Text recognition method applied to text detection system and text detection system

Abstract

The invention relates to the technical field of data recognition, and discloses a text recognition method applied to a text detection system and the text detection system.The method comprises the steps that a keyword set is extracted from text information through the text detection system, and sensitive word recognition is conducted on the keyword set according to a sensitive word set; when the sensitive word recognition result shows that at least one target keyword matched with the sensitive word set exists in the keyword set, the text detection system generates prompt information of each target keyword, and sending the prompt information of each target keyword to an operation management end of the target game. Therefore, by implementing the method, the comprehensiveness and the accuracy of identifying the sensitive information can be improved, the probability that sensitive words appear in the game text is reduced, the health degree and the safety of the game are improved, meanwhile, the probability that improper speech appears during chatting of players in the game is reduced, chatting behaviors of the players in the game are standardized, and the health of the game is improved.

Description

Technical Field

[0001] The present invention relates to the technical field of data recognition, and in particular to a text recognition method and a text detection system applied to a text detection system. Background Art

[0002] With the rapid development of Internet technology, online games have become one of the important ways for people to relax and entertain. In the game, players can not only experience rich plots, but also communicate with other players in real time through the chat system. However, this open communication environment also brings a series of problems, such as the spread of bad information and malicious speech, which may have a negative impact on the healthy operation of the game and the game experience of players.

[0003] In order to effectively supervise the text information in the game and ensure the healthy and civilized operation of the game, the text detection system came into being. However, the detection accuracy and comprehensiveness of sensitive words in the existing text detection system are low. Therefore, it is particularly important to propose a technical solution that can improve the detection accuracy and comprehensiveness of sensitive words in game texts. Summary of the invention

[0004] The present invention provides a text recognition method and a text detection system applied to a text detection system, which can help improve the detection accuracy and comprehensiveness of sensitive words in game texts.

[0005] In order to solve the above technical problems, the first aspect of the present invention discloses a text recognition method applied to a text detection system, the method comprising:

[0006] The text detection system obtains text information of the target game and extracts a keyword set from the text information, wherein the text information includes game content text information and / or user chat text information;

[0007] The text detection system obtains a preset sensitive word set, and performs sensitive word recognition on the keyword set according to the sensitive word set to obtain a sensitive word recognition result corresponding to the keyword set;

[0008] The text detection system analyzes the sensitive word recognition result. When the sensitive word recognition result indicates that there is at least one target keyword in the keyword set that matches the sensitive word set, the text detection system generates prompt information for each of the target keywords and sends the prompt information for each of the target keywords to the operation and management end of the target game.

[0009] As an optional implementation, in the first aspect of the present invention, the sensitive word recognition result includes a first sub-recognition result and a second sub-recognition result;

[0010] The text detection system performs sensitive word recognition on the keyword set according to the sensitive word set to obtain a sensitive word recognition result corresponding to the keyword set, including:

[0011] The text detection system performs sensitive word recognition on each keyword in the keyword set according to the sensitive word set to obtain the first sub-recognition result corresponding to the keyword set;

[0012] The text detection system determines keyword information of each keyword in the keyword set, and divides the keyword set according to the keyword information of each keyword to obtain at least two sub-keyword sets;

[0013] The text detection system determines the semantic features corresponding to each of the sub-keyword sets, and performs sensitive semantic recognition on the semantic features corresponding to each of the sub-keyword sets according to the sensitive word set to obtain the second sub-recognition result corresponding to the keyword set.

[0014] As an optional implementation, in the first aspect of the present invention, the text detection system determines the keyword information of each keyword in the keyword set, including:

[0015] The text detection system determines target information of each keyword in the keyword set, wherein the target information includes keyword occurrence frequency information and keyword meaning information;

[0016] The text detection system calculates a keyword score corresponding to each keyword according to the target information corresponding to each keyword;

[0017] The text detection system calculates the string distance between each of the keywords in the keyword set;

[0018] The text detection system determines the keyword information of each keyword according to the keyword score corresponding to each keyword and the character string distance between each keyword.

[0019] As an optional implementation, in the first aspect of the present invention, the text detection system divides the keyword set according to the keyword information of each keyword to obtain at least two sub-keyword sets, including:

[0020] The text detection system obtains a preset weight allocation rule, and determines a keyword scoring weight corresponding to each of the keywords and a string distance weight between each of the keywords according to the weight allocation rule;

[0021] The text detection system calculates the association parameter between each of the keywords according to the keyword score and the keyword score weight corresponding to each of the keywords, the string distance between each of the keywords and the string distance weight;

[0022] The text detection system divides the keyword set according to a preset association parameter threshold and an association parameter between each of the keywords to obtain at least two sub-keyword sets;

[0023] Each of the sub-keyword sets includes at least two associated keywords, and an associated parameter between each of the associated keywords in each of the sub-keyword sets is greater than the associated parameter threshold.

[0024] As an optional implementation, in the first aspect of the present invention, the text detection system determines the semantic features corresponding to each of the sub-keyword sets, including:

[0025] For each of the sub-keyword sets, the text detection system determines the expanded word information of each of the associated keywords in the sub-keyword set according to the keyword meaning of each associated keyword in the sub-keyword set, wherein the expanded word meaning information includes synonymous expanded word information and / or antonymous expanded word information;

[0026] For each of the sub-keyword sets, the text detection system determines contextual semantic information of each of the associated keywords in the sub-keyword set;

[0027] For each of the sub-keyword sets, the text detection system determines a first coefficient corresponding to the expanded word information and a second coefficient corresponding to the contextual semantic information of each of the associated keywords in the sub-keyword set;

[0028] For each of the sub-keyword sets, the text detection system randomly combines each of the associated keywords in the sub-keyword set to obtain at least one keyword combination, each of which includes at least two associated keywords;

[0029] For each of the sub-keyword sets and each of the keyword combinations, the text detection system determines the combined semantic features corresponding to the keyword combination according to the expanded word information corresponding to the keyword combination, the first coefficient, the contextual semantic information and the second coefficient;

[0030] For each of the sub-keyword sets, the text detection system determines the semantic features corresponding to the sub-keyword set according to the combined semantic features corresponding to each of the keyword combinations in the sub-keyword set.

[0031] As an optional embodiment, in the first aspect of the present invention, the method further comprises:

[0032] The text detection system obtains a set of ignored words for the target game, and updates the set of sensitive words according to the set of ignored words;

[0033] Furthermore, the text detection system performs sensitive word recognition on each keyword in the keyword set according to the sensitive word set to obtain the first sub-recognition result corresponding to the keyword set, including:

[0034] The text detection system performs sensitive word recognition on the keyword set according to the sensitive word set and the ignored word set to obtain a first target recognition result corresponding to the keyword set;

[0035] The text detection system selects a related word set corresponding to the first target recognition result from the keyword set, wherein the related word set includes at least one of a synonym, a synonym, and antonym corresponding to the first target recognition result;

[0036] The text detection system performs sensitive word recognition on the related word set according to the sensitive word set and the ignored word set to obtain a second target recognition result corresponding to the keyword set;

[0037] The text detection system determines the first sub-recognition result corresponding to the keyword set according to the first target recognition result and the second target recognition result.

[0038] As an optional implementation, in the first aspect of the present invention, when the sensitive word recognition result indicates that there is at least one target keyword in the keyword set that matches the sensitive word set, the method further includes:

[0039] The text detection system determines feature information of each of the keywords in the keyword set, wherein the feature information includes attribute label features and / or fine-grained features;

[0040] For each of the keywords, the text detection system classifies and stores the keyword into a preset feature word library according to feature information of the keyword;

[0041] And, the method further comprises:

[0042] When the text detection system receives the sensitive word recognition range and the sensitive word configuration requirements sent by the operation management end, the text detection system configures a sensitive word set based on the feature word library according to the sensitive word recognition range and the sensitive word configuration requirements.

[0043] A second aspect of the present invention discloses a text detection system, the text detection system comprising:

[0044] An acquisition module, used to acquire text information of a target game and extract a keyword set from the text information, wherein the text information includes game content text information and / or user chat text information;

[0045] The acquisition module is further used to acquire a preset sensitive word set, and perform sensitive word recognition on the keyword set according to the sensitive word set to obtain a sensitive word recognition result corresponding to the keyword set;

[0046] The analysis module is used to analyze the sensitive word recognition result. When the sensitive word recognition result indicates that there is at least one target keyword in the keyword set that matches the sensitive word set, prompt information for each target keyword is generated, and the prompt information for each target keyword is sent to the operation management end of the target game.

[0047] As an optional implementation, in the second aspect of the present invention, the sensitive word recognition result includes a first sub-recognition result and a second sub-recognition result;

[0048] The acquisition module performs sensitive word recognition on the keyword set according to the sensitive word set to obtain the sensitive word recognition result corresponding to the keyword set, specifically including:

[0049] Performing sensitive word recognition on each keyword in the keyword set according to the sensitive word set to obtain the first sub-recognition result corresponding to the keyword set;

[0050] Determine keyword information of each keyword in the keyword set, and divide the keyword set according to the keyword information of each keyword to obtain at least two sub-keyword sets;

[0051] Determine the semantic features corresponding to each of the sub-keyword sets, and perform sensitive semantic recognition on the semantic features corresponding to each of the sub-keyword sets according to the sensitive word set to obtain the second sub-recognition result corresponding to the keyword set.

[0052] As an optional implementation, in the second aspect of the present invention, the manner in which the acquisition module determines the keyword information of each keyword in the keyword set specifically includes:

[0053] Determine target information of each keyword in the keyword set, wherein the target information includes keyword occurrence frequency information and keyword meaning information;

[0054] Calculate the keyword score corresponding to each keyword according to the target information corresponding to each keyword;

[0055] Calculating the string distance between each of the keywords in the keyword set;

[0056] The keyword information of each keyword is determined according to the keyword score corresponding to each keyword and the character string distance between each keyword.

[0057] As an optional implementation, in the second aspect of the present invention, the acquisition module divides the keyword set according to the keyword information of each keyword to obtain at least two sub-keyword sets, specifically including:

[0058] Obtaining a preset weight allocation rule, and determining a keyword score weight corresponding to each of the keywords and a string distance weight between each of the keywords according to the weight allocation rule;

[0059] Calculate the association parameter between each of the keywords according to the keyword score and the keyword score weight corresponding to each of the keywords, the string distance between each of the keywords and the string distance weight;

[0060] Dividing the keyword set according to a preset association parameter threshold and an association parameter between each of the keywords to obtain at least two sub-keyword sets;

[0061] Each of the sub-keyword sets includes at least two associated keywords, and an associated parameter between each of the associated keywords in each of the sub-keyword sets is greater than the associated parameter threshold.

[0062] As an optional implementation, in the second aspect of the present invention, the manner in which the acquisition module determines the semantic features corresponding to each of the sub-keyword sets specifically includes:

[0063] For each of the sub-keyword sets, according to the keyword meaning of each associated keyword in the sub-keyword set, determining the expanded word information of each of the associated keywords in the sub-keyword set, wherein the expanded word meaning information includes synonymous expanded word information and / or antonymous expanded word information;

[0064] For each of the sub-keyword sets, determining contextual semantic information of each of the associated keywords in the sub-keyword set;

[0065] For each of the sub-keyword sets, determining a first coefficient corresponding to the expanded word information and a second coefficient corresponding to the contextual semantic information of each of the associated keywords in the sub-keyword set;

[0066] For each of the sub-keyword sets, randomly combine each of the associated keywords in the sub-keyword set to obtain at least one keyword combination, each of the keyword combinations including at least two of the associated keywords;

[0067] For each of the sub-keyword sets and each of the keyword combinations, determining the combination semantic features corresponding to the keyword combination according to the expansion word information corresponding to the keyword combination, the first coefficient, the context semantic information and the second coefficient;

[0068] For each of the sub-keyword sets, the semantic feature corresponding to the sub-keyword set is determined according to the combined semantic feature corresponding to each of the keyword combinations in the sub-keyword set.

[0069] As an optional implementation, in the second aspect of the present invention, the acquisition module is further used to acquire an ignored word set for the target game, and update the sensitive word set according to the ignored word set;

[0070] Furthermore, the acquisition module performs sensitive word recognition on each keyword in the keyword set according to the sensitive word set to obtain the first sub-recognition result corresponding to the keyword set, which specifically includes:

[0071] Performing sensitive word recognition on the keyword set according to the sensitive word set and the ignored word set to obtain a first target recognition result corresponding to the keyword set;

[0072] Filtering a related word set corresponding to the first target recognition result in the keyword set, wherein the related word set includes at least one of a synonym, a synonym, and antonym corresponding to the first target recognition result;

[0073] Performing sensitive word recognition on the related word set according to the sensitive word set and the ignored word set to obtain a second target recognition result corresponding to the keyword set;

[0074] The first sub-recognition result corresponding to the keyword set is determined according to the first target recognition result and the second target recognition result.

[0075] As an optional implementation, in the second aspect of the present invention, the text detection system further includes:

[0076] a determination module, configured to determine feature information of each keyword in the keyword set when the sensitive word recognition result indicates that there is at least one target keyword matching the sensitive word set in the keyword set, wherein the feature information includes attribute label features and / or fine-grained features;

[0077] A storage module, for classifying and storing each keyword into a preset feature word library according to feature information of the keyword;

[0078] And, the text detection system also includes:

[0079] The configuration module is used to configure the sensitive word set based on the feature word library according to the sensitive word recognition range and the sensitive word configuration requirements when receiving the sensitive word recognition range and the sensitive word configuration requirements sent by the operation management terminal.

[0080] A third aspect of the present invention discloses a text recognition device applied to a text detection system, the device comprising:

[0081] A memory storing executable program code;

[0082] a processor coupled to the memory;

[0083] The processor calls the executable program code stored in the memory to execute the text recognition method applied to the text detection system disclosed in the first aspect of the present invention.

[0084] The fourth aspect of the present invention discloses a computer storage medium, which stores computer instructions. When the computer instructions are called, they are used to execute the text recognition method applied to the text detection system disclosed in the first aspect of the present invention.

[0085] Compared with the prior art, the embodiments of the present invention have the following beneficial effects:

[0086] In the embodiment of the present invention, the text information of the target game can be obtained through the text detection system, and a keyword set can be extracted from the text information to obtain a preset sensitive word set, and the keyword set can be identified according to the sensitive word set to obtain a sensitive word identification result corresponding to the keyword set, and the sensitive word identification result can be analyzed. When the sensitive word identification result indicates that there is at least one target keyword matching the sensitive word set in the keyword set, the text detection system generates prompt information for each target keyword, and sends the prompt information for each target keyword to the operation management end of the target game. It can be seen that the implementation of the present invention can identify sensitive words in the game text based on the text detection system, and perform sensitive word reminders, improve the comprehensiveness and accuracy of identifying sensitive information, reduce the probability of sensitive words appearing in the game text during the game development process, improve the health and safety of the game, reduce the cost of game development and maintenance, and at the same time reduce the probability of inappropriate remarks when players chat in the game, standardize the chat behavior of players in the game, and improve the health of game operations. BRIEF DESCRIPTION OF THE DRAWINGS

[0087] In order to more clearly illustrate the technical solutions in the embodiments of the present invention, the following briefly introduces the drawings required for use in the description of the embodiments. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without creative work.

[0088] Figure 1 It is a flowchart of a text recognition method applied to a text detection system disclosed in an embodiment of the present invention;

[0089] Figure 2 is a structural schematic diagram of a text detection system disclosed in an embodiment of the present invention;

[0090] Figure 3 is a flow chart of another text recognition method applied to a text detection system disclosed in an embodiment of the present invention;

[0091] Figure 4 is a structural schematic diagram of another text detection system disclosed in an embodiment of the present invention;

[0092] Figure 5 is a structural schematic diagram of another text detection system disclosed in an embodiment of the present invention;

[0093] Figure 6 It is a structural schematic diagram of a text recognition device applied to a text detection system disclosed in an embodiment of the present invention. DETAILED DESCRIPTION

[0094] In order to enable those skilled in the art to better understand the scheme of the present invention, the technical scheme in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0095] The terms "first", "second", etc. in the specification and claims of the present invention and the above-mentioned drawings are used to distinguish different objects rather than to describe a specific order. In addition, the terms "including" and "having" and any variations thereof are intended to cover non-exclusive inclusions. For example, a process, method, device, product or end including a series of steps or units is not limited to the listed steps or units, but may optionally include steps or units that are not listed, or may optionally include other steps or units inherent to these processes, methods, products or ends.

[0096] Reference to "embodiments" herein means that a particular feature, structure, or characteristic described in conjunction with the embodiments may be included in at least one embodiment of the present invention. The appearance of the phrase in various places in the specification does not necessarily refer to the same embodiment, nor is it an independent or alternative embodiment that is mutually exclusive with other embodiments. It is explicitly and implicitly understood by those skilled in the art that the embodiments described herein may be combined with other embodiments.

[0097] The present invention discloses a text recognition method and a text detection system applied to a text detection system, which can identify sensitive words in game texts based on the text detection system, and provide sensitive word reminders, thereby improving the comprehensiveness and accuracy of identifying sensitive information, reducing the probability of sensitive words appearing in game texts during game development, improving the health and safety of games, reducing game development and maintenance costs, and reducing the probability of inappropriate remarks by players in the game during chats, regulating the chat behavior of players in the game, and improving the health of game operations. The following are detailed descriptions.

[0098] Embodiment 1

[0099] See also Figure 1 , Figure 1 1 is a flowchart of a text recognition method applied to a text detection system disclosed in an embodiment of the present invention. Figure 1 The text recognition method applied to the text detection system described above can be applied to the text detection system, and the text detection system may include an intelligent server or intelligent platform for performing keyword and / or sensitive word recognition and detection on the text, and the intelligent server includes a local server or a cloud server, which is not limited in the embodiment of the present invention. Figure 1 As shown, the text recognition method applied to the text detection system may include the following operations:

[0100] 101. The text detection system obtains text information of the target game and extracts a keyword set from the text information.

[0101] In the embodiment of the present invention, optionally, the text detection system can be used to identify keywords or sensitive words in the text, such as Figure 2 As shown, Figure 2 is a structural diagram of a text detection system disclosed in an embodiment of the present invention, wherein: Figure 2The text detection system shown can include a database layer, a service layer, a gateway layer, an access layer and an early warning layer from the bottom layer to the upper layer. The database layer supports active-standby switching and cluster mode deployment. The service layer supports multi-node deployment to eliminate single point failures. If any server goes down, other servers can continue to provide services. The gateway layer supports Tencent load balancing CLB (Cloud LoadBalancer). CLB automatically detects the health status of connected servers and automatically switches after a server goes down. The access layer supports a business fuse mechanism and automatic protection of API timeout. Even if the service goes down, it will not affect the normal process of the game. The early warning layer includes a complete early warning mechanism: early warnings such as vocabulary failure, response delay, request QPS (Query Per Second), server load, and code operation abnormalities, so as to achieve minute-level problem discovery and abnormal response.

[0102] In an embodiment of the present invention, optionally, the text information of the target game may include game content text information and / or user chat text information, the game content text information may include one or more of game plot text information, game system text information, game description text information, game tutorial text information, game achievement and / or reward text information, and game environment text information, the user chat text information may include chat text information between game players and / or game announcement text information, and the keyword set extracted from the text information may include one or more combinations of game character keywords, game environment keywords, game prop keywords, game task keywords, game plot keywords, game player behavior keywords, etc., which are not limited by the present invention.

[0103] 102. The text detection system obtains a preset sensitive word set, and performs sensitive word recognition on the keyword set according to the sensitive word set to obtain a sensitive word recognition result corresponding to the keyword set.

[0104] In the embodiment of the present invention, optionally, the preset sensitive word set can be configured by the game manager at the operation and management end, or the game manager sets sensitive word conditions at the operation and management end and then configures the text detection system according to the sensitive word conditions, or the game manager configures the sensitive word set based on the national sensitive word library at the operation and management end, which is not limited by the present invention.

[0105] 103. The text detection system analyzes the sensitive word recognition result. When the sensitive word recognition result indicates that there is at least one target keyword in the keyword set that matches the sensitive word set, the text detection system generates prompt information for each target keyword and sends the prompt information for each target keyword to the operation management end of the target game.

[0106] In an embodiment of the present invention, optionally, when the sensitive word recognition result indicates that there is at least one target keyword in the keyword set that matches the sensitive word set, the text detection system generates prompt information for each target keyword, the target keyword is a sensitive word, and the prompt information for each target keyword may include the sensitive word type and severity corresponding to the target keyword, which is not limited by the present invention.

[0107] It can be seen that the implementation Figure 1 The described text recognition method applied to a text detection system can obtain text information of a target game through a text detection system, extract a keyword set from the text information, obtain a preset sensitive word set, and perform sensitive word recognition on the keyword set according to the sensitive word set to obtain a sensitive word recognition result corresponding to the keyword set, analyze the sensitive word recognition result, and when the sensitive word recognition result indicates that there is at least one target keyword in the keyword set that matches the sensitive word set, the text detection system generates prompt information for each target keyword, and sends the prompt information for each target keyword to the operation management end of the target game, and can identify sensitive words in the game text based on the text detection system, and perform sensitive word reminders, thereby improving the comprehensiveness and accuracy of identifying sensitive information, reducing the probability of sensitive words appearing in the game text during the game development process, improving the health and safety of the game, reducing the cost of game development and maintenance, and at the same time reducing the probability of inappropriate remarks when players chat in the game, standardizing the chat behavior of players in the game, and improving the health of game operations.

[0108] In an optional embodiment, the prompt information of each target keyword includes the sensitive word type and severity level corresponding to the target keyword;

[0109] The text recognition method applied to the text detection system may also include the following operations:

[0110] For each target keyword, the operation management end determines the processing urgency of the target keyword according to the prompt information of the target keyword;

[0111] For each target keyword, the operation management end determines the original text corresponding to the target keyword, where the original text includes game content text or user chat text;

[0112] For each target keyword, when the original text corresponding to the target keyword includes the game content text, the operation management end performs the game text processing operation on the target keyword according to the prompt information and processing urgency of the target keyword;

[0113] For each target keyword, when the original text corresponding to the target keyword includes user chat text, the operation management end determines the user information of the game user corresponding to the target keyword, and determines the game standard processing operation for the game user based on the prompt information, processing urgency and user information of the target keyword, and performs user text operations on the target keyword.

[0114] In this optional embodiment, optionally, the prompt information of each target keyword may include the sensitive word type and severity corresponding to the target keyword, and the processing urgency of the target keyword may indicate the processing priority of the target keyword, which is not limited in this embodiment.

[0115] In this optional embodiment, optionally, when the original text corresponding to the target keyword includes game content text, the operation management end may perform a game text processing operation on the target keyword according to the prompt information and processing urgency of the target keyword, and the game text processing operation may include one or more processing operations such as deleting, modifying and blocking the target keyword. When the original text corresponding to the target keyword includes user chat text, the game specification processing operation for the game user may include one or more processing operations such as prompting, warning or banning the game user. When the game user has not sent the target keyword, performing a user text operation on the target keyword may include prohibiting the game user from sending the target keyword. When the game user has already sent the target keyword, performing a user text operation on the target keyword may include performing a symbol replacement operation on the target keyword. This is not limited in this embodiment.

[0116] It can be seen that the implementation of this optional embodiment can determine the urgency of processing the target keyword according to the prompt information of the target keyword through the operation management end, and process the target keyword according to the original text corresponding to the target keyword, perform game standard processing on game users, improve the efficiency of early warning processing for sensitive words, and at the same time reduce the probability of inappropriate remarks during in-game player chats, standardize the chat behavior of in-game players, and improve the health of game operations.

[0117] Embodiment 2

[0118] See also Figure 3 , Figure 3 1 is a flowchart of a text recognition method applied to a text detection system disclosed in an embodiment of the present invention. Figure 3 The text recognition method applied to the text detection system described above can be applied to the text detection system, and the text detection system may include an intelligent server or intelligent platform for performing keyword and / or sensitive word recognition and detection on the text, and the intelligent server includes a local server or a cloud server, which is not limited in the embodiment of the present invention. Figure 3 As shown, the text recognition method applied to the text detection system may include the following operations:

[0119] 201. The text detection system obtains text information of the target game and extracts a keyword set from the text information.

[0120] 202. The text detection system obtains a preset sensitive word set, and the text detection system performs sensitive word recognition on each keyword in the keyword set according to the sensitive word set to obtain a first sub-recognition result corresponding to the keyword set.

[0121] In an embodiment of the present invention, optionally, the first sub-recognition result corresponding to the keyword set may include the same target keyword in the keyword set as in the sensitive word set, that is, the text detection system compares and identifies each keyword in the keyword set one by one according to the sensitive word set, and the present invention does not limit this.

[0122] 203. The text detection system determines keyword information of each keyword in the keyword set, and divides the keyword set according to the keyword information of each keyword to obtain at least two sub-keyword sets.

[0123] In an embodiment of the present invention, optionally, the keyword information of each keyword in the keyword set may include keyword meaning information of each keyword, a keyword score, and a string distance between each keyword. The keyword set may be divided according to the keyword information of each keyword to obtain at least two sub-keyword sets, each sub-keyword set includes multiple keywords, and the string distance between each keyword in the same sub-keyword set is less than a preset string distance threshold, that is, each keyword in the same sub-keyword set may be a keyword in the same sentence or a keyword in the same paragraph.

[0124] 204. The text detection system determines the semantic features corresponding to each sub-keyword set, and performs sensitive semantic recognition on the semantic features corresponding to each sub-keyword set according to the sensitive word set to obtain a second sub-recognition result corresponding to the keyword set.

[0125] In an embodiment of the present invention, optionally, the semantic features corresponding to each sub-keyword set can represent the semantic features of a combined word obtained by randomly combining multiple keywords in each sub-keyword set, and sensitive semantic recognition is performed on the semantic features corresponding to each sub-keyword set according to the sensitive word set, that is, it is determined whether the semantic features of the combined word obtained by randomly combining multiple keywords in each sub-keyword set express sensitive semantics, that is, two or more keywords themselves may not be sensitive words, but the semantics expressed after the two or more keywords are combined may be sensitive semantics, and a second sub-identification result corresponding to the keyword set is obtained, which is not limited by the present invention.

[0126] 205. The text detection system analyzes the sensitive word recognition result. When the sensitive word recognition result indicates that there is at least one target keyword in the keyword set that matches the sensitive word set, the text detection system generates prompt information for each target keyword and sends the prompt information for each target keyword to the operation management end of the target game.

[0127] In the embodiment of the present invention, for other descriptions of step 201 and step 205, please refer to the detailed description of step 101 and step 103 in the first embodiment, and the embodiment of the present invention will not be repeated.

[0128] It can be seen that the implementation Figure 3The described text recognition method for a text detection system can obtain text information of a target game through a text detection system, and extract a keyword set from the text information, obtain a preset sensitive word set, perform sensitive word recognition on each keyword in the keyword set according to the sensitive word set, obtain a first sub-recognition result corresponding to the keyword set, determine keyword information of each keyword in the keyword set, and divide the keyword set according to the keyword information of each keyword to obtain at least two sub-keyword sets, determine the semantic features corresponding to each sub-keyword set, and perform sensitive semantic recognition on the semantic features corresponding to each sub-keyword set according to the sensitive word set to obtain a second sub-recognition result corresponding to the keyword set, and can perform sensitive recognition on the keywords in the text one by one. And after the combination, semantic sensitivity recognition is performed to improve the comprehensiveness and accuracy of sensitive word recognition, and the sensitive word recognition results are analyzed. When the sensitive word recognition result indicates that there is at least one target keyword in the keyword set that matches the sensitive word set, the text detection system generates prompt information for each target keyword, and sends the prompt information for each target keyword to the operation and management end of the target game. The text detection system can identify sensitive words in the game text based on the text detection system, and provide sensitive word reminders, thereby improving the comprehensiveness and accuracy of identifying sensitive information, reducing the probability of sensitive words appearing in the game text during the game development process, and improving the health and safety of the game. At the same time, it reduces the probability of inappropriate remarks made by players during in-game chats, standardizes the chat behavior of players during in-game chats, and improves the health of game operations.

[0129] In an optional embodiment, the text detection system determining the keyword information of each keyword in the keyword set may include the following operations:

[0130] The text detection system determines target information of each keyword in the keyword set, the target information including keyword occurrence frequency information and keyword meaning information;

[0131] The text detection system calculates the keyword score corresponding to each keyword based on the target information corresponding to each keyword;

[0132] The text detection system calculates the string distance between each keyword in the keyword set;

[0133] The text detection system determines the keyword information of each keyword based on the keyword score corresponding to each keyword and the string distance between each keyword.

[0134] In this optional embodiment, optionally, the target information of each keyword in the keyword set may include keyword occurrence frequency information and keyword meaning information, and the sensitivity level of the keyword can be determined based on the keyword meaning information of the keyword, and then the keyword score of the keyword can be calculated based on the keyword occurrence frequency information, the sensitivity level and a preset formula of the keyword. Optionally, the higher the keyword occurrence frequency of the keyword, the higher the keyword score of the keyword can be, which is not limited in this embodiment.

[0135] In this optional embodiment, optionally, the text detection system calculates the string distance between each keyword in the keyword set. The string distance between keywords can reflect the degree of semantic association between keywords. When the string distance between multiple keywords is less than a preset string distance threshold, it can indicate that the multiple keywords are in the same sentence or the same paragraph, that is, the probability that the multiple keywords are in the same context is relatively high. The multiple keywords can be combined to determine the semantic features of the multiple keywords after combination. When the string distance between multiple keywords is greater than the preset string distance threshold, the context distance between the multiple keywords in the text is large, that is, the probability that the multiple keywords are in different contexts is relatively high. At this time, after combining the multiple keywords, it is determined that the semantic features of the multiple keywords after combination are highly likely to be inconsistent with the semantic logic. At this time, the degree of semantic association between the keywords can be ignored, and this embodiment does not make any limitation.

[0136] It can be seen that implementing this optional embodiment can determine the target information of each keyword in the keyword set, calculate the keyword score corresponding to each keyword according to the target information corresponding to each keyword, calculate the string distance between each keyword in the keyword set, and determine the keyword information of each keyword according to the keyword score corresponding to each keyword and the string distance between each keyword. It is possible to determine the keyword information through the keyword score and the string distance, improve the accuracy of determining the keyword information, and at the same time improve the rationality of dividing the keyword set through the keyword information.

[0137] In another optional embodiment, the text detection system divides the keyword set according to the keyword information of each keyword to obtain at least two sub-keyword sets, which may include the following operations:

[0138] The text detection system obtains a preset weight allocation rule, and determines the keyword score weight corresponding to each keyword and the string distance weight between each keyword according to the weight allocation rule;

[0139] The text detection system calculates the association parameters between each keyword based on the keyword score and keyword score weight corresponding to each keyword, the string distance between each keyword, and the string distance weight;

[0140] The text detection system divides the keyword set according to a preset association parameter threshold and an association parameter between each keyword to obtain at least two sub-keyword sets;

[0141] Each sub-keyword set includes at least two related keywords, and a correlation parameter between each related keyword in each sub-keyword set is greater than a correlation parameter threshold.

[0142] In this optional embodiment, optionally, the keyword score weight corresponding to each keyword and the string distance weight between each keyword can be determined according to the weight allocation rule, and the association parameters between each keyword can be calculated according to the keyword score corresponding to each keyword and the keyword score weight, the string distance between each keyword and the string distance weight.

[0143] In this optional embodiment, optionally, when calculating the association parameters between each keyword, the matching degree between the keyword scores corresponding to each keyword can be calculated, for example, the higher the matching degree between the keyword scores corresponding to two keywords, the higher the word meaning matching degree and word frequency matching degree between the two keywords; determine whether the string distance between each keyword is less than a preset string distance threshold, for example, when the string distance between two keywords is less than the preset string distance threshold, the higher the correlation degree between the two keywords, and then calculate the association parameters between each keyword based on the matching degree between the keyword scores corresponding to each keyword and whether the string distance between each keyword is less than the preset string distance threshold. Optionally, the string distance weight between each keyword in this text detection system can be much greater than the keyword score weight, that is, the string distance between each keyword has a greater impact on the association parameters between each keyword, which is not limited in this embodiment.

[0144] In this optional embodiment, optionally, the keyword set can be divided according to a preset association parameter threshold and the association parameters between each keyword to obtain at least two sub-keyword sets, each sub-keyword set includes at least two associated keywords, and the association parameters between each associated keyword in each sub-keyword set are greater than the association parameter threshold, that is, each associated keyword in the same sub-keyword set can be randomly combined to determine the semantic features of the combination, which is not limited in this embodiment.

[0145] It can be seen that the implementation of this optional embodiment can calculate the association parameters between each keyword based on the keyword score and the string distance, and then divide the keywords whose association parameters are greater than the association parameter threshold into the same sub-keyword set, thereby improving the degree of association between each associated keyword in the keyword set in terms of semantics and contextual text information, and can improve the rationality of combining each associated keyword in the keyword set, thereby improving the rationality and accuracy of sensitive semantic recognition through the semantic features corresponding to each sub-keyword set.

[0146] In yet another optional embodiment, the text detection system may determine the semantic features corresponding to each sub-keyword set by performing the following operations:

[0147] For each sub-keyword set, the text detection system determines the expanded word information of each associated keyword in the sub-keyword set according to the keyword meaning of each associated keyword in the sub-keyword set, wherein the expanded word meaning information includes synonymous expanded word information and / or antonymous expanded word information;

[0148] For each sub-keyword set, the text detection system determines contextual semantic information of each associated keyword in the sub-keyword set;

[0149] For each sub-keyword set, the text detection system determines a first coefficient corresponding to the expanded word information of each associated keyword in the sub-keyword set and a second coefficient corresponding to the contextual semantic information;

[0150] For each sub-keyword set, the text detection system randomly combines each associated keyword in the sub-keyword set to obtain at least one keyword combination, each keyword combination including at least two associated keywords;

[0151] For each sub-keyword set and each keyword combination, the text detection system determines the combination semantic feature corresponding to the keyword combination according to the expansion word information, the first coefficient, the context semantic information and the second coefficient corresponding to the keyword combination;

[0152] For each sub-keyword set, the text detection system determines the semantic features corresponding to the sub-keyword set according to the combined semantic features corresponding to each keyword combination in the sub-keyword set.

[0153] In this optional embodiment, optionally, the expanded word information of each associated keyword in the sub-keyword set may include synonymous expanded word information and / or antonymous expanded word information, that is, determining the synonymous semantics and / or antonymous semantics of the associated keyword in the current context, the text detection system determines the contextual semantic information of each associated keyword in the sub-keyword set, the text detection system determines the first coefficient corresponding to the expanded word information of each associated keyword in the sub-keyword set and the second coefficient corresponding to the contextual semantic information, in the text detection system, the second coefficient corresponding to the contextual semantic information is greater than the first coefficient corresponding to the expanded word information, that is, the contextual semantic information of the associated keyword has a greater impact on the combined semantic features than the expanded word information. For example, when a game user sends "You are awesome!" in a game, it may be an expression of praise to a teammate or opponent, but when the contextual semantic information indicates that the teammate or opponent is at a disadvantage or fails in the game, it may be an expression of sarcasm or blame.

[0154] In this optional embodiment, optionally, the text detection system randomly combines each associated keyword in the sub-keyword set to obtain at least one keyword combination, and each keyword combination may include at least two associated keywords. For each keyword combination, the text detection system determines the combined semantic features corresponding to the keyword combination based on the extended word information, the first coefficient, the contextual semantic information and the second coefficient corresponding to the keyword combination. For each sub-keyword set, the text detection system determines the semantic features corresponding to the sub-keyword set based on the combined semantic features corresponding to each keyword combination in the sub-keyword set, that is, the semantic features corresponding to the sub-keyword set may include multiple combined semantic features, which is not limited in this embodiment.

[0155] It can be seen that the implementation of this optional embodiment can determine the combined semantic features corresponding to the keyword combination by combining the expanded word information and contextual semantic information of the related keywords in the process of determining the semantic features corresponding to each sub-keyword set through the text detection system, comprehensively consider the semantics and context, and improve the accuracy and rationality of the determined combined semantic features.

[0156] In yet another optional embodiment, the text recognition method applied to the text detection system may further include the following operations:

[0157] The text detection system obtains a set of ignored words for the target game, and updates the set of sensitive words based on the set of ignored words;

[0158] Furthermore, the text detection system performs sensitive word recognition on each keyword in the keyword set according to the sensitive word set to obtain a first sub-recognition result corresponding to the keyword set, which may include the following operations:

[0159] The text detection system performs sensitive word recognition on the keyword set according to the sensitive word set and the ignored word set, and obtains a first target recognition result corresponding to the keyword set;

[0160] The text detection system selects a related word set corresponding to the first target recognition result from the keyword set, wherein the related word set includes at least one of a synonym, a synonym, and antonym corresponding to the first target recognition result;

[0161] The text detection system performs sensitive word recognition on the related word set according to the sensitive word set and the ignored word set, and obtains a second target recognition result corresponding to the keyword set;

[0162] The text detection system determines a first sub-recognition result corresponding to the keyword set according to the first target recognition result and the second target recognition result.

[0163] In this optional embodiment, optionally, the ignored word set for the target game can be configured by the game manager at the operation management end according to the needs of the target game, and the keyword set can be compared and identified one by one according to the sensitive word set and the ignored word set to obtain a first target recognition result corresponding to the keyword set. When the first target recognition result indicates that the keyword set includes at least one first sub-target keyword that matches the sensitive word set and the ignored word set, the related word set corresponding to the first target recognition result can be screened in the keyword set. The related word set can include at least one of the synonyms, antonyms and words corresponding to the first target recognition result, that is, the synonyms, antonyms and / or words corresponding to each first sub-target keyword are screened in the keyword set to obtain a related word set, which is not limited in this embodiment.

[0164] In this optional embodiment, optionally, sensitive word identification is performed on the related word set according to the sensitive word set and the ignored word set to obtain a second target identification result corresponding to the keyword set, that is, the second sub-target keyword in the related word set is identified in the sensitive word set and the ignored word set. For example, when the first target identification result indicates that word A in the keyword set is a sensitive word or an ignored word in the sensitive word set and the ignored word set, the synonyms, synonyms and / or antonyms corresponding to word A are screened in the keyword set to obtain a related word set, such as word B, word C and word D. Sensitive word identification is performed on the related word set according to the sensitive word set and the ignored word set to obtain the second target identification result corresponding to the keyword set. When the second target identification result indicates that word B and word D in the related word set are sensitive words or ignored words in the sensitive word set and the ignored word set, it is determined that the first sub-identification result corresponding to the keyword set includes word A, word B and word D. This is not limited in this embodiment.

[0165] It can be seen that implementing this optional embodiment can obtain a set of ignored words for the target game, and update the sensitive word set according to the ignored word set, perform sensitive word identification on the keyword set according to the sensitive word set and the ignored word set, and further identify the related word set based on the sensitive word identification result, thereby improving the accuracy and comprehensiveness of sensitive word identification, thereby reducing the probability of sensitive words appearing in the game text during the game development process, improving the health and safety of the game, and at the same time reducing the probability of inappropriate remarks during player chats in the game, standardizing the chat behavior of players in the game, and improving the health of game operations.

[0166] In yet another optional embodiment, when the sensitive word recognition result indicates that there is at least one target keyword in the keyword set that matches the sensitive word set, the text recognition method applied to the text detection system may further include the following operations:

[0167] The text detection system determines feature information of each keyword in the keyword set, where the feature information includes attribute label features and / or fine-grained features;

[0168] For each keyword, the text detection system classifies and stores the keyword into a preset feature word library based on the feature information of the keyword;

[0169] Furthermore, the text recognition method applied to the text detection system may further include the following operations:

[0170] When the text detection system receives the sensitive word recognition range and sensitive word configuration requirements sent by the operation management end, the text detection system configures the sensitive word set based on the feature word library according to the sensitive word recognition range and the sensitive word configuration requirements.

[0171] In this optional embodiment, optionally, the feature information of each keyword in the keyword set may include attribute label features and / or fine-grained features, and the keyword can be classified and stored in a preset feature vocabulary based on the feature information of the keyword. The feature vocabulary is continuously updated and improved with each new text recognition processing, gradually accumulating richer and more comprehensive vocabulary and feature information.

[0172] In this optional embodiment, optionally, the game administrator can set the sensitive word recognition range and sensitive word configuration requirements at the operation management end. The sensitive word recognition range can include global recognition and / or local recognition. For example, sensitive word recognition can be performed only on the game text during the game development or update process, and sensitive word recognition can be performed only on the game user chat text after the game is launched. The sensitive word configuration requirements may include sensitive word content requirements and / or sensitive word scope requirements. The sensitive word scope requirements can indicate that the recognition range of certain special sensitive words is limited, such as global recognition of a certain sensitive word or only recognition of a certain text part. When the text detection system receives the sensitive word recognition range and sensitive word configuration requirements sent by the operation management end, the text detection system configures the sensitive word set based on the feature word library according to the sensitive word recognition range and the sensitive word configuration requirements. This embodiment does not make any limitation.

[0173] It can be seen that the implementation of this optional embodiment can determine the feature information of each keyword in the keyword set, and classify and store the keyword into a preset feature word library based on the feature information of the keyword. The feature word library is continuously updated and improved with each new text recognition process, gradually accumulating richer and more comprehensive vocabulary and feature information, forming a self-learning closed loop, improving the ability to identify and classify sensitive words, and continuously improving the accuracy and efficiency of sensitive word recognition; when the text detection system receives the sensitive word recognition range and sensitive word configuration requirements sent by the operation management end, the text detection system configures the sensitive word set based on the feature word library according to the sensitive word recognition range and sensitive word configuration requirements, which can enable game managers to customize the configuration of sensitive word content and recognition range, improve the flexibility of sensitive word recognition, and improve user experience.

[0174] Embodiment 3

[0175] See also Figure 4 , Figure 4 is a structural diagram of a text detection system disclosed in an embodiment of the present invention. Figure 4 The text detection system described may include an intelligent server or intelligent platform for detecting keywords and / or sensitive words in text, and the intelligent server may include a local server or a cloud server, which is not limited in the embodiment of the present invention. Figure 4 As shown, the text detection system may include:

[0176] An acquisition module 301 is used to acquire text information of a target game and extract a keyword set from the text information, wherein the text information includes game content text information and / or user chat text information;

[0177] The acquisition module 301 is further used to acquire a preset sensitive word set, and perform sensitive word recognition on the keyword set according to the sensitive word set to obtain a sensitive word recognition result corresponding to the keyword set;

[0178] The analysis module 302 is used to analyze the sensitive word recognition result. When the sensitive word recognition result indicates that there is at least one target keyword in the keyword set that matches the sensitive word set, prompt information for each target keyword is generated and the prompt information for each target keyword is sent to the operation management end of the target game.

[0179] It can be seen that the implementation Figure 4 The described text detection system can obtain text information of the target game through the text detection system, extract a keyword set from the text information, obtain a preset sensitive word set, and perform sensitive word recognition on the keyword set according to the sensitive word set to obtain a sensitive word recognition result corresponding to the keyword set, and analyze the sensitive word recognition result. When the sensitive word recognition result indicates that there is at least one target keyword in the keyword set that matches the sensitive word set, the text detection system generates prompt information for each target keyword, and sends the prompt information for each target keyword to the operation and management end of the target game. It can identify sensitive words in the game text based on the text detection system and provide sensitive word reminders, thereby improving the comprehensiveness and accuracy of identifying sensitive information, reducing the probability of sensitive words appearing in the game text during the game development process, improving the health and safety of the game, reducing the cost of game development and maintenance, and at the same time reducing the probability of inappropriate remarks made by players during in-game chats, standardizing the chat behavior of players during in-game chats, and improving the health of game operations.

[0180] In an optional embodiment, if Figure 5 As shown, the sensitive word recognition result includes a first sub-recognition result and a second sub-recognition result;

[0181] The acquisition module 301 performs sensitive word recognition on the keyword set according to the sensitive word set, and obtains the sensitive word recognition result corresponding to the keyword set in the following specific manners:

[0182] Perform sensitive word recognition on each keyword in the keyword set according to the sensitive word set to obtain a first sub-recognition result corresponding to the keyword set;

[0183] Determine keyword information of each keyword in the keyword set, and divide the keyword set according to the keyword information of each keyword to obtain at least two sub-keyword sets;

[0184] The semantic features corresponding to each sub-keyword set are determined, and sensitive semantic recognition is performed on the semantic features corresponding to each sub-keyword set according to the sensitive word set to obtain a second sub-recognition result corresponding to the keyword set.

[0185] It can be seen that the implementation Figure 5The described text detection system can obtain the text information of the target game through the text detection system, and extract the keyword set from the text information, obtain the preset sensitive word set, perform sensitive word recognition on each keyword in the keyword set according to the sensitive word set, obtain the first sub-recognition result corresponding to the keyword set, determine the keyword information of each keyword in the keyword set, and divide the keyword set according to the keyword information of each keyword to obtain at least two sub-keyword sets, determine the semantic features corresponding to each sub-keyword set, and perform sensitive semantic recognition on the semantic features corresponding to each sub-keyword set according to the sensitive word set to obtain the second sub-recognition result corresponding to the keyword set, which can perform sensitive recognition on the keywords in the text one by one and combine them. Perform semantic sensitivity recognition, improve the comprehensiveness and accuracy of sensitive word recognition, analyze the sensitive word recognition results, and when the sensitive word recognition results indicate that there is at least one target keyword in the keyword set that matches the sensitive word set, the text detection system generates prompt information for each target keyword, and sends the prompt information for each target keyword to the operation and management end of the target game. The text detection system can identify sensitive words in the game text based on the text detection system, and provide sensitive word reminders, thereby improving the comprehensiveness and accuracy of identifying sensitive information, reducing the probability of sensitive words appearing in the game text during the game development process, improving the health and safety of the game, and at the same time reducing the probability of inappropriate remarks by players chatting in the game, standardizing the chat behavior of players in the game, and improving the health of game operations.

[0186] In another optional embodiment, Figure 5 As shown, the specific manner in which the acquisition module 301 determines the keyword information of each keyword in the keyword set includes:

[0187] Determine target information for each keyword in the keyword set, the target information including keyword occurrence frequency information and keyword meaning information;

[0188] Calculate the keyword score corresponding to each keyword based on the target information corresponding to each keyword;

[0189] Calculate the string distance between each keyword in the keyword set;

[0190] The keyword information of each keyword is determined according to the keyword score corresponding to each keyword and the character string distance between each keyword.

[0191] It can be seen that the implementation Figure 5The described text detection system can determine the target information of each keyword in the keyword set, calculate the keyword score corresponding to each keyword based on the target information corresponding to each keyword, calculate the string distance between each keyword in the keyword set, and determine the keyword information of each keyword based on the keyword score corresponding to each keyword and the string distance between each keyword. It can determine the keyword information through the keyword score and the string distance, thereby improving the accuracy of determining the keyword information and at the same time improving the rationality of dividing the keyword set through the keyword information.

[0192] In yet another optional embodiment, Figure 5 As shown, the specific manner in which the acquisition module 301 divides the keyword set according to the keyword information of each keyword to obtain at least two sub-keyword sets includes:

[0193] Obtaining a preset weight allocation rule, and determining the keyword score weight corresponding to each keyword and the string distance weight between each keyword according to the weight allocation rule;

[0194] Calculate the association parameters between each keyword based on the keyword score and keyword score weight corresponding to each keyword, the string distance between each keyword and the string distance weight;

[0195] Dividing the keyword set according to a preset association parameter threshold and an association parameter between each keyword to obtain at least two sub-keyword sets;

[0196] Each sub-keyword set includes at least two related keywords, and a correlation parameter between each related keyword in each sub-keyword set is greater than a correlation parameter threshold.

[0197] It can be seen that the implementation Figure 5 The described text detection system can calculate the association parameters between each keyword based on the keyword score and the string distance, and then divide the keywords whose association parameters are greater than the association parameter threshold into the same sub-keyword set, thereby improving the degree of association between each associated keyword in the keyword set in terms of semantics and contextual text information, and can improve the rationality of combining each associated keyword in the keyword set, thereby improving the rationality and accuracy of sensitive semantic recognition through the semantic features corresponding to each sub-keyword set.

[0198] In yet another optional embodiment, Figure 5 As shown, the specific manner in which the acquisition module 301 determines the semantic features corresponding to each sub-keyword set includes:

[0199] For each sub-keyword set, according to the keyword meaning of each associated keyword in the sub-keyword set, determine the expanded word information of each associated keyword in the sub-keyword set, the expanded word meaning information including synonymous expanded word information and / or antonymous expanded word information;

[0200] For each sub-keyword set, determining contextual semantic information of each associated keyword in the sub-keyword set;

[0201] For each sub-keyword set, determining a first coefficient corresponding to the expanded word information of each associated keyword in the sub-keyword set and a second coefficient corresponding to the contextual semantic information;

[0202] For each sub-keyword set, randomly combine each associated keyword in the sub-keyword set to obtain at least one keyword combination, each keyword combination including at least two associated keywords;

[0203] For each sub-keyword set and each keyword combination, determine the combination semantic feature corresponding to the keyword combination according to the expansion word information, the first coefficient, the context semantic information and the second coefficient corresponding to the keyword combination;

[0204] For each sub-keyword set, the semantic feature corresponding to the sub-keyword set is determined according to the combined semantic feature corresponding to each keyword combination in the sub-keyword set.

[0205] It can be seen that the implementation Figure 5 The described text detection system can, in the process of determining the semantic features corresponding to each sub-keyword set through the text detection system, combine the expanded word information and contextual semantic information of the associated keywords to determine the combined semantic features corresponding to the keyword combination, comprehensively consider the semantics and context, and improve the accuracy and rationality of the determined combined semantic features.

[0206] In yet another optional embodiment, Figure 5 As shown, the acquisition module 301 is also used to acquire an ignored word set for the target game, and update the sensitive word set according to the ignored word set;

[0207] Furthermore, the acquisition module 301 performs sensitive word recognition on each keyword in the keyword set according to the sensitive word set, and obtains the first sub-recognition result corresponding to the keyword set in a specific manner including:

[0208] Performing sensitive word recognition on the keyword set according to the sensitive word set and the ignored word set to obtain a first target recognition result corresponding to the keyword set;

[0209] Filtering a related word set corresponding to the first target recognition result in the keyword set, wherein the related word set includes at least one of a synonym, a synonym, and antonym corresponding to the first target recognition result;

[0210] Perform sensitive word recognition on the related word set according to the sensitive word set and the ignored word set to obtain a second target recognition result corresponding to the keyword set;

[0211] A first sub-recognition result corresponding to the keyword set is determined according to the first target recognition result and the second target recognition result.

[0212] It can be seen that the implementation Figure 5 The described text detection system can obtain a set of ignored words for a target game, and update a set of sensitive words based on the set of ignored words, identify sensitive words for a set of keywords based on the set of sensitive words and the set of ignored words, and further identify related word sets based on the sensitive word identification results, thereby improving the accuracy and comprehensiveness of sensitive word identification, thereby reducing the probability of sensitive words appearing in game texts during game development, improving the health and safety of games, and at the same time reducing the probability of inappropriate remarks made by players during in-game chats, standardizing the chat behavior of players within the game, and improving the health of game operations.

[0213] In yet another optional embodiment, Figure 5 As shown, the text detection system may also include:

[0214] A determination module 303 is configured to determine feature information of each keyword in the keyword set when the sensitive word recognition result indicates that there is at least one target keyword matching the sensitive word set in the keyword set, wherein the feature information includes attribute label features and / or fine-grained features;

[0215] The storage module 304 is used to classify and store each keyword into a preset feature word library according to the feature information of the keyword;

[0216] And, the text detection system may also include:

[0217] The configuration module 305 is used to configure a sensitive word set based on a feature word library according to the sensitive word recognition range and the sensitive word configuration requirements when receiving the sensitive word recognition range and the sensitive word configuration requirements sent by the operation management terminal.

[0218] It can be seen that the implementation Figure 5The described text detection system can determine the feature information of each keyword in the keyword set, and classify and store the keyword into a preset feature word library according to the feature information of the keyword. The feature word library is continuously updated and improved with each new text recognition process, gradually accumulating richer and more comprehensive vocabulary and feature information, forming a self-learning closed loop, improving the sensitive word recognition and classification capabilities, and continuously improving the accuracy and efficiency of sensitive word recognition; when the text detection system receives the sensitive word recognition range and sensitive word configuration requirements sent by the operation management end, the text detection system configures the sensitive word set based on the feature word library according to the sensitive word recognition range and sensitive word configuration requirements, which can enable game managers to customize the sensitive word content and recognition range, improve the flexibility of sensitive word recognition, and improve user experience.

[0219] Embodiment 4

[0220] See also Figure 6 , Figure 6 Schematic diagram of a text recognition device applied to a text detection system disclosed in an embodiment of the present invention. Figure 6 As shown, the text recognition device applied to the text detection system may include:

[0221] A memory 401 storing executable program codes;

[0222] a processor 402 coupled to the memory 401;

[0223] The processor 402 calls the executable program code stored in the memory 401 to execute the steps of the text recognition method applied to the text detection system described in the first embodiment of the present invention or the second embodiment of the present invention.

[0224] Embodiment 5

[0225] An embodiment of the present invention discloses a computer storage medium, which stores computer instructions. When the computer instructions are called, they are used to execute the steps of the text recognition method applied to the text detection system described in the first embodiment of the present invention or the second embodiment of the present invention.

[0226] Embodiment 6

[0227] An embodiment of the present invention discloses a computer program product, which includes a non-transitory computer-readable storage medium storing a computer program, and the computer program is operable to enable a computer to execute the steps of the text recognition method applied to a text detection system described in Embodiment 1 or Embodiment 2.

[0228] The device embodiments described above are only illustrative, wherein the modules described as separate components may or may not be physically separated, and the components displayed as modules may or may not be physical modules, i.e., they may be located in one place, or they may be distributed on multiple network modules. Some or all of the modules may be selected according to actual needs to achieve the purpose of the scheme of this embodiment. Those of ordinary skill in the art may understand and implement it without creative work.

[0229] Through the specific description of the above embodiments, those skilled in the art can clearly understand that each implementation method can be implemented by means of software plus a necessary general hardware platform, and of course, it can also be implemented by hardware. Based on such an understanding, the above technical solution can be essentially or partly contributed to the prior art in the form of a software product, and the computer software product can be stored in a computer-readable storage medium, and the storage medium includes a read-only memory (ROM), a random access memory (RAM), a programmable read-only memory (PROM), an erasable programmable read-only memory (EPROM), a one-time programmable read-only memory (OTPROM), an electronically erasable rewritable read-only memory (EEPROM), a compact disc (CD-ROM) or other optical disc storage, magnetic disk storage, magnetic tape storage, or any other computer-readable medium that can be used to carry or store data.

[0230] Finally, it should be noted that the text recognition method and text detection system applied to a text detection system disclosed in the embodiments of the present invention are only preferred embodiments of the present invention, which are only used to illustrate the technical solution of the present invention, rather than to limit it. Although the present invention has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that the technical solutions described in the aforementioned embodiments can still be modified, or some of the technical features therein can be replaced by equivalents. However, these modifications or replacements do not deviate the essence of the corresponding technical solutions from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. A text recognition method applied to a text detection system, characterized in that: The method comprises: The text detection system obtains text information of the target game and extracts a keyword set from the text information, wherein the text information includes game content text information and / or user chat text information; The text detection system obtains a preset sensitive word set, and performs sensitive word recognition on the keyword set according to the sensitive word set to obtain a sensitive word recognition result corresponding to the keyword set; The text detection system analyzes the sensitive word recognition result. When the sensitive word recognition result indicates that there is at least one target keyword in the keyword set that matches the sensitive word set, the text detection system generates prompt information for each of the target keywords and sends the prompt information for each of the target keywords to the operation and management end of the target game.

2. The text recognition method applied to a text detection system according to claim 1, characterized in that: The sensitive word recognition result includes a first sub-recognition result and a second sub-recognition result; The text detection system performs sensitive word recognition on the keyword set according to the sensitive word set to obtain a sensitive word recognition result corresponding to the keyword set, including: The text detection system performs sensitive word recognition on each keyword in the keyword set according to the sensitive word set to obtain the first sub-recognition result corresponding to the keyword set; The text detection system determines keyword information of each keyword in the keyword set, and divides the keyword set according to the keyword information of each keyword to obtain at least two sub-keyword sets; The text detection system determines the semantic features corresponding to each of the sub-keyword sets, and performs sensitive semantic recognition on the semantic features corresponding to each of the sub-keyword sets according to the sensitive word set to obtain the second sub-recognition result corresponding to the keyword set.

3. The text recognition method applied to a text detection system according to claim 2, characterized in that: The text detection system determines keyword information of each keyword in the keyword set, including: The text detection system determines target information of each keyword in the keyword set, wherein the target information includes keyword occurrence frequency information and keyword meaning information; The text detection system calculates a keyword score corresponding to each keyword according to the target information corresponding to each keyword; The text detection system calculates the string distance between each of the keywords in the keyword set; The text detection system determines the keyword information of each keyword according to the keyword score corresponding to each keyword and the character string distance between each keyword.

4. The text recognition method applied to a text detection system according to claim 3, characterized in that: The text detection system divides the keyword set according to the keyword information of each keyword to obtain at least two sub-keyword sets, including: The text detection system obtains a preset weight allocation rule, and determines a keyword scoring weight corresponding to each of the keywords and a string distance weight between each of the keywords according to the weight allocation rule; The text detection system calculates the association parameter between each of the keywords according to the keyword score and the keyword score weight corresponding to each of the keywords, the string distance between each of the keywords and the string distance weight; The text detection system divides the keyword set according to a preset association parameter threshold and an association parameter between each of the keywords to obtain at least two sub-keyword sets; Each of the sub-keyword sets includes at least two associated keywords, and an associated parameter between each of the associated keywords in each of the sub-keyword sets is greater than the associated parameter threshold.

5. The text recognition method applied to a text detection system according to any one of claims 2 to 4, characterized in that: The text detection system determines the semantic features corresponding to each of the sub-keyword sets, including: For each of the sub-keyword sets, the text detection system determines the expanded word information of each of the associated keywords in the sub-keyword set according to the keyword meaning of each associated keyword in the sub-keyword set, wherein the expanded word meaning information includes synonymous expanded word information and / or antonymous expanded word information; For each of the sub-keyword sets, the text detection system determines contextual semantic information of each of the associated keywords in the sub-keyword set; For each of the sub-keyword sets, the text detection system determines a first coefficient corresponding to the expanded word information and a second coefficient corresponding to the contextual semantic information of each of the associated keywords in the sub-keyword set; For each of the sub-keyword sets, the text detection system randomly combines each of the associated keywords in the sub-keyword set to obtain at least one keyword combination, each of which includes at least two associated keywords; For each of the sub-keyword sets and each of the keyword combinations, the text detection system determines the combined semantic features corresponding to the keyword combination according to the expanded word information corresponding to the keyword combination, the first coefficient, the contextual semantic information and the second coefficient; For each of the sub-keyword sets, the text detection system determines the semantic features corresponding to the sub-keyword set according to the combined semantic features corresponding to each of the keyword combinations in the sub-keyword set.

6. The text recognition method applied to a text detection system according to any one of claims 2 to 4, characterized in that: The method further comprises: The text detection system obtains a set of ignored words for the target game, and updates the set of sensitive words according to the set of ignored words; Furthermore, the text detection system performs sensitive word recognition on each keyword in the keyword set according to the sensitive word set to obtain the first sub-recognition result corresponding to the keyword set, including: The text detection system performs sensitive word recognition on the keyword set according to the sensitive word set and the ignored word set to obtain a first target recognition result corresponding to the keyword set; The text detection system selects a related word set corresponding to the first target recognition result from the keyword set, wherein the related word set includes at least one of a synonym, a synonym, and antonym corresponding to the first target recognition result; The text detection system performs sensitive word recognition on the related word set according to the sensitive word set and the ignored word set to obtain a second target recognition result corresponding to the keyword set; The text detection system determines the first sub-recognition result corresponding to the keyword set according to the first target recognition result and the second target recognition result.

7. The text recognition method applied to a text detection system according to claim 6, characterized in that: When the sensitive word recognition result indicates that there is at least one target keyword in the keyword set that matches the sensitive word set, the method further includes: The text detection system determines feature information of each keyword in the keyword set, wherein the feature information includes attribute label features and / or fine-grained features; For each of the keywords, the text detection system classifies and stores the keyword into a preset feature word library according to feature information of the keyword; And, the method further comprises: When the text detection system receives the sensitive word recognition range and the sensitive word configuration requirements sent by the operation management end, the text detection system configures a sensitive word set based on the feature word library according to the sensitive word recognition range and the sensitive word configuration requirements.

8. A text detection system, characterized in that: The text detection system comprises: An acquisition module, used to acquire text information of a target game and extract a keyword set from the text information, wherein the text information includes game content text information and / or user chat text information; The acquisition module is further used to acquire a preset sensitive word set, and perform sensitive word recognition on the keyword set according to the sensitive word set to obtain a sensitive word recognition result corresponding to the keyword set; The analysis module is used to analyze the sensitive word recognition result. When the sensitive word recognition result indicates that there is at least one target keyword in the keyword set that matches the sensitive word set, prompt information for each target keyword is generated, and the prompt information for each target keyword is sent to the operation management end of the target game.

9. A text recognition device applied to a text detection system, characterized in that: The device comprises: A memory storing executable program code; a processor coupled to the memory; The processor calls the executable program code stored in the memory to execute the text recognition method applied to a text detection system as described in any one of claims 1 to 7.

10. A computer storage medium, characterized in that: The computer storage medium stores computer instructions, and when the computer instructions are called, they are used to execute the text recognition method applied to the text detection system as described in any one of claims 1-7.

Images