Comment processing method and apparatus

Abstract

The application provides a comment processing method and device, comprising: obtaining comments of a target object on each game level in a game, and determining a first feature based on the comments of the target object on each game level, the first feature being used to indicate an emotional attribute of the target object represented by the comments and a game progress corresponding to the comments; for the comments in each game level, iteratively obtaining a second feature of the comments in each game level based on the first feature of the comments, the second feature being used to indicate a change in the emotional attribute of the target object; determining a behavior event sequence of the target object in the game, and performing feature extraction on the behavior event sequence to obtain a third feature, the third feature being used to indicate the emotional attribute of the target object represented by the behavior event sequence; and detecting the game based on a fusion feature obtained by fusing the second feature and the third feature to obtain a first detection result of the game. In this way, the accuracy of the detection result of the game can be improved.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of computer, in particular to a comment processing method and device. BACKGROUND

[0002] In the related art, when detecting a game based on user comments, the accuracy of the detection result of the game is low due to a large amount of meaningless data in the user comments. SUMMARY

[0003] The embodiments of the present application provide a comment processing method and device, an electronic device, a computer readable storage medium and a computer program product, which can improve the accuracy of the detection result of the game.

[0004] The technical solutions of the embodiments of the present application are implemented as follows:

[0005] The embodiments of the present application provide a comment processing method, which comprises:

[0006] Obtaining comments of a target object on each game level in a game, and determining a first feature based on the comments of the target object on each game level, the first feature being used to indicate an emotional attribute of the target object to the game and a game progress corresponding to the comments represented by the comments;

[0007] Iteratively obtaining a second feature of the comments in each game level based on the first feature of the comments for the comments in each game level, the second feature being used to indicate a change in the emotional attribute of the target object to the game;

[0008] Determining a behavior event sequence of the target object in the game, and performing feature extraction on the behavior event sequence to obtain a third feature, the third feature being used to indicate an emotional attribute of the target object to the game represented by the behavior event sequence;

[0009] Fusing the second feature and the third feature to obtain a fused feature, and detecting the game based on the fused feature to obtain a first detection result of the game, the first detection result being used to indicate a difficulty of a target game level in the game.

[0010] The embodiments of the present application provide a comment processing device, which comprises:

[0011] A first obtaining module is configured to obtain comments of a target object on each game level in a game, and determine a first feature based on the comments of the target object on each game level, the first feature being used to indicate an emotional attribute of the target object to the game and a game progress corresponding to the comments represented by the comments.

[0012] The second acquisition module is configured to acquire, for each game level, a second feature of the comment in each game level based on a first feature of the comment, the second feature being used to indicate a change in the emotional attribute of the target object to the game;

[0013] The determination module is configured to determine a behavior event sequence of the target object in the game, and perform feature extraction on the behavior event sequence to obtain a third feature, the third feature being used to indicate an emotional attribute of the target object to the game represented by the behavior event sequence;

[0014] The detection module is configured to fuse the second feature and the third feature to obtain a fused feature, and perform detection on the game based on the fused feature to obtain a first detection result of the game, the first detection result being used to indicate a difficulty of a target game level in the game.

[0015] An electronic device is provided in an embodiment of the present application, and the electronic device comprises:

[0016] A memory is configured to store executable instructions.

[0017] A processor is configured to execute the executable instructions stored in the memory, so as to implement the comment processing method provided in the embodiments of the present application.

[0018] A computer readable storage medium is provided in an embodiment of the present application, and the computer readable storage medium stores computer executable instructions. When the computer executable instructions are executed by a processor, the processor executes the comment processing method provided in the embodiments of the present application.

[0019] An embodiment of the present application provides a computer program product, which comprises a computer program or computer executable instructions stored in a computer readable storage medium. A processor of an electronic device reads the computer executable instructions from the computer readable storage medium, and the processor executes the computer executable instructions, so that the electronic device executes the comment processing method provided in the embodiments of the present application.

[0020] The embodiments of the present application have the following beneficial effects:

[0021] Based on the target object's comments on each game level, after determining the first feature indicating the target object's emotional attributes towards the game and the corresponding game progress, for each game level's comments, based on the first feature, iteratively obtain the second feature indicating the changes in the target object's emotional attributes towards the game. Then, determine the target object's behavioral event sequence in the game and extract features from the behavioral event sequence to obtain the third feature indicating the target object's emotional attributes towards the game. Finally, based on the fused feature obtained by fusing the second and third features, the game is detected to obtain the feature indicating the target object's emotional attributes towards the game. The first detection result is the difficulty of the game levels. Therefore, compared to game detection based solely on user reviews, combining user reviews and user behavior enriches the information used in game detection, thus improving the accuracy of the results. Simultaneously, a second feature indicating changes in the target audience's emotional attributes towards the game is determined through an iterative processing process. Combining this second feature with the detection results further enhances accuracy. Furthermore, during detection, user behavior verifies the validity of user reviews, while user reviews interpret user behavior, further improving the precision of the detection results. Attached Figure Description

[0022] Figure 1 This is a schematic diagram of the architecture of the comment processing system 100 provided in an embodiment of this application;

[0023] Figure 2 This is a schematic diagram of the structure of the electronic device provided in the embodiments of this application;

[0024] Figure 3 This is a flowchart illustrating the comment processing method provided in an embodiment of this application;

[0025] Figure 4 This is a schematic diagram of the process for iteratively obtaining the second feature of a comment, provided in an embodiment of this application;

[0026] Figure 5 This is a schematic diagram of the behavioral event sequence provided in the embodiments of this application;

[0027] Figure 6 This is a technical architecture diagram of the data preprocessing process provided in the embodiments of this application. Detailed Implementation

[0028] In order to make the purposes, technical solutions and advantages of the present application clearer, the following further describes the present application in conjunction with the accompanying drawings, the described embodiments should not be regarded as limiting the present application, all other embodiments obtained by those of ordinary skill in the art without creative labor fall within the scope of protection of the present application.

[0029] In the following description, "some embodiments" are referred to, which describe a subset of all possible embodiments, but it can be understood that "some embodiments" can be the same subset or different subsets of all possible embodiments, and can be combined with each other without conflict.

[0030] In the following description, the terms "first\second\third" are only distinguished from similar objects, and do not represent a specific order of the objects, and it can be understood that "first\second\third" can be interchanged in a specific order or sequence as allowed, so that the embodiments of the present application described herein can be implemented in an order other than that illustrated or described herein.

[0031] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which the present application belongs. The terms used herein are only for the purpose of describing the embodiments of the present application and are not intended to limit the present application.

[0032] Before the embodiments of the present application are further described in detail, the terms and phrases involved in the embodiments of the present application are explained, and the terms and phrases involved in the embodiments of the present application are applicable to the following explanations.

[0033] 1) Artificial intelligence (AI, Artificial Intelligence) is to use digital computers or digital computer controlled machines to simulate, extend and expand human intelligence, perceive environment, acquire knowledge and use knowledge to obtain the best results. Artificial intelligence technology is a comprehensive discipline, involving a wide range of fields, both hardware and software level technology. Artificial intelligence basic technology generally includes technologies such as sensors, special artificial intelligence chips, cloud computing, distributed storage, big data processing technology, operation / interaction system, mechatronics, etc.

[0034] 2) Client (Client), also known as user end, is a program that provides local service for users corresponding to the server. Except for some application programs that can only run locally, it is generally installed on a common client and needs to run with the server, that is, it needs a corresponding server and service program in the network to provide corresponding services. Therefore, a specific communication connection needs to be established between the client and the server to ensure the normal operation of the application program.

[0035] 3) Long Short-Term Memory (LSTM), a special recurrent neural network, which solves the problem of gradient disappearance or gradient explosion in traditional recurrent neural networks when processing long sequence data, can effectively capture long-term dependencies in sequences, and is one of the core models in natural language processing, time series prediction and other fields.

[0036] 4) BERT model (Bidirectional Encoder Representations from Transformer), a pre-training model for natural language processing, which is used to train large-scale unlabeled corpus to obtain semantic representation of text containing rich semantic information, and then fine-tune the semantic representation of the text in a specific natural language processing task, and finally apply it to the natural language processing task.

[0037] Referring to Figure 1 , Figure 1 is an architecture diagram of a comment processing system 100 provided by an embodiment of the present application. To realize the application scenario of comment processing, a terminal (an example of a terminal 400) is connected to a server 200 through a network 300. The network 300 can be a wide area network or a local area network, or a combination of the two. The terminal 400 is used for a user to use a client 401 to display on a display interface (an example of a display interface 401-1). The terminal 400 and the server 200 are connected to each other through a wired or wireless network.

[0038] The server 200 is configured to obtain comments of a target object on each game level in a game, and determine a first feature based on the comments of the target object on each game level. The first feature is used to indicate the emotional attributes of the target object represented by the comments, and the game progress corresponding to the comments. For the comments in each game level, the second feature of the comments in each game level is iteratively obtained based on the first feature of the comments. The second feature is used to indicate the change in the emotional attributes of the target object. The behavior event sequence of the target object in the game is determined, and the third feature is obtained by feature extraction on the behavior event sequence. The third feature is used to indicate the emotional attributes of the target object represented by the behavior event sequence. The second feature and the third feature are fused to obtain a fusion feature, and the game is detected based on the fusion feature to obtain a first detection result of the game. The first detection result is used to indicate the difficulty of the target game level in the game. The first detection result is sent to the terminal.

[0039] The terminal 400 is configured to display the first detection result.

[0040] In some embodiments, the server 200 can be a standalone physical server, a server cluster composed of multiple physical servers, or a distributed system, and can also be a cloud server providing cloud services, cloud databases, cloud computing, cloud functions, cloud storage, network services, cloud communication, middleware services, domain name services, security services, content delivery networks (CDN), and basic cloud computing services such as big data and artificial intelligence platforms. The terminal 400 can be a smart phone, a tablet computer, a notebook computer, a desktop computer, a set-top box, a smart voice interaction device, a smart home appliance, a vehicle-mounted terminal, an aircraft, and a mobile device (for example, a mobile phone, a portable music player, a personal digital assistant, a dedicated message device, a portable game device, a smart speaker, and a smart watch), but is not limited thereto. The terminal device and the server can be connected directly or indirectly through wired or wireless communication, and the embodiments of the present application are not limited.

[0041] Referring to Figure 2 , Figure 2 is a structural schematic diagram of an electronic device provided by an embodiment of the present application. In actual application, the electronic device can be Figure 1 the server 200 or the terminal 400 shown in the figure, referring to Figure 2 , Figure 2 The electronic device shown in the figure includes at least one processor 410, a memory 450, at least one network interface 420, and a user interface 430. The various components in the terminal 400 are coupled together through a bus system 440. It can be understood that the bus system 440 is used to realize the connection and communication between the components. In addition to the data bus, the bus system 440 also includes a power bus, a control bus, and a status signal bus. However, for the purpose of clear illustration, all kinds of buses are marked as the bus system 440 in the figure. Figure 2

[0042] The processor 410 can be an integrated circuit chip with signal processing capability, such as a general-purpose processor, a digital signal processor (DSP), or other programmable logic devices, discrete gates or transistor logic devices, discrete hardware components, etc. The general-purpose processor can be a microprocessor or any conventional processor.

[0043] The user interface 430 includes one or more output devices 431 that enable the presentation of media content, including one or more speakers and / or one or more visual display screens. The user interface 430 also includes one or more input devices 432 that facilitate user input, such as a keyboard, a mouse, a microphone, a touch screen display, a camera, other input buttons and controls.

[0044] ​The memory 450 can be removable, non-removable, or a combination thereof. Exemplary hardware devices include solid-state memory, hard drives, optical disc drives, etc. The memory 450 optionally includes one or more storage devices remotely located from the processor(s) 410.

[0045] The memory 450 includes volatile memory or nonvolatile memory, and can also include both volatile and nonvolatile memory. Nonvolatile memory can be read only memory (ROM), volatile memory can be random access memory (RAM). The memory 450 described in the embodiments of the present application is intended to include any suitable type of memory.

[0046] In some embodiments, the memory 450 is capable of storing data to support various operations, examples of which include programs, modules, and data structures or a subset or superset thereof, which are exemplarily illustrated below.

[0047] The operating system 451 includes system programs for processing various basic system services and performing hardware-related tasks, such as a framework layer, a core library layer, a driver layer, etc., for implementing various basic services and processing hardware-based tasks;

[0048] The network communication module 452 is used to communicate with other electronic devices via one or more (wired or wireless) network interfaces 420, exemplary network interfaces 420 include Bluetooth, wireless compatibility certification (WiFi), and universal serial bus (USB), etc.

[0049] The presentation module 453 is used to enable the presentation of information via one or more output devices 431 associated with the user interface 430 (e.g., display screen, speaker, etc.), such as a user interface for operating peripheral devices and displaying content and information.

[0050] The input processing module 454 is used to detect and interpret one or more user inputs or interactions from one or more input devices 432.

[0051] In some embodiments, the apparatus provided by the embodiments of the present application can be implemented in software, Figure 2The comment processing apparatus 455 stored in the memory 450 is shown, which can be software in the form of programs and plug-ins, etc., including the following software modules: a first acquisition module 4551, a second acquisition module 4552, a determination module 4553, and a detection module 4554, which are logical, and thus can be combined or further split according to the implemented functions. The functions of the respective modules will be described below.

[0052] In some embodiments, the apparatus provided by the embodiments of the present application can be implemented in a hardware manner. For example, the comment processing apparatus provided by the embodiments of the present application can be a processor in the form of a hardware decoding processor, which is programmed to execute the comment processing method provided by the embodiments of the present application. For example, the processor in the form of a hardware decoding processor can use one or more application-specific integrated circuits (ASICs), DSPs, programmable logic devices (PLDs), complex programmable logic devices (CPLDs), field-programmable gate arrays (FPGAs), or other electronic elements.

[0053] In some embodiments, the terminal or server can implement the comment processing method provided by the embodiments of the present application by running a computer program. For example, the computer program can be a native program or a software module in an operating system; can be a native application program (APP), i.e., a program that needs to be installed in an operating system to run, such as an instant messaging APP or a web browser APP; can also be a mini-program, i.e., a program that only needs to be downloaded into a browser environment to run; and can also be a mini-program that can be embedded into any APP. In summary, the above computer program can be any form of application program, module or plug-in.

[0054] Based on the above description of the comment processing system and the electronic device provided by the embodiments of the present application, the comment processing method provided by the embodiments of the present application will be described below. In actual implementation, the comment processing method provided by the embodiments of the present application can be implemented by a terminal or a server alone, or by a terminal and a server cooperatively, to execute the comment processing method provided by the embodiments of the present application by the server 200 in the terminal 100 and the server 200 in the system 400 respectively. Figure 1 The server 200 in the system 400 will be taken as an example to illustrate the comment processing method provided by the embodiments of the present application, which is implemented by the server 200 alone. Figure 3 , Figure 3 is a flowchart of the comment processing method provided by the embodiments of the present application. Next, the illustrated steps will be described in combination with Figure 3 .

[0055] In step 101, the server obtains comments of the target object on each game level in the game, and determines a first feature based on the comments of the target object on each game level, the first feature being used to indicate an emotional attribute of the target object represented by the comments on the game, and a game progress corresponding to the comments.

[0056] In actual implementation, there are various ways for the server to obtain the comments of the target object on each game level in the game. In some embodiments, the comments input by the target object (i.e., a user or a player) in a comment submission interface can be obtained. For example, at the end of a level, the terminal can display a comment pop-up window and display prompt information in the comment pop-up window, such as “What suggestions do you have for this level?” or “How do you think about the difficulty of this level?” Then, in response to a comment operation performed based on the comment pop-up window, the server is sent with the comments input and submitted by the user, so that the server automatically associates the current level identifier and stores the comments. Alternatively, the terminal sets level scoring controls of multiple dimensions in a level interface (such as a pause menu or a level settlement page), the multiple dimensions including level difficulty, user satisfaction, etc., and then, in response to a scoring operation for the level performed based on the level scoring controls of each dimension, the scores of the user for each dimension are taken as comments of the user for the game level, and the corresponding comments are sent to the server. In other embodiments, the comments related to the level can also be extracted from in-game player social interactions. For example, the server detects world channels, team chat, etc. in real time to extract user comments (such as “How to beat the boss in Level 12” or “There are too many bugs in Level 5”), or extracts from the strategies or dynamics published by players, i.e., if the game supports players to publish level strategies or dynamics, the server takes the content published by the players as comment data when the players publish the content.

[0057] It should be noted that the way for the server to obtain the comments of the target object on each game level in the game includes but is not limited to the above-mentioned ways, which are not limited in the embodiments of the present application.

[0058] In some embodiments, after obtaining the comments of the target object on each game level, the comments can be filtered to obtain target comments. Then, the process of determining the first feature based on the comments of the target object on each game level can be determining the first feature based on the target comments of the target object on each game level.

[0059] It should be noted that the process of filtering the comments to obtain the target comments can be, first, filtering the comments according to rules to obtain first filtered comments, then filtering the first filtered comments according to semantics to obtain second filtered comments, and finally filtering the second filtered comments according to statistics to obtain the target comments.

[0060] Wherein, the comments are filtered by rules to obtain the first filtered comments, specifically, pre-defined comment types to be filtered are obtained, the comment types to be filtered at least include advertisement types and brushing types, wherein the comment of the advertisement type is, for example, the comment of fixed format or keyword combination such as "contact information (xxx-xxxx)", "link (http / https / short chain)", "low price recharge / agent practice", "scan code to get benefits", etc. The comment of the brushing type is the comment determined based on pre-set rules such as "pure number / symbol ratio ≥ 80%", "text repetition degree ≥ 90%", for example, meaningless brushing (such as "111", "not bad, not bad"), repeated brushing (such as the same account sending the same text), and induced brushing (such as "review 5 stars to get cashback") text mode.

[0061] Then, after obtaining the pre-defined comment types to be filtered, the comments are parsed to obtain core information and auxiliary information of the comments, wherein the core information includes keywords (such as "link", "agent practice") in the text, special symbols (such as "@", "#", "two-dimensional code description"), format structure (such as "xxx-xxxx", "xxx@xxx.com"), and the auxiliary information includes account information (such as whether newly registered, whether multiple comments are sent in a short time) of the comment publisher, publishing scene (such as whether there are a large number of same texts in the same level), for supplementing the judgment dimension; based on the core information and the auxiliary information of the comments, each comment is detected, when the detection result represents that the comment belongs to the comment type to be filtered, the corresponding comment is determined as a violation comment; finally, the violation comment in the obtained comments is removed to obtain the first filtered comments.

[0062] For the process of performing semantic filtering on the first filtered comments to obtain the second filtered comments, specifically, the semantic recognition is performed on the first filtered comments to obtain the semantic results of the first filtered comments, and based on the semantic results, the effective comments, i.e. the second filtered comments, are selected from the first filtered comments; wherein, the process of selecting the effective comments from the first filtered comments based on the semantic results, for example, is to select the comments containing the corresponding target words (such as the words of "level", "skill", "gold coin", etc.) from the first filtered comments based on the pre-set game term dictionary, as the effective comments, and delete meaningless short texts.

[0063] For the statistical filtering of the second filtered comment, the process of obtaining the target comment specifically includes, for each second filtered comment, performing word segmentation processing on the second filtered comment to obtain a plurality of word units included in the second filtered comment, then selecting invalid word units (such as "haha", "666", etc.) from the plurality of word units, and counting the proportion of the invalid word units in the plurality of word units included in the corresponding second filtered comment; when the proportion does not exceed a target proportion (such as 60%), the corresponding second filtered comment is taken as the target comment, and when the proportion exceeds the target proportion, the corresponding second filtered comment is filtered out.

[0064] In some embodiments, the process of determining the first feature based on the comments of the target object on each game level can be that, based on the comments of the target object on each game level, first sub-features are extracted to indicate the emotional attributes of the target object to the game represented by the comments; the game levels corresponding to each comment are determined, and second sub-features are extracted based on the game levels corresponding to each comment; the second sub-features are used to indicate the game progress corresponding to the comments; and the first sub-features and the second sub-features are spliced to obtain the first feature.

[0065] It should be noted that the first feature is a two-dimensional vector, the first dimension of which is the first sub-feature used to indicate the emotional attributes of the target object to the game represented by the comments, and the second dimension is the second sub-feature used to indicate the game progress corresponding to the comments; or the first dimension is the second sub-feature used to indicate the game progress corresponding to the comments, and the second dimension is the first sub-feature used to indicate the emotional attributes of the target object to the game.

[0066] It should be noted that the emotional attributes of the target object to the game are the feelings of the target object to the game, for example, thinking that the picture is high quality but the game is difficult, or the picture is low quality, or the operation is simple, etc. For example, when the comment is "picture is exquisite but operation is anti-human", the emotional attributes of the target object to the game are that the target object thinks that the picture is high quality but the game is difficult; and the game progress corresponding to the comment refers to the game level to which the comment belongs.

[0067] In actual implementation, the process of determining the first feature based on the comments of the target object on each game level is equivalent to performing two-dimensional structural conversion based on the comments of the target object on each game level to obtain a two-dimensional array corresponding to the comments, that is, a two-dimensional vector, which is the first feature. The process of determining the first feature based on the comments of the target object on each game level can be realized by a pre-trained model (such as a BERT model), which is not limited by the embodiments of the present application.

[0068] In step 102, for the comments in each game level, the second feature of the comments in each game level is iteratively obtained based on the first feature of the comments, and the second feature is used to indicate the change of the emotional attributes of the target object to the game.

[0069] In actual implementation, for the comments in each game level, the process of iteratively obtaining the second feature of the comments in each game level based on the first feature of the comments is as follows, referring to Figure 4 , Figure 4 is a flowchart of the process of iteratively obtaining the second feature of the comments provided by the embodiments of the present application, based on Figure 4 , for the comments in each game level, the process of iteratively obtaining the second feature of the comments in each game level based on the first feature of the comments is implemented by the following steps.

[0070] Step 1021, for the comments in the first game level, the second feature of the comments is determined based on the first feature of the comments.

[0071] In actual implementation, there are various methods for determining the second feature of the comments based on the comments in the first game level, specifically, in some embodiments, the first feature of the comments in the first game level can be directly taken as the second feature of the comments; in other embodiments, a comment is randomly sampled from the comments of other game levels, where the comments of other game levels are the comments of the game levels included in the plurality of game levels except the comments of the first game level, and then the first feature of the sampled comment and the first feature of the comments in the first game level are fused to obtain the second feature of the comments in the first game level.

[0072] It should be noted that the fusion of the first feature of the sampled comment and the first feature of the comments in the first game level can be addition or multiplication, etc., and the embodiments of the present application do not limit this; or, similar to the process of fusing the second feature of the comments in the i-1th game level and the first feature of the comments in the ith game level to obtain the second feature of the comments in the ith game level, which will be described below, the related process will be described below, which will not be described here.

[0073] Step 1022, for the comments in the ith game level, the following processing is performed to obtain the second feature of the comments in the ith game level: obtaining the second feature of the comments in the i-1th game level; where i is an integer greater than 1 and less than or equal to N, and N is the number of game levels.

[0074] It should be noted that the second feature of the comments in the i-1th game level is obtained, that is, when the comments in the i-1th game level are traversed, the second feature of the comments in the i-1th game level is determined according to a process similar to the process of determining the second feature of the comments in the ith game level described below.

[0075] In step 1023, the second feature of the comment in the i-1th game level and the first feature of the comment in the ith game level are fused to obtain the second feature of the comment in the ith game level.

[0076] In actual implementation, for the process of fusing the second feature of the comment in the i-1th game level and the first feature of the comment in the ith game level to obtain the second feature of the comment in the ith game level, specifically includes: based on the first feature of the comment in the ith game level, performing feature extraction on the second feature of the comment in the i-1th game level to obtain a first key feature; based on the second feature of the comment in the i-1th game level, performing feature extraction on the first feature of the comment in the ith game level to obtain a second key feature; and based on the first key feature and the second key feature, determining the second feature of the comment in the ith game level.

[0077] It should be noted that the first key feature is an important feature in the second feature of the comment in the i-1th game level, and the process of performing feature extraction on the second feature of the comment in the i-1th game level based on the first feature of the comment in the ith game level is that, based on the first feature of the comment in the ith game level, the features that need to be removed in the second feature of the comment in the i-1th game level are determined, in other words, based on the first feature of the comment in the ith game level, the important features in the second feature of the comment in the i-1th game level, that is, the features that are valuable to the prediction result, are determined, so as to obtain the first key feature.

[0078] Correspondingly, the second key feature is an important feature in the first feature of the comment in the ith game level, and the process of performing feature extraction on the first feature of the comment in the ith game level based on the second feature of the comment in the i-1th game level is that, based on the second feature of the comment in the i-1th game level, the features that need to be written, extracted or fused in the first feature of the comment in the ith game level are determined, that is, the features that are valuable to the prediction result, in other words, based on the second feature of the comment in the i-1th game level, the important features in the first feature of the comment in the ith game level are determined, so as to obtain the second key feature.

[0079] In actual implementation, for the process of determining the second feature of the comment in the ith game level based on the first key feature and the second key feature, the first key feature and the second key feature can be fused to obtain the second feature of the comment in the ith game level. Wherein, the first key feature and the second key feature can be added or multiplied, and the present application does not limit the same.

[0080] It should be noted that, after the first feature of the comments in the i-th game level is extracted based on the second feature of the comments in the i-1-th game level to obtain the first key feature, and the first feature of the comments in the i-th game level is extracted based on the second feature of the comments in the i-1-th game level to obtain the second key feature, the process of determining the second feature of the comments in the i-th game level based on the first key feature and the second key feature includes but is not limited to the above method, and the embodiments of the present application are not limited thereto.

[0081] In step 1024, i is traversed to obtain the second feature of the comments in each game level.

[0082] It should be noted that, when i is equal to N, that is, the number of comments in the plurality of game levels, the second feature of the comments in each game level can be obtained.

[0083] In step 103, the behavior event sequence of the target object in the game is determined, and the third feature is extracted from the behavior event sequence to obtain the third feature, which is used to indicate the emotional attribute of the target object in the behavior event sequence.

[0084] In some embodiments, the process of determining the behavior event sequence of the target object in the game can be that, for each game level, the following processing is performed to obtain the behavior event sequence of the target object in the game level: obtaining at least one behavior event performed by the target object in the game level, and obtaining the execution time of each behavior event; sorting the at least one behavior event according to the execution time of each behavior event; and constructing the behavior event sequence in the game level based on the sorted at least one behavior event.

[0085] It should be noted that the behavior events herein include, for example, level failure, reviving by watching advertisements, reviving by spending gold coins, using props, level victory, exiting the game, in-app purchase behavior, etc.; in a game level, the execution time of each behavior event of the user is obtained, and the at least one behavior event is sorted based on the execution time of each behavior event to obtain the behavior event sequence in the game level. For example, see Figure 5 , Figure 5 is a schematic diagram of the behavior event sequence provided by the embodiments of the present application, based on Figure 5 In a game level, four behavior events performed by the target object in the game level, i.e., level failure (i.e., death in the first level), reviving by watching advertisements, level victory (i.e., passing the first level), and exiting the game, are obtained, and the four behavior events are sorted based on the execution time of each behavior event to obtain the behavior event sequence as shown in Figure 5 .

[0086] In actual implementation, the process of feature extraction on the sequence of behavior events to obtain the third feature can be: feature extraction on each behavior event in the sequence of behavior events to obtain a fourth feature; based on the fourth feature of each behavior event in the sequence of behavior events, iteratively obtaining a fifth feature of each behavior event; and determining the fifth feature of the last behavior event in the sequence of behavior events as the third feature of the sequence of behavior events.

[0087] It should be noted that the process of iteratively obtaining the fifth feature of each behavior event based on the fourth feature of each behavior event in the sequence of behavior events includes: for the first behavior event, determining the fifth feature of the behavior event based on the fourth feature of the first behavior event; for the jth behavior event, performing the following processing to obtain the fifth feature of the jth behavior event: obtaining the fifth feature of the (j-1)th behavior event; where j is an integer greater than 1 and less than or equal to M, M is the number of behavior events in the sequence of behavior events; and fusing the fifth feature of the (j-1)th behavior event and the fourth feature of the jth behavior event to obtain the fifth feature of the jth behavior event; and traversing j to obtain the fifth feature of each behavior event.

[0088] In actual implementation, as described above for the process of determining the second feature of the comment in the first game level based on the first feature of the comment, there are multiple methods for determining the fifth feature of the first behavior event based on the fourth feature of the first behavior event. Specifically, in some embodiments, the fourth feature of the first behavior event can be directly used as the fifth feature of the behavior event; in other embodiments, a behavior event is randomly sampled from other behavior events, where the other behavior events are the behavior events in the sequence of behavior events except the first behavior event, and then the fourth feature of the sampled behavior event and the fourth feature of the first behavior event are fused to obtain the fifth feature of the first behavior event.

[0089] It should be noted that the fusion of the fourth feature of the sampled behavior event and the fourth feature of the first behavior event can be addition or multiplication, and the present embodiments do not limit the fusion method. Alternatively, the fusion process is similar to the process of fusing the fifth feature of the (j-1)th behavior event and the fourth feature of the jth behavior event to obtain the fifth feature of the jth behavior event, which will be described below.

[0090] In actual implementation, the following process is performed to obtain the second feature of the comment in the ith game level: the process of obtaining the second feature of the comment in the (i-1)th game level. The following process is performed to obtain the fifth feature of the jth behavior event: the process of obtaining the fifth feature of the (j-1)th behavior event, that is, the fifth feature of the (j-1)th behavior event is determined according to a process similar to the process of determining the fifth feature of the jth behavior event described below when the (j-1)th behavior event is traversed.

[0091] In actual implementation, the following process is performed to obtain the second feature of the comment in the ith game level: the process of obtaining the second feature of the comment in the (i-1)th game level, and the first feature of the comment in the ith game level is fused. The following process is performed to obtain the fifth feature of the jth behavior event: the process of fusing the fifth feature of the (j-1)th behavior event and the fourth feature of the jth behavior event. Specifically, the fifth feature of the (j-1)th behavior event is feature-extracted based on the fourth feature of the jth behavior event to obtain a third key feature; the fourth feature of the jth behavior event is feature-extracted based on the fifth feature of the (j-1)th behavior event to obtain a fourth key feature; and the fifth feature of the jth behavior event is determined based on the third key feature and the fourth key feature.

[0092] It should be noted that the third key feature is an important feature in the fifth feature of the (j-1)th behavior event. The process of feature-extracting the fifth feature of the (j-1)th behavior event based on the fourth feature of the jth behavior event is that, based on the fourth feature of the jth behavior event, the features that need to be removed from the fifth feature of the (j-1)th behavior event are determined, in other words, based on the fourth feature of the jth behavior event, the important features in the fifth feature of the (j-1)th behavior event, that is, the features that are valuable to the prediction result, are determined, thereby obtaining the third key feature.

[0093] Correspondingly, the second key feature is an important feature in the fourth feature of the jth behavior event. The process of feature-extracting the fourth feature of the jth behavior event based on the fifth feature of the (j-1)th behavior event is that, based on the fifth feature of the (j-1)th behavior event, the features that need to be written, extracted or fused from the fourth feature of the jth behavior event are determined, that is, the features that are valuable to the prediction result, in other words, based on the fifth feature of the (j-1)th behavior event, the important features in the fourth feature of the jth behavior event are determined, thereby obtaining the fourth key feature.

[0094] Then, for determining the fifth feature of the jth behavior event based on the third key feature and the fourth key feature, a process can be that the third key feature and the fourth key feature are fused to obtain the fifth feature of the jth behavior event. Wherein, the third key feature and the fourth key feature are fused, for example, can be that the third key feature and the fourth key feature are added or multiplied, etc., for which the embodiments of the present application are not limited.

[0095] Finally, j is traversed, when j is equal to M, that is, the number of behavior events in the behavior event sequence, the fifth feature of each behavior event can be obtained.

[0096] In actual implementation, after obtaining the fifth feature of each behavior event, the fifth feature of the last behavior event in the behavior event sequence is determined as the third feature of the behavior event sequence. Thus, since the fifth feature of the last behavior event has fused the features of all previous behavior events in the traversal process, the fifth feature of the last behavior event can reflect the characteristics of the behavior event sequence, that is, the fifth feature of the last behavior event in the behavior event sequence can be determined as the third feature of the behavior event sequence.

[0097] Wherein, here the third feature is used to indicate the emotional attribute of the target object of the behavior event sequence to the game, which refers to the feeling of the target object of the behavior event sequence to the corresponding game level in the game. For example, if the behavior event sequence of the target object, that is, the user, is "continuous death→watching advertisements→quitting the game level", it can be determined that the feeling of the user to the corresponding game level in the game can include frustration, etc., for which the embodiments of the present application are not limited.

[0098] It should be noted that the process of determining the second feature and determining the third feature described above can be realized by a pre-trained model (such as an LSTM model, etc.), for which the embodiments of the present application are not limited.

[0099] Step 104, the second feature and the third feature are fused to obtain a fused feature, and the game is detected based on the fused feature to obtain a first detection result of the game, the first detection result is used to indicate the difficulty of the target game level in the game.

[0100] In actual implementation, the process of fusing the second feature and the third feature can include fusing the second feature and the third feature by using attention learning, for example, multiplying or adding the second feature and the third feature, and the like, which is not limited by the embodiments of the present application. Then, the game is detected based on the fused feature to obtain a first detection result for indicating the difficulty of a target game level in the game; wherein the number of the target game level can be one or more, which is not limited by the embodiments of the present application; and the first detection result is equivalent to the analysis of the game, for example, the 10th level is too difficult, the 20th level is too simple, and the like.

[0101] It should be noted that in addition to the comment processing method in the above process, after obtaining the comments of the target object on each game level and the behavior events of the target object in the game, a topological graph corresponding to the target object can be constructed based on the comments and the behavior events, so as to detect the game based on the topological graph to obtain a first detection result of the game, which is not limited by the embodiments of the present application.

[0102] In some embodiments, after fusing the second feature and the third feature to obtain the fused feature, the game can be detected based on the fused feature to obtain at least one of a second detection result and a third detection result; wherein the second detection result is used to indicate a game parameter that has a problem in the game; and the third detection result is used to indicate a game design mechanism preferred by the target object in the game.

[0103] It should be noted that the game parameter at least includes frame rate, network rate, and the like, which is not limited by the embodiments of the present application; and the game design mechanism preferred by the target object in the game is, for example, an excellent design in the game, such as the design of a paid gift package that pops up after the user passes through a game level after performing a paid operation in the game level.

[0104] When the second detection result is obtained, a prompt information can be output based on the second detection result, and the prompt information is used to prompt the game parameter that has a problem in the game and a way to solve the corresponding problem; for example, if the game parameter that has a problem in the game is the network rate, it means that there is a network lag when the user plays the game, and a performance optimization prompt is triggered, and the like, which is not limited by the embodiments of the present application.

[0105] In some embodiments, after detecting the game based on the fused feature to obtain a first detection result of the game, a first detection result of another object for the game can be obtained; based on a plurality of first detection results, in response to at least one game level having a game level to be adjusted, a prompt information is output; wherein the prompt information is used to prompt that the difficulty of the game level to be adjusted needs to be adjusted.

[0106] It should be noted that the other object is an object participating in the game other than the target object, that is, other users. Here, the first detection result of the other object for the game is similar to the process of determining the first detection result of the target object for the game described above, that is, the first detection result of the other object for the game is obtained through the comment processing method described above. For this, the embodiments of the present application do not repeat the description.

[0107] In actual implementation, the difficulty of the target game level indicated by the first detection result includes difficult and simple, so after obtaining the first detection result of the other object for the game, for each game level, based on a plurality of first detection results, a first number of first detection results indicating that the difficulty of the corresponding game level is difficult and a second number of first detection results indicating that the difficulty of the corresponding game level is simple are counted; then, based on the first number and the second number, a game level to be adjusted is determined from at least one game level.

[0108] Among them, for the process of determining the game level to be adjusted from at least one game level based on the first number and the second number, specifically, for each game level, a first ratio of the first number to the total number of a plurality of first detection results and a second ratio of the second number to the total number of a plurality of first detection results are obtained; in response to the first ratio reaching a target ratio and / or the second ratio reaching a target ratio, the corresponding game level is determined as the game level to be adjusted; here, the target ratio is pre-set, and for this, the embodiments of the present application are not limited.

[0109] Based on this, when a certain number of users feel that the difficulty of the same game level is too difficult or too simple, it is determined that the difficulty of the game level is unbalanced, so as to output prompt information for prompting that the difficulty of the corresponding game level needs to be adjusted, so that the staff adjusts the difficulty of the corresponding game level.

[0110] With the above embodiments of the application, based on the comments of the target object on each game level, the first feature of the emotional attribute of the target object to the game indicated by the comment representation and the game progress corresponding to the comment are determined, and for the comments in each game level, the second feature of the comments in each game level for indicating the change of the emotional attribute of the target object to the game is iteratively obtained based on the first feature of the comments, and then the behavior event sequence of the target object in the game is determined, and the behavior event sequence is feature extracted to obtain the third feature of the emotional attribute of the target object to the game indicated by the behavior event sequence representation, and finally the game is detected based on the fusion feature obtained by fusing the second feature and the third feature to obtain the first detection result for indicating the difficulty of the target game level in the game; in this way, compared with the game detection based on only the single dimension of user comments, the game is detected in combination with user comments and user behavior, the information relied on in the game detection is enriched, and therefore the accuracy of the detection result is improved; meanwhile, the second feature for indicating the change of the emotional attribute of the target object to the game is determined through the iterative processing process, so that the detection result is obtained in combination with the second feature in the detection process, and the accuracy of the detection result can also be improved; in addition, in the detection process, the user behavior can verify the effectiveness of the user comments, and the user comments can interpret the user behavior, so that the accuracy of the detection result can be further improved.

[0111] In the following, an exemplary application of the embodiments of the application in an actual application scenario will be described.

[0112] In the related art game, when detecting the game based on user evaluation, the accuracy of the detection result for the game is low due to the large amount of meaningless data in the user evaluation.

[0113] Based on this, the embodiments of the application provide a comment processing method, and propose a "evaluation-behavior-game event" three-dimensional analysis framework: first, a game evaluation special noise filter is designed and implemented to identify and filter meaningless text; second, the comment is converted into a two-dimensional space-time array, that is, the time axis indicates the game progress stage, and the space axis indicates the game subsystem score; third, combined with in-game event stream data such as level failure, prop consumption, and advertisement clicks, the causal chain of negative comments and game design is mined through LSTM.

[0114] It should be noted that the key points of the technical solutions of the application include: first, a structured preprocessing engine converts the original text into a two-dimensional numerical representation through a game evaluation special noise filter to solve the noise interference problem; second, a dynamic gated LSTM network performs efficient feature extraction and multi-dimensional classification on the structured text; third, an evaluation-behavior joint analysis model is constructed in combination with user behavior time series data; fourth, an interpretable game analysis result is generated based on cross-modal attention.

[0115] In actual implementation, refer to Figure 6 , Figure 6 is a technical architecture diagram of a data preprocessing process provided by an embodiment of the application, based on Figure 6 , the comment processing method provided by the application first preprocesses the comments. Specifically, first, as shown in the dashed box 601, intelligent preprocessing of game comments is performed (a three-level filtering engine is used to solve the noise problem). First, rule layer filtering is performed, that is, a pre-defined template library is used to remove fake comments and advertising text. Then, semantic layer filtering is performed, that is, based on a game terminology dictionary (containing 300+ words such as "level", "skill", and "gold coins"), valid comments are retained, and meaningless short texts are deleted. Finally, statistical layer filtering is performed, that is, the TF-IDF value is calculated, and high-frequency invalid words (such as comments with a proportion of "haha" and "666" greater than 60%) are dynamically filtered. Second, as shown in the dashed box 602, two-dimensional structured conversion of the comments is performed, that is, the comments are converted through a time dimension (divided according to the game progress stage of the player) and a space dimension (game subsystem scores are extracted). The conversion process can use a BERT model to extract subsystem sentiment values (sentiment attributes) from the text. At the same time of preprocessing the comments, behavior data and game event encoding are also performed. Specifically, a key behavior sequence (behavior event sequence) is constructed. Here, the key events include level failure, watching advertisements to revive, spending gold coins to revive, using props, level victory, quitting the game, and in-app purchase behavior.

[0116] In actual implementation, after the comment preprocessing and the construction of the behavior sequence, double-channel LSTM joint modeling can be performed. Specifically, first, model input is performed. Channel one inputs a two-dimensional matrix of comments (first feature), and channel two inputs a behavior event sequence (such as time sequence encoding of [death, advertising, and payment]). Then, feature extraction is performed based on the two-dimensional matrix of comments and the behavior event sequence. Specifically, through an LSTM model, for the two-dimensional matrix of comments, the matrix rows are read in the order of game stages to capture the sentiment changes in the progress evolution (second feature), for example, discovering that the control score suddenly drops in the novice period (indicating a defect in novice guide). Through an LSTM model, for the behavior event sequence, the event sequence pattern (such as "continuous death → advertising → exit" indicating frustration) is analyzed. Finally, attention mechanism fusion is performed to automatically calculate the correlation weight of behavior and comments, so as to obtain the output result of the model (first detection result). Here, the output result of the model is an analysis of the game, such as that the 10th level is too difficult and the 20th level is too simple.

[0117] It should be noted that for the construction of the data set, specifically, first, the evaluation team evaluates the existing content of the game, and converts the evaluation content into the output result of the model, then all the evaluation contents are shuffled and randomly divided into a test set and a validation set in a ratio of 9:1, and then the learning rate and other parameters are adjusted to train the model, and the results with high efficiency and accuracy are screened.

[0118] In actual implementation, the technical solution of the present application can analyze game design problems, such as when the negative evaluation rate of a certain level is >60%, and the number of deaths in the associated behavior increases sharply, it is determined that the difficulty is unbalanced; it can also analyze device compatibility problems, such as negative reviews containing "stuttering" and the player's device being a low-end device, triggering performance optimization warning; it can also analyze excellent design, such as a certain level after the user gives a good review, and there is a user payment record in the associated behavior, then the pay package design after the level is popped out is excellent.

[0119] It should be noted that because of different gameplays, different types of games need to train different models, but similar types of games can use the same model parameters, and when the analysis of the game in the output result exceeds a certain number of problems, the corresponding team needs to modify the problems.

[0120] In this way, through the present application, the following technical effects can be achieved: first, noise immunity: accurately retaining effective information such as "the 5th level is too difficult", and filtering worthless text such as "fun"; second, causality explicitness: establishing a quantifiable correlation between behavior data (such as 7 deaths) and text evaluation (such as "can't beat"); third, dynamic optimization: adjusting game parameters (such as difficulty / advertising frequency) in real time based on attribution analysis; fourth, resource efficiency: the two-dimensional structure improves the training speed of LSTM, meeting the lightweight needs of mobile terminals.

[0121] With the above embodiments of the present application, based on the comments of the target object on each game level, the first feature of the emotional attribute of the target object to the game indicated by the comments and the game progress corresponding to the comments are determined, and then for the comments in each game level, the second feature of the comments in each game level for indicating the change of the emotional attribute of the target object to the game is iteratively obtained based on the first feature of the comments, the behavior event sequence of the target object in the game is determined again, and feature extraction is performed on the behavior event sequence to obtain the third feature of the emotional attribute of the target object to the game indicated by the behavior event sequence feature, and finally the game is detected based on the fusion feature obtained by fusing the second feature and the third feature to obtain the first detection result for indicating the difficulty of the target game level in the game; in this way, compared with the game detection based on only the single dimension of user comments, the game is detected in combination with user comments and user behavior, the information relied on in the game detection is enriched, and the accuracy of the detection result is improved; at the same time, the second feature for indicating the change of the emotional attribute of the target object to the game is determined through the iterative processing process, so that the detection result is obtained in combination with the second feature in the detection process, and the accuracy of the detection result can also be improved; in addition, in the detection process, the user behavior can verify the effectiveness of the user comments, and the user comments can interpret the user behavior, so that the accuracy of the detection result can be further improved.

[0122] The following continues to illustrate an exemplary structure of the implementation of the comment processing apparatus 455 provided by the embodiments of the present application as a software module, in some embodiments, as shown in Figure 2 The software module stored in the comment processing apparatus 455 of the memory 450 can include:

[0123] The first obtaining module 4551 is configured to obtain the comments of the target object on each game level in the game, and determine the first feature based on the comments of the target object on each game level, the first feature being used to indicate the emotional attribute of the target object to the game indicated by the comments and the game progress corresponding to the comments;

[0124] The second obtaining module 4552 is configured to, for the comments in each game level, iteratively obtain the second feature of the comments in each game level based on the first feature of the comments, the second feature being used to indicate the change of the emotional attribute of the target object to the game;

[0125] The determining module 4553 is configured to determine the behavior event sequence of the target object in the game, and perform feature extraction on the behavior event sequence to obtain the third feature, the third feature being used to indicate the emotional attribute of the target object to the game indicated by the behavior event sequence feature;

[0126] The detection module 4554 is configured to fuse the second feature and the third feature to obtain a fused feature, and detect the game based on the fused feature to obtain a first detection result of the game, where the first detection result is used to indicate a difficulty of a target game level in the game.

[0127] In some embodiments, the apparatus further includes a filtering module configured to filter the comments to obtain target comments; and the first obtaining module 4551 is further configured to determine the first feature based on the target comments of the target object on each game level.

[0128] In some embodiments, the first obtaining module 4551 is further configured to perform feature extraction on the comments of the target object on each game level to obtain a first sub-feature, where the first sub-feature is used to indicate a sentiment attribute of the target object on the game represented by the comments; determine a game level corresponding to each comment, and perform feature extraction on the game level corresponding to each comment to obtain a second sub-feature, where the second sub-feature is used to indicate a game progress corresponding to the comment; and splice the first sub-feature and the second sub-feature to obtain the first feature.

[0129] In some embodiments, the second obtaining module 4552 is further configured to, for a comment in a first game level, determine a second feature of the comment based on a first feature of the comment; and for a comment in an i-th game level, perform the following processing to obtain the second feature of the comment in the i-th game level: obtain a second feature of a comment in an (i-1)-th game level; where i is an integer greater than 1 and less than or equal to N, and N is a number of the game levels; fuse the second feature of the comment in the (i-1)-th game level and the first feature of the comment in the i-th game level to obtain the second feature of the comment in the i-th game level; and iterate i to obtain the second feature of the comment in each game level.

[0130] In some embodiments, the second obtaining module 4552 is further configured to perform feature extraction on the second feature of the comment in the (i-1)-th game level based on the first feature of the comment in the i-th game level to obtain a first key feature; perform feature extraction on the first feature of the comment in the i-th game level based on the second feature of the comment in the (i-1)-th game level to obtain a second key feature; and determine the second feature of the comment in the i-th game level based on the first key feature and the second key feature.

[0131] In some embodiments, the determining module 4553 is further configured to perform the following processing for each game level to obtain a sequence of behavioral events of the target object in the game level: obtaining at least one behavioral event performed by the target object in the game level and obtaining the execution time of each behavioral event; sorting the at least one behavioral event according to the execution time of each behavioral event; and constructing a sequence of behavioral events in the game level based on the sorted at least one behavioral event.

[0132] In some embodiments, the determining module 4553 is further configured to extract features from each of the behavioral events in the behavioral event sequence to obtain a fourth feature; for each of the behavioral events in the behavioral event sequence, based on the fourth feature of the behavioral event, iteratively obtain a fifth feature of each of the behavioral events; and determine the fifth feature of the last behavioral event in the behavioral event sequence as the third feature of the behavioral event sequence.

[0133] In some embodiments, the detection module 4554 is further configured to detect the game based on the fusion features to obtain at least one of a second detection result and a third detection result; wherein the second detection result is used to indicate problematic game parameters in the game; and the third detection result is used to indicate the game design mechanism preferred by the target object in the game.

[0134] In some embodiments, the detection module 4554 is further configured to acquire first detection results of other objects for the game; based on multiple first detection results, in response to the existence of a game level to be adjusted in at least one game level, output prompt information; wherein, the prompt information is used to indicate that the difficulty of the game level to be adjusted in the game needs to be adjusted.

[0135] This application provides a computer program product or computer program that includes computer instructions stored in a computer-readable storage medium. A processor of an electronic device reads the computer instructions from the computer-readable storage medium and executes the computer instructions, causing the electronic device to perform the comment processing method or job matching method described in this application, for example... Figure 3 The comment handling method is shown.

[0136] This application provides a computer-readable storage medium storing executable instructions. When these executable instructions are executed by a processor, they cause the processor to execute the comment processing method or job matching method provided in this application, for example... Figure 3 The comment handling method is shown.

[0137] In some embodiments, the computer-readable storage medium can be a Read-Only Memory (ROM), a Random Access Memory (RAM), an Erasable Programmable Read-Only Memory (EPROM), an Electrically Erasable Programmable Read-Only Memory (EEPROM), a flash memory, a magnetic surface memory, an optical disc, or a CD-ROM, etc. memory; or can be various devices including one or any combination of the above memories.

[0138] In some embodiments, the executable instructions can be in the form of a program, software, a software module, a script, or code, written in any form of programming language, including compiled or interpreted languages, or declarative or procedural languages, and can be deployed in any form, including as a stand-alone program or as a module, component, subroutine, or other unit suitable for use in a computing environment.

[0139] By way of example, the executable instructions can or can not correspond to a file in a file system, can be stored in a part of a file that holds other programs or data, for example, in one or more scripts stored in a Hyper Text Markup Language (HTML) document, in a single file dedicated to the program in question, or in multiple coordinated files, for example, files that store one or more modules, sub programs, or portions of code.

[0140] By way of example, the executable instructions can be deployed to execute on one electronic device or on multiple electronic devices that are located at one site, or that are distributed across multiple sites and that are interconnected by a communication network.

[0141] It should be noted that, in the embodiments of the present application, when the embodiments of the present application are applied to specific products or technologies, the user's permission or consent needs to be obtained, and the collection, use and processing of related data need to comply with relevant laws, regulations and standards of the country and region.

[0142] The above merely illustrates the embodiments of the present application, but is not intended to limit the protection scope of the present application. Any modification, equivalent replacement and improvement made within the spirit and scope of the present application shall be included in the protection scope of the present application.

Claims

1. A comment processing method, characterized in that, The method includes: In the game, the comments of the target object on each game level are obtained, and a first feature is determined based on the comments of the target object on each game level. The first feature is used to indicate the emotional attribute of the target object to the game represented by the comments, as well as the game progress corresponding to the comments. For the comments in the first game level, a second feature of the comments is determined based on a first feature of the comments; For the comments in the i-th game level, the following processing is performed to obtain the second feature of the comments in the i-th game level: Obtain the second feature of the comments in the (i-1)-th game level; where i is an integer greater than 1 and less than or equal to N, and N is the number of game levels; fuse the second feature of the comments in the (i-1)-th game level with the first feature of the comments in the i-th game level to obtain the second feature of the comments in the i-th game level; iterate through i to obtain the second feature of the comments in each game level, where the second feature indicates the change in the target object's emotional attributes towards the game; The sequence of behavioral events of the target object in the game is determined, and features are extracted from the sequence of behavioral events to obtain a third feature. The third feature is used to indicate the emotional attributes of the target object to the game as represented by the sequence of behavioral events. The second feature and the third feature are fused to obtain a fused feature, and the game is detected based on the fused feature to obtain a first detection result of the game. The first detection result is used to indicate the difficulty of the target game level in the game.

2. The method according to claim 1, characterized in that, After obtaining the target object's comments on each game level in the game, the method further includes: The comments are filtered to obtain the target comments; The process of determining the first feature based on the comments of the target object on each game level includes: The first feature is determined based on the target object's comments on each game level.

3. The method according to claim 1, characterized in that, The process of determining the first feature based on the comments of the target object on each game level includes: Based on the target object, feature extraction is performed on the comments of each game level to obtain a first sub-feature. The first sub-feature is used to indicate the emotional attribute of the target object to the game represented by the comments. The game level corresponding to each comment is determined, and features are extracted based on the game level corresponding to each comment to obtain a second sub-feature; the second sub-feature is used to indicate the game progress corresponding to the comment. The first sub-feature and the second sub-feature are concatenated to obtain the first feature.

4. The method according to claim 1, characterized in that, The process of fusing the second feature of the comments in the (i-1)th game level and the first feature of the comments in the ith game level to obtain the second feature of the comments in the ith game level includes: Based on the first feature of the comments in the i-th game level, feature extraction is performed on the second feature of the comments in the (i-1)-th game level to obtain the first key feature; Based on the second feature of the comments in the (i-1)th game level, feature extraction is performed on the first feature of the comments in the i-th game level to obtain the second key feature; Based on the first key feature and the second key feature, the second feature of the comments in the i-th game level is determined.

5. The method according to claim 1, characterized in that, Determining the sequence of behavioral events of the target object in the game includes: For each of the aforementioned game levels, the following processing is performed to obtain the sequence of behavioral events of the target object in the game level: Obtain at least one behavioral event performed by the target object in the game level, and obtain the execution time of each behavioral event; The at least one behavioral event is sorted according to its execution time. Based on the sorted at least one behavioral event, a sequence of behavioral events in the game level is constructed.

6. The method according to claim 1, characterized in that, The feature extraction of the behavioral event sequence to obtain the third feature includes: Feature extraction is performed on each behavioral event in the behavioral event sequence to obtain a fourth feature; For each behavioral event in the behavioral event sequence, a fifth feature of each behavioral event is iteratively obtained based on the fourth feature of the behavioral event; The fifth feature of the last behavioral event in the behavioral event sequence is determined as the third feature of the behavioral event sequence.

7. The method according to claim 1, characterized in that, After fusing the second feature and the third feature to obtain the fused feature, the method further includes: Based on the fusion features, the game is detected to obtain at least one of a second detection result and a third detection result; The second detection result is used to indicate the problematic game parameters in the game; the third detection result is used to indicate the game design mechanism preferred by the target object in the game.

8. The method according to claim 1, characterized in that, After detecting the game based on the fusion features and obtaining the first detection result of the game, the method further includes: Obtain the first detection results of other objects for the game; Based on multiple first detection results, in response to the existence of a game level that needs adjustment in at least one game level, a prompt message is output; The prompt message is used to indicate that the difficulty of the game level to be adjusted needs to be adjusted.

9. A comment processing device, characterized in that, The device includes: The first acquisition module is used to acquire comments from a target object on each game level in the game, and determine a first feature based on the comments from the target object on each game level. The first feature is used to indicate the emotional attributes of the target object towards the game represented by the comments, as well as the game progress corresponding to the comments. The second acquisition module is used to determine a second feature of a comment based on a first feature of the comment in the first game level; and to perform the following processing on a comment in the i-th game level to obtain the second feature of the comment in the i-th game level: acquiring the second feature of the comment in the (i-1)-th game level; where i is an integer greater than 1 and less than or equal to N, and N is the number of game levels; fusing the second feature of the comment in the (i-1)-th game level and the first feature of the comment in the i-th game level to obtain the second feature of the comment in the i-th game level; and iterating through i to obtain the second feature of the comment in each game level, wherein the second feature is used to indicate the change in the emotional attribute of the target object towards the game; The determination module is used to determine the sequence of behavioral events of the target object in the game, and to extract features from the sequence of behavioral events to obtain a third feature, wherein the third feature is used to indicate the emotional attributes of the target object to the game as represented by the sequence of behavioral events. The detection module is used to fuse the second feature and the third feature to obtain a fused feature, and to detect the game based on the fused feature to obtain a first detection result of the game. The first detection result is used to indicate the difficulty of the target game level in the game.

Images