of troubleshooting in this embodiment will be described below. As specific examples of troubleshooting, we will describe the case of executing a process that restricts user gameplay and the case of executing a process that restricts user comment posting. Note that the troubleshooting in this embodiment is not limited to these specific examples and can be broadly applied to other specific examples.

[0110] <Specific Example 1> The operation of restricting user gameplay as a troubleshooting measure in this embodiment will be explained using Figure 12. Figure 12 is a flowchart showing an example of the operation of executing troubleshooting processing to restrict user gameplay in this embodiment.

[0111] As shown in Figure 12, the game control unit 101 of the server device 10 matches users who will be opponents against each other by combining two or more users from among the multiple users participating in the game (step S21).

[0112] Next, the troubleshooting unit 103 of the server device 10 refers to the troubleshooting flag information set in the user information shown in Figure 6 and determines whether or not there is a user set as a troubleshooting user among the users who have become opponents through matching (step S22).

[0113] Next, if the troubleshooting unit 103 of the server device 10 determines that there is a user designated as a troubleshooting user among the users who have become opponents through matching (step S22: YES), it restricts gameplay as a troubleshooting measure for that user (step S23).

[0114] In this embodiment, the troubleshooting unit 103 resolves the matching that includes the user set as the troubleshooting user, and then performs matching again, excluding the user set as the troubleshooting user.

[0115] In this way, users designated as troubleshooters will have their gameplay restricted, thus preventing problems that could arise from communication between users.

[0116] Furthermore, even if a user designated as a troubleshooting user has their gameplay restricted, it is possible to configure the system so that the restriction is lifted once certain conditions are met (for example, after a certain amount of time has elapsed). Alternatively, the user can be removed from the troubleshooting user designation altogether.

[0117] <Specific Example 2> The operation regarding restricting user comment posting as a troubleshooting measure in this embodiment will be explained using Figure 13. Figure 13 is a flowchart showing an example of the operation of the troubleshooting process for restricting user comment posting in this embodiment.

[0118] As shown in Figure 13, the comment posting control unit 102 of the server device 10 determines whether or not the user who obtained the game result posted a comment after the game result has been determined for the game play that took place between matched users (step S31).

[0119] In this embodiment, the acquisition unit 102A of the server device 10 determines whether or not it has acquired, via the network, a posted comment submitted by a user who obtained game results during gameplay using the client terminal 20 from the client terminal 20.

[0120] Next, if the troubleshooting unit 103 of the server device 10 determines that a user who obtained a game result in that game has posted a comment (step S31: YES), it refers to the troubleshooting flag information set in the user information shown in Figure 6 to determine whether the user who posted the comment is set as a troubleshooting user (step S32).

[0121] Next, if the troubleshooting unit 103 of the server device 10 determines that the user who posted the comment is set as a troubleshooting user (step S32: YES), it restricts the posting of comments as a troubleshooting measure for that user (step S33).

[0122] In this embodiment, the troubleshooting unit 103 restricts the display of a user who has been set as a troubleshooting user so that the posted comment is not displayed in the posted comment display area 502 of the game screen 500 shown in Figure 11.

[0123] In this way, users designated as troubleshooters will have their comment posting restricted, thus preventing problems that could arise from user-to-user communication.

[0124] Furthermore, even if a user designated as a troubleshooting user has their comment posting restricted, it is possible to configure the system so that the restriction is lifted once certain conditions (for example, after a specified amount of time has elapsed) are met. In addition, the user may be removed from the troubleshooting user designation altogether.

[0125] ===Other Embodiments=== The embodiments described above are for the purpose of facilitating understanding of the present invention and are not intended to limit the interpretation of the present invention. The present invention can be modified and improved without departing from its spirit, and equivalents thereof are also included in the present invention. In particular, embodiments described below are also included in the present invention. That is, the present invention can also be constructed by appropriately combining the embodiments described above and the embodiments described below.

[0126] Furthermore, it is possible to construct various inventions by appropriately combining the multiple inventive features disclosed in the above-described embodiments. For example, an invention may be constructed by deleting some of the inventive features disclosed in the above-described embodiments, or an invention may be constructed by appropriately combining the inventive features disclosed in the above-described embodiments with the inventive features disclosed in the embodiments described below.

[0127] Furthermore, while the claims of this application describe the present invention using claims that selectively reference two or more other claims (referred to as "multi-claims"), the invention is not limited thereto. For example, the present invention also includes claims that describe the invention using multi-claims that reference at least one of the multi-claims described in the claims of this application (referred to as "multi-multi-claims").

[0128] <Game Results> In the above embodiment, when inputting the game results and posted comments of users who posted during gameplay as input text to the machine learning model, in addition to inputting the game result of "win" or "lose," the score or amount of points earned as a result of playing, or the time taken to reach the goal may also be input as game results. In this case, the input text of the training data is also modified and further trained on the machine learning model.

[0129] <Posted Comments> In the above embodiment, when inputting game results and posted comments from users who posted during gameplay into the machine learning model as input text, the model may be configured to perform data preprocessing on the posted comments before inputting them. Data preprocessing can remove noise such as anagrams and variations in spelling (bad characters) by performing a process to convert inappropriate characters into correct characters.

[0130] Furthermore, although the above-described embodiment explained an example in which users can post comments by entering text, the present invention is not limited to this. For example, in addition to text, users may also be able to post pre-prepared stamp images, voice messages, etc.

[0131] <Troubleshooting> In the above-described embodiment, an example was given of a case where a process to restrict the user's gameplay is executed as a troubleshooting measure, but the present invention is not limited to this.

[0132] For example, the troubleshooting unit 103 of the server device 10 may execute a process to restrict the actions of the character used by a user designated as a troubleshooting user within the game. Specifically, even if a user designated as a troubleshooting user inputs commands to a character, the system may restrict the character from performing certain actions (for example, actions such as bending the knees and lowering the hips, repeatedly crouching, or lowering the head).

[0133] Furthermore, while the above-described embodiment explained the case where a process is implemented to restrict user comment posting as a troubleshooting measure, the present invention is not limited to this.

[0134] For example, the troubleshooting unit 103 of the server device 10 may execute a process to delete a post made by a user designated as a troubleshooting user. Specifically, even if a post made by a user designated as a troubleshooting user is already displayed in the post comment display area 502 of the game screen shown in Figure 11, the post may be deleted from the post comment display area 502.

[0135] For example, if the system is configured to allow users to post not only text but also pre-prepared stamp images, voice messages, etc., the troubleshooting unit 103 of the server device 10 may perform a process to restrict users designated as troubleshooting users from posting not only text but also non-text items such as stamp images and voice messages, or it may perform a process to allow text posting but restrict only non-text items such as stamp images and voice messages.

[0136] Furthermore, in the above-described embodiment, the trouble countermeasure unit 103 of the server device 10 may be configured to gradually increase the trouble level for users designated as trouble countermeasure users each time they are evaluated as posting malicious comments. In this case, the trouble level gradually increases, and only when it eventually reaches a predetermined level will restrictions on game play and comment posting for users designated as trouble countermeasure users become effective. In this way, by quantifying and visualizing users who post malicious comments, it becomes possible to suppress troubles that develop from communication between users.

[0137] Furthermore, in the above-described embodiment, the troubleshooting unit 103 of the server device 10 may execute a process to restrict screenshots of the game screen for users designated as troubleshooting users. Screenshots of specific game screens, such as the game screen containing the opponent's profile, may be restricted. The screenshot restriction may be to prevent screenshots of the game screen altogether, or to allow screenshots of the game screen but restrict the screenshot to mask a portion of the game screen (for example, the portion containing the opponent's profile).

[0138] Furthermore, in the embodiment described above, the troubleshooting unit 103 of the server device 10 may perform troubleshooting actions for users designated as troubleshooting users, by appropriately combining the specific examples described above.

[0139] <Learning> In the above embodiment, the learning of the machine learning model was described using the example of the case where it is performed on the server device 10, but it may also be performed on an information processing device other than the server device 10. In this case, in the information processing system 1 according to this embodiment, the other information processing device and the server device 10 are connected via a network N, and the other information processing device will have the function of a learning unit.

[0140] <Game> In the embodiments described above, we explained using the example of a case where each user playing a competitive or cooperative game can communicate with each other by posting comments within the game. However, in these competitive and cooperative games, users may be assigned to one of the teams, and the teams may play against each other or play cooperatively. In the case of a competitive game, each user belonging to the winning team will receive a "victory" game result, and each user belonging to the losing team will receive a "defeat" game result.

[0141] 1 Information Processing System 10 Server Device 20 Client Terminal 50 Computer 51 CPU 52 RAM 53 ROM 54 Communication Interface 55 Input Device 56 Display Device 57 External Interface 58 HDD 100 Server Control Unit 101 Game Control Unit 102 Posted Comment Control Unit 102A Acquisition Unit 102B Evaluation Unit 103 Troubleshooting Unit 104 Learning Unit 120 Server Storage Unit 121 Machine Learning Model Storage Unit 122 Training Data Storage Unit 123 Character Information Storage Unit 124 User Information Storage Unit 125 Gameplay Information Storage Unit 140 Server Communication Unit 200 Terminal Control Unit 201 Server Access Unit 202 Display Control Unit 220 Terminal Storage Unit 240 Terminal Communication Unit 250 Operation Input Reception Unit 260 Screen Display Unit 500 Game Screen 501 Gameplay Area 502 Comment display area 503 Comment input area 504 Post button B Bus line N Network

Claims

1. A game system configured to allow multiple users participating in a game to interact with each other by posting comments, comprising: a game control unit that determines the game results for each user participating in the game; an evaluation unit that uses a machine learning model to generate an evaluation result of the posted comments as output data for input data including game results and posted comments, to evaluate whether the posted comments of users who have obtained game results are malicious or not; and a trouble prevention unit that, if a posted comment is evaluated as malicious, sets the user who posted the comment as a trouble prevention user.

2. The information processing system according to claim 1, wherein the game control unit determines whether each user wins or loses as a game result in a competitive game in which multiple users participate, and the evaluation unit uses a machine learning model that generates an evaluation result of the posted comments as output data for input data including the game result of win or loss and the posted comments to evaluate whether the posted comments of the user who obtained the game result of win or loss are malicious or not.

3. The information processing system according to claim 1, wherein the troubleshooting unit executes a process to restrict gameplay for the user designated as a troubleshooting user.

4. The information processing system according to claim 1, wherein the troubleshooting unit executes a process to restrict the movement of characters used in the game for the user who has been set as a troubleshooting user.

5. The information processing system according to claim 1, wherein the troubleshooting unit executes a process to restrict the posting of comments by the user designated as a troubleshooting user.

6. The information processing system according to claim 1, wherein the troubleshooting unit executes a process to delete posted comments made by the user who has been designated as a troubleshooting user.

7. The information processing system according to claim 1, wherein the trouble countermeasure unit progressively increases the trouble level of a user designated as a trouble countermeasure user each time that user's posted comment is evaluated as malicious.

8. An information processing device configured to allow multiple users participating in a game to interact with each other by posting comments, comprising: a game control unit that determines the game results for each user participating in the game; an evaluation unit that evaluates whether a user's posted comment is malicious or not, using a machine learning model that generates an evaluation result of the posted comment as output data for input data including the game results and the posted comment; and a trouble prevention unit that, if the posted comment is evaluated as malicious, sets the user who posted the comment as a trouble prevention user.

9. A program that enables an information processing device configured to allow multiple users participating in a game to interact with each other by posting comments to one another to function as: a game control means for determining the game results of each user participating in the game; an evaluation means for evaluating whether a user's posted comment is malicious or not, using a machine learning model that generates an evaluation result of the posted comment as output data for input data including the game result and the posted comment; and a trouble prevention means for setting the user who posted the comment as a trouble prevention user if the comment is evaluated as malicious.

Images