Conversation interaction control method and device and electronic equipment

Abstract

The invention provides a dialogue interaction control method and device and electronic equipment, and the method comprises the steps: responding to a dialogue triggering instruction, and starting a current dialogue between a first user and a dialogue object; wherein a target state is preset for the dialogue object; the dialogue object is controlled by the dialogue robot; receiving a first dialogue content sent by a first user, and generating an evaluation parameter corresponding to the first dialogue content; wherein the evaluation parameter is used for indicating a state distance between an object state of the dialogue object for the first dialogue content and a target state; and generating reply content corresponding to the first dialogue content, and sending the reply content and the evaluation parameter to the first user. According to the method, the user can be guided to send the subsequent dialogue content through the evaluation parameters and the target state, so that the user and the dialogue robot can carry out targeted dialogue interaction, and the interactivity and dialogue experience of the user and the dialogue robot are improved.

Description

Technical Field

[0001] The present invention relates to the field of game technology, and in particular to a method, device and electronic equipment for interactive control of dialogue. Background Art

[0002] In the related art, a real user can have a conversation with a chatbot. After the real user sends a conversation, the chatbot generates a reply based on predefined rules and keyword matching, or retrieves the reply from a predefined answer library. The chatbot's reply content has limited flexibility and scalability, and can only have simple small talk with real users, resulting in poor interactivity between the user and the chatbot. Summary of the invention

[0003] In view of this, an object of the present invention is to provide a method, device and electronic device for interactive control of dialogue, so as to enhance the interactivity between users and dialogue robots.

[0004] In a first aspect, an embodiment of the present invention provides a method for interactive control of a conversation, the method comprising: in response to a conversation trigger instruction, starting a current conversation between a first user and a conversation object; wherein the conversation object is preset with a target state; the conversation object is controlled by a conversation robot; receiving a first conversation content sent by the first user, and generating an evaluation parameter corresponding to the first conversation content; wherein the evaluation parameter is used to indicate: the state distance between the object state of the conversation object with respect to the first conversation content and the target state; generating reply content corresponding to the first conversation content, and sending the reply content and the evaluation parameter to the first user.

[0005] In a second aspect, an embodiment of the present invention further provides an interactive control device for a conversation, the device comprising: a first start module, for starting a current conversation between a first user and a conversation object in response to a conversation trigger instruction; wherein the conversation object is preset with a target state; the conversation object is controlled by a conversation robot; a first generation module, for receiving a first conversation content sent by the first user, and generating an evaluation parameter corresponding to the first conversation content; wherein the evaluation parameter is used to indicate: the state distance between the object state of the conversation object for the first conversation content and the target state; a first sending module, for generating a reply content corresponding to the first conversation content, and sending the reply content and the evaluation parameter to the first user.

[0006] In a third aspect, an embodiment of the present invention provides an electronic device, including a processor and a memory, wherein the memory stores machine executable instructions that can be executed by the processor, and the processor executes the machine executable instructions to implement the above-mentioned interactive control method of the dialogue.

[0007] In a fourth aspect, an embodiment of the present invention provides a machine-readable storage medium, which stores machine-executable instructions. When the machine-executable instructions are called and executed by a processor, the machine-executable instructions prompt the processor to implement the interactive control method of the above-mentioned dialogue.

[0008] The embodiments of the present invention bring the following beneficial effects:

[0009] The interactive control method, device and electronic device of the above-mentioned dialogue, the method includes: responding to a dialogue trigger instruction, starting the current dialogue between the first user and the dialogue object; wherein the dialogue object is preset with a target state; the dialogue object is controlled by a dialogue robot; receiving the first dialogue content sent by the first user, and generating an evaluation parameter corresponding to the first dialogue content; wherein the evaluation parameter is used to indicate: the state distance between the object state of the dialogue object for the first dialogue content and the target state; generating reply content corresponding to the first dialogue content, and sending the reply content and the evaluation parameter to the first user.

[0010] In this method, the dialogue object is preset with a target state. After the user sends a dialogue content, an evaluation parameter corresponding to the dialogue content is generated to indicate the state distance between the object state of the dialogue object with respect to the dialogue content and the target state, and a reply content corresponding to the dialogue content is generated. Through the evaluation parameters and the target state, the user can be guided to send subsequent dialogue content, so that the user and the dialogue robot can have a purposeful dialogue interaction, thereby improving the interactivity and dialogue experience between the user and the dialogue robot.

[0011] Other features and advantages of the present invention will be described in the following description, and partly become apparent from the description, or understood by practicing the present invention. The purpose and other advantages of the present invention are realized and obtained by the structures particularly pointed out in the description, claims and drawings.

[0012] In order to make the above-mentioned objects, features and advantages of the present invention more obvious and easy to understand, preferred embodiments are given below and described in detail with reference to the accompanying drawings. BRIEF DESCRIPTION OF THE DRAWINGS

[0013] In order to more clearly illustrate the specific implementation methods of the present invention or the technical solutions in the prior art, the drawings required for use in the specific implementation methods or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are some implementation methods of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without paying any creative work.

[0014] Figure 1 A flowchart of a method for interactive control of a dialogue provided by an embodiment of the present invention;

[0015] Figure 2A schematic diagram of a current conversation between a first user and a conversation partner provided by an embodiment of the present invention;

[0016] Figure 3 Another schematic diagram of a current conversation between a first user and a conversation partner provided by an embodiment of the present invention;

[0017] Figure 4 A schematic diagram of an interactive control device for a dialogue provided by an embodiment of the present invention;

[0018] Figure 5 A schematic diagram of the structure of an electronic device provided by an embodiment of the present invention. DETAILED DESCRIPTION

[0019] In order to make the purpose, technical solution and advantages of the embodiments of the present invention clearer, the technical solution of the present invention will be clearly and completely described below in conjunction with the accompanying drawings. Obviously, the described embodiments are 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 those skilled in the art without creative work are within the scope of protection of the present invention.

[0020] First, the key terms involved in this embodiment are explained:

[0021] 1. Large Language Model;

[0022] A large language model is a language model consisting of an artificial neural network with many parameters (usually billions of weights or more), which is trained on a large amount of unlabeled text using self-supervised learning or semi-supervised learning, and can understand and generate human language. It is a model based on deep learning technology, which mainly understands and masters the grammar, semantics, and contextual relationships of a language by learning from a large amount of text data. The large language model has powerful text understanding and generation capabilities and can be used to handle various natural language processing tasks, such as machine translation, text summarization, question-answering systems, etc.

[0023] 2. NPC (Non-Player Character);

[0024] Game NPCs refer to non-player characters that play a specific role in the game and can interact with players according to the rules and situations preset by the program. In many games, NPCs play a key role, such as providing tasks, trading items, providing information, etc.

[0025] 3. Prompt;

[0026] Prompt can be understood as a way to start a machine learning model. It is a piece of text or sentence that guides the machine learning model to generate output of a specific type, topic or format. In the field of natural language processing, prompts usually consist of a question or task description, such as "write me an article about artificial intelligence", "translate this English sentence into French", and so on.

[0027] In the related art, the conversations that users can have with chatbots usually only allow players to have simple small talk with the chatbots. When users interact with the chatbots, the flexibility and scalability of the chatbots' reply content are limited. The chatbots exist only as a small talk tool, and users cannot get an intelligent and strategic gaming experience from the interaction process. This form of interaction is too single and lacks an in-depth interactive experience.

[0028] Based on the above, an embodiment of the present invention provides an interactive control method, device and electronic device for dialogue. This technology can be applied to character dialogues in game scenes, and can also be applied to dialogues between virtual objects, virtual characters, etc. in other virtual scenes.

[0029] In an optional implementation, various cloud applications can be run under the cloud interaction system, such as cloud games. Taking cloud games as an example, cloud games refer to a game mode based on cloud computing. In the operation mode of cloud games, the operating body of the game program and the main body of the game screen presentation are separated. The storage and operation of the interactive control method of the dialogue are completed on the cloud game server. The role of the client device is used for receiving and sending data and presenting the game screen. For example, the client device can be a display device with data transmission function close to the user side, such as a mobile terminal, a TV, a computer, a handheld computer, etc.; but the cloud game server in the cloud is used for information processing. When playing the game, the player operates the client device to send an operation instruction to the cloud game server. The cloud game server runs the game according to the operation instruction, encodes and compresses the game screen and other data, and returns it to the client device through the network. Finally, the client device decodes and outputs the game screen.

[0030] In an optional embodiment, taking a game as an example, a local terminal device stores a game program and is used to present a game screen. The local terminal device is used to interact with the player through a graphical user interface, that is, the game program is downloaded and installed by an electronic device and run conventionally. The local terminal device may provide the graphical user interface to the player in a variety of ways, for example, it may be rendered and displayed on a display screen of the terminal, or provided to the player through a holographic projection. For example, the local terminal device may include a display screen and a processor, the display screen is used to present a graphical user interface, the graphical user interface includes a game screen, and the processor is used to run the game, generate a graphical user interface, and control the display of the graphical user interface on the display screen.

[0031] To facilitate understanding of this embodiment, a method for interactive control of a conversation disclosed in an embodiment of the present invention is first introduced in detail. A graphical user interface can be provided through a terminal device, wherein the terminal device can be the local terminal device mentioned above, or a client device in the cloud interactive system mentioned above. A conversation window can be displayed in the graphical user interface. The conversation window displays the conversation content sent by the user, as well as the generated evaluation parameters, reply content, etc.

[0032] like Figure 1 As shown, the method comprises the following steps:

[0033] Step S102: In response to the dialogue triggering instruction, the current dialogue between the first user and the dialogue object is started; wherein the dialogue object is preset with a target state; and the dialogue object is controlled by the dialogue robot;

[0034] The above-mentioned dialogue object is the object with which the first user conducts a dialogue, and may be a virtual character, a virtual animal, a cartoon character, etc. The above-mentioned dialogue triggering instruction is an instruction for starting the current dialogue between the first user and the dialogue object. Specifically, the dialogue triggering instruction may be actively triggered by the first user, for example, the first user actively sends a dialogue message to the dialogue object to trigger the dialogue with the dialogue object.

[0035] The dialogue trigger instruction may be triggered based on a preset condition, such as recognizing that the game character controlled by the first user in the game scene is within a preset distance range from the object model of the dialogue object, and controlling the dialogue object to actively send a dialogue message to the first user. The dialogue object may be a virtual object that is not controlled by the user, such as an NPC in the game scene.

[0036] The dialogue object is preset with a target state, which can be understood as the object state that the dialogue object needs to achieve when the first user realizes the preset dialogue intention. The target state can be the target emotional state of the dialogue object after the emotional state changes, or the target behavior of the dialogue object after the behavior changes.

[0037] For example, if the first user's conversation intention with the dialogue object is to persuade the dialogue object to read a book, then the target state is that after one or more rounds of conversation, the dialogue object agrees to read a book. For another example: if the first user's conversation intention with the dialogue object is to persuade the dialogue object to get rid of pessimism, then the target state is that after one or more rounds of conversation, the dialogue object is in a positive emotional state.

[0038] The above-mentioned dialogue object is controlled by a dialogue robot, which can be built based on a neural network model. The neural network model has powerful text understanding and generation capabilities and can be used to handle various natural language processing tasks, such as machine translation, text summarization, question-answering systems, etc. The neural network model included in the dialogue robot also needs to be pre-trained to enable it to have the ability to interact with the first user in real-time intelligent dialogue, and after obtaining the dialogue content, it can also control and update the emotional state, behavior content, etc. of the dialogue object according to the pre-set game rules, and output the reply content corresponding to the dialogue content and the evaluation parameters for the dialogue content.

[0039] The evaluation parameter is used to indicate the state distance between the object state of the dialogue object and the target state for the dialogue content. In other words, the evaluation parameter can be used to indicate the distance between the emotional state of the dialogue object after receiving the dialogue content or the behavior content to be performed by the dialogue object and the preset target state.

[0040] For example, the evaluation parameter can be described by a score value under a preset score system. For example, if the score corresponding to the preset target state is 100 points, the evaluation parameter value of the evaluation parameter can be a score value not greater than 100 points. The greater the difference between the evaluation parameter value of the evaluation parameter and 100 points, the greater the gap between the object state of the dialogue object for the dialogue content and the target state.

[0041] Here, the conversation robot needs to be trained through prompt words, so that it can understand the rules of the game, control the conversation partner to play the corresponding role and interact in the game.

[0042] Specifically, firstly, a prompt word corresponding to the dialogue object is created. The prompt word includes the object setting information of the dialogue object, the game rules, etc. In one mode, the object setting information of the dialogue object may include: the identity information of the dialogue object, the character information of the dialogue object, the skill information of the dialogue object, etc. The game rules may include: the format information of the reply content, the pre-set alternative strategy that can control the change of the object state of the dialogue object, the initial evaluation parameter, the evaluation standard of the evaluation parameter, the end condition of the dialogue, the dialogue content constraint information, etc.

[0043] Among them, the alternative strategy refers to a pre-set dialogue strategy that can control the change of the object state of the dialogue object. When the dialogue robot recognizes that the user content contains the alternative strategy, it can control the object state of the dialogue object to undergo a specified change. The evaluation criteria of the evaluation parameters are used to determine the evaluation parameters of the dialogue content sent by the user each time. The evaluation criteria of the evaluation parameters include: the corresponding relationship between the alternative strategy and the change value of the evaluation parameter, and the change criteria of the change value of the evaluation parameter under other dialogue strategies.

[0044] For example, the evaluation criteria for the evaluation parameter may be: if the conversation content includes alternative strategy 1, the evaluation parameter change value decreases by 5 points; if the conversation content includes alternative strategy 2, the evaluation parameter change value increases by 3 points; if the conversation content is content sent repeatedly by the user, the evaluation parameter change value remains unchanged.

[0045] The initial evaluation parameter refers to the initial value of the evaluation parameter at the beginning of the current conversation. After the user sends the first conversation content, the evaluation parameter of the conversation content is obtained by adding or subtracting the evaluation parameter change value corresponding to the conversation content based on the initial evaluation parameter. Thereafter, the evaluation parameter corresponding to each conversation content sent by the user is obtained by adding or subtracting the evaluation parameter change value corresponding to the current conversation content based on the evaluation parameter of the previous conversation content.

[0046] The end condition of the conversation can be that the evaluation parameter reaches the preset parameter threshold, at which time the dialogue object reaches the target state and the game is successfully completed. The end condition of the conversation can also be that the generation round of the reply content in the current conversation reaches the preset round threshold, and the dialogue object has not yet reached the target state. At this time, the game fails to pass the level, or the current conversation content triggers a conversation effect that is opposite to the preset conversation intention, resulting in the failure of the game, and the end of the current conversation, etc. Conversation content constraint information refers to topic information that the conversation robot can explicitly refuse to discuss. For example, the conversation content constraint information includes: refusing to talk about any illegal or derogatory words to others. When the first user sends a conversation content that contains derogatory words to others, the conversation robot can directly reply: "Please don't belittle others."

[0047] Furthermore, the prompt words corresponding to the dialogue object are used to train the dialogue robot that controls the dialogue object, so that the dialogue robot can learn and understand the object setting information and game rules, so as to evaluate the game strategy in the dialogue content sent by the user and generate a reply that conforms to the object setting information of the dialogue object.

[0048] Here, the conversation robot includes multiple neural network models of specified functions, and the neural network models of different functions can be targeted to train part or all of the content in the prompt words to achieve the corresponding functions. For example, the conversation robot includes an evaluation model. After the evaluation model is pre-trained with the prompt words corresponding to the conversation object, the current conversation content, the correspondence between the pre-set alternative strategies and the change value of the evaluation parameter, and the historical conversation content before the current conversation content can be input into the pre-trained evaluation model to output the evaluation parameters corresponding to the current conversation content. The conversation robot also includes a language model. After the language model is pre-trained with the prompt words corresponding to the conversation object, the object setting information of the conversation object, the current conversation content, the evaluation parameters corresponding to the current conversation content, and the historical conversation content before the current conversation content can be input into the pre-trained language model to output the reply content that conforms to the object setting information of the conversation object and conforms to the format of the reply content.

[0049] In this step, after the conversation triggering instruction is triggered, the current conversation between the first user and the conversation object begins. It should be noted that after the conversation triggering instruction is triggered, the game system can send the conversation background to the conversation robot and the first user, as well as the pre-set conversation intention that the first user wants to achieve in the current conversation.

[0050] Step S104: receiving a first conversation content sent by a first user, and generating an evaluation parameter corresponding to the first conversation content; wherein the evaluation parameter is used to indicate: a state distance between an object state of a conversation object and a target state for the first conversation content;

[0051] The above-mentioned first conversation content is any conversation content sent by the first user after starting the current conversation between the first user and the conversation object. The first conversation content can be at least one of a conversation text, a conversation voice, a conversation video and a virtual prop. The above-mentioned object state refers to the emotional state of the conversation object after receiving the first conversation content or the behavioral content to be performed by the conversation object. The evaluation parameter corresponding to the above-mentioned first conversation content can be understood as the gap between the object state of the conversation object and the target state after receiving the first conversation content. The evaluation parameter corresponding to the first conversation content can be determined by the evaluation parameter corresponding to the previous conversation content before the first conversation content sent by the first user and the change value of the evaluation parameter corresponding to the first conversation content.

[0052] After receiving the first conversation content sent by the first user, the conversation robot generates an evaluation parameter corresponding to the first conversation content. Here, the conversation robot can identify the state transition strategy contained in the first conversation content, and the state transition strategy can be understood as a conversation strategy that can be used to control the change of the object state of the conversation object.

[0053] For example, in the current conversation, the first user's conversation intention is to persuade the conversation partner to release the first user. If the first conversation content is a conversation text: "I have a stomachache, can you let me out so I can go to the toilet, so as not to stink up my brothers", the conversation robot controlling the conversation partner can identify the state transition strategy as "I have a stomachache and want you to release him" from the conversation text. If the first conversation content is that the first user sent a gold coin virtual item, the conversation robot can identify the state transition strategy as "I send you a gold coin and want you to release him".

[0054] Furthermore, the conversational robot will determine whether the state transition strategy contained in the first conversation content belongs to the alternative strategy contained in the prompt word. If it is determined to be an alternative strategy, the evaluation parameter change value corresponding to the state transition strategy can be determined from the correspondence between the pre-set alternative strategy and the evaluation parameter change value. Then, the evaluation parameter corresponding to the first conversation content is determined according to the current value of the evaluation parameter and the evaluation parameter change value, wherein the current value of the evaluation parameter is the evaluation parameter corresponding to the previous conversation content before the first conversation content sent by the first user.

[0055] If the conversational robot determines that the state transition strategy is not the alternative strategy contained in the prompt word, it obtains the alternative strategy with the greatest similarity to the state transition strategy from the corresponding relationship, and then determines the evaluation parameter change value corresponding to the alternative strategy as the evaluation parameter change value corresponding to the state transition strategy, and determines the evaluation parameter corresponding to the first conversation content based on the current value of the evaluation parameter and the evaluation parameter change value.

[0056] It should be noted that the evaluation parameter can be described by a score value under a preset score system. For example, if the score corresponding to the target state is 100 points, the score value of the evaluation parameter may be no greater than 100 points. The greater the difference between the score value and 100 points, the greater the gap between the object state of the dialogue object for the dialogue content and the target state.

[0057] This step provides a way to quantify the player's language expression. Through evaluation parameters, the dialogue strategies contained in the dialogue content sent by the user are objectively evaluated, thereby guiding the user to continue or change the dialogue strategy and realize the preset dialogue intention. It solves the limitation that traditional dialogue robots can only talk and chat with players, realizes the game behavior interaction between users and dialogue objects, and improves the player's gaming experience.

[0058] Step S106: Generate reply content corresponding to the first conversation content, and send the reply content and evaluation parameters to the first user.

[0059] Here, the conversation robot can also generate reply content that matches the identity and personality settings of the conversation object based on the first conversation content and the identity information and personality information in the object setting information of the conversation object. Finally, the reply content and evaluation parameters are sent to the first user, thereby guiding the realization of the preset conversation intention.

[0060] In this way, the conversational robot can not only interact with players as an NPC, but also understand the rules and difficulty of the game, and interact with players dynamically, which greatly improves the interactivity of the game and the player's gaming experience.

[0061] The interactive control method of the above-mentioned dialogue comprises: in response to a dialogue trigger instruction, starting the current dialogue between the first user and the dialogue object; wherein the dialogue object is preset with a target state; the dialogue object is controlled by a dialogue robot; receiving the first dialogue content sent by the first user, and generating an evaluation parameter corresponding to the first dialogue content; wherein the evaluation parameter is used to indicate: the state distance between the object state of the dialogue object with respect to the first dialogue content and the target state; generating reply content corresponding to the first dialogue content, and sending the reply content and the evaluation parameter to the first user.

[0062] In this method, the dialogue object is preset with a target state. After the user sends a dialogue content, an evaluation parameter corresponding to the dialogue content is generated to indicate the state distance between the object state of the dialogue object with respect to the dialogue content and the target state, and a reply content corresponding to the dialogue content is generated. Through the evaluation parameters and the target state, the user can be guided to send subsequent dialogue content, so that the user and the dialogue robot can have a purposeful dialogue interaction, thereby improving the interactivity and dialogue experience between the user and the dialogue robot.

[0063] The following provides an implementation method for generating evaluation parameters corresponding to the first conversation content.

[0064] In one method, a state transition strategy included in the first dialogue content is identified; wherein the state transition strategy is used to: control the change of the object state of the dialogue object; and generate an evaluation parameter corresponding to the state transition strategy.

[0065] The evaluation parameters corresponding to the above-mentioned state transition strategy, that is, the evaluation parameters corresponding to the first conversation content, are determined by the evaluation parameters corresponding to the previous conversation content before the first conversation content sent by the first user and the change value of the evaluation parameters corresponding to the state transition strategy in the first conversation content.

[0066] Here, the dialogue robot can identify the state transition strategy contained in the first dialogue content. The state transition strategy is a dialogue strategy identified by the dialogue robot that can be used to control the change of the object state of the dialogue object.

[0067] For example, in the current conversation, the first user's conversation intention is to persuade the conversation partner to release the first user. If the first conversation content is a conversation text: "I have a stomachache, can you let me out so I can go to the toilet, so as not to stink up my brothers", the conversation robot can identify the state transition strategy as "I have a stomachache and want you to release him" from the conversation text. If the first conversation content is that the first user sent a gold coin virtual item, the conversation robot can identify the state transition strategy as "I sent a gold coin and want you to release him".

[0068] After the conversation robot recognizes the state transition strategy included in the first conversation content, it needs to generate the evaluation parameter corresponding to the state transition strategy by itself according to the state transition strategy. Specifically, the conversation robot can determine the evaluation parameter change value corresponding to the state transition strategy from the evaluation criteria of the evaluation parameter included in the prompt word, and obtain the evaluation parameter corresponding to the state transition strategy based on the evaluation parameter change value and the current value of the evaluation parameter.

[0069] Here, the conversational robot will determine whether the state transition strategy contained in the first conversation content is the alternative strategy contained in the prompt word. In one case, if it is determined to be an alternative strategy, the target change value corresponding to the state transition strategy is obtained from the correspondence between the pre-set alternative strategy and the evaluation parameter change value; based on the target change value and the current value of the evaluation parameter, the evaluation parameter corresponding to the state transition strategy is generated.

[0070] The above-mentioned alternative strategy refers to a pre-set dialogue strategy that can control the change of the object state of the dialogue object. When the dialogue robot recognizes that the dialogue content contains the alternative strategy, it can control the specified change of the object state of the dialogue object. The above-mentioned target change value refers to the evaluation parameter change value corresponding to the state transition strategy. The current value of the above-mentioned evaluation parameter is the evaluation parameter corresponding to the previous dialogue content before the first dialogue content sent by the first user. The above-mentioned correspondence indicates the relationship between each alternative strategy and the evaluation parameter change value. For example, the corresponding relationship can be: when the dialogue content contains alternative strategy 1, the evaluation parameter change value decreases by 5 points; when the dialogue content contains alternative strategy 2, the evaluation parameter change value increases by 3 points.

[0071] That is to say, the conversational robot can obtain the target change value corresponding to the state transition strategy from the correspondence between the pre-set alternative strategies and the evaluation parameter change values, and then add or subtract the current value of the evaluation parameter according to the target change value to generate the evaluation parameter corresponding to the state transition strategy, that is, the evaluation parameter corresponding to the first conversation content.

[0072] In another case, if it is determined that the state transition strategy contained in the first dialogue content is not the alternative strategy contained in the prompt word, it is necessary to obtain the target alternative strategy with the greatest similarity to the state transition strategy from the corresponding relationship, and determine the target change value corresponding to the state transition strategy based on the target alternative strategy.

[0073] Specifically, if the target change value corresponding to the state transition strategy cannot be obtained from the corresponding relationship, the target alternative strategy with the greatest similarity to the state transition strategy is obtained from the corresponding relationship; the evaluation parameter change value corresponding to the target alternative strategy is determined as the target change value corresponding to the state transition strategy.

[0074] The above-mentioned similarity includes semantic similarity. Here, if the target change value corresponding to the state transition strategy cannot be obtained from the corresponding relationship, the dialogue robot can infer the semantic similarity between the state transition strategy and each candidate strategy in the first dialogue content, obtain the target candidate strategy with the greatest similarity to the state transition strategy, determine the target change value corresponding to the target candidate strategy as the target change value corresponding to the state transition strategy, and then add or subtract the current value of the evaluation parameter according to the target change value to generate the evaluation parameter corresponding to the state transition strategy, that is, the evaluation parameter corresponding to the first dialogue content.

[0075] It should be noted that after the dialogue object receives the first dialogue content, the dialogue robot will obtain the historical dialogue content before the first dialogue content sent by the first user in the current dialogue; determine that the first dialogue content is the same as at least part of the historical dialogue content, and determine the evaluation parameters before the first dialogue content is sent as the evaluation parameters corresponding to the first dialogue content.

[0076] That is to say, after the dialogue object receives the first dialogue content, the dialogue robot will obtain the historical dialogue content before the first dialogue content sent by the first user in the current dialogue. If the dialogue robot determines that the first dialogue content is the same as at least part of the historical dialogue content sent by the first user, the target change value corresponding to the state transition strategy contained in the first dialogue content is 0, that is, the evaluation parameter corresponding to the previous dialogue content before the first dialogue content is determined as the evaluation parameter corresponding to the first dialogue content.

[0077] In this method, for the state transition strategy used in the user's conversation content, the target change value corresponding to the state transition strategy can be directly determined through the correspondence between the alternative strategy and the change value of the evaluation parameter. In the case where the state transition strategy does not belong to the alternative strategy, the conversation robot will infer the semantic similarity between the state transition strategy in the first conversation content and each alternative strategy based on the meaning of the discourse in the conversation content, determine the target change value corresponding to the state transition strategy, and then determine the evaluation parameter corresponding to the conversation content.

[0078] In actual implementation, the output process of the evaluation parameter corresponding to the first dialogue content can be performed in the following manner.

[0079] Specifically, the correspondence between the first conversation content, the pre-set alternative strategy and the change value of the evaluation parameter, and the historical conversation content before the first conversation content sent by the first user are input into the pre-trained evaluation model, and the evaluation parameter corresponding to the first conversation content is output.

[0080] Here, the conversational robot includes a pre-trained evaluation model, which may be a large language model. The correspondence between the first conversation content, the pre-set alternative strategy and the change value of the evaluation parameter, and the historical conversation content before the first conversation content sent by the first user may be input into the pre-trained evaluation model, and the evaluation model outputs the evaluation parameters corresponding to the first conversation content.

[0081] Taking the dialogue object as an NPC in the game scene as an example, the first user dialogue sends the dialogue content to the dialogue object, and the dialogue content will first be extracted by the multimodal model, and after security filtering, it will be submitted to the game server. The game server transmits the dialogue content after security filtering to the model server, and the model server runs a dialogue robot, which is composed of multiple neural network models. The game server inputs the first dialogue content, the correspondence between the preset alternative strategy and the change value of the evaluation parameter, and the historical dialogue content before the first dialogue content sent by the first user into the pre-trained evaluation model. The evaluation model can determine the state transition strategy contained in the first dialogue content and the target change value corresponding to the state transition strategy based on the first dialogue content, the correspondence between the preset alternative strategy and the change value of the evaluation parameter, and the historical dialogue content before the first dialogue content sent by the first user, and output the evaluation parameter corresponding to the first dialogue content according to the current value and the target change value of the evaluation parameter.

[0082] The conversation robot can also generate reply content that matches the identity and personality settings of the conversation partner based on the content of the first conversation.

[0083] Specifically, based on the object setting information of the dialogue object and the first dialogue content, the reply content corresponding to the first dialogue content is generated; wherein the object setting information includes: identity information and / or personality information; the reply content matches the object setting information.

[0084] The above-mentioned object setting information includes identity information and personality information, and also includes one of identity information or personality information. Identity information can include the image of the dialogue object, the name, gender, social identity, etc. of the dialogue object. Personality information includes personality traits, such as irritability, timidity, optimism, etc. For example, the object setting information of the dialogue object can be a young man named Li Yi, who has a bad temper, is sensitive and suspicious. The above-mentioned reply content is a reply to the first dialogue content, including a reply dialogue, and also includes at least one of action information, appearance information, and expression information. Among them, the action information, appearance information, and expression information can be expressed through text descriptions, or through the expressions and actions of the dialogue object in the game scene.

[0085] Here, the conversation robot can generate reply content that matches the object setting information according to the object setting information of the conversation object and the first conversation content. For example, the reply content is described in the form of text in the dialog box: "The conversation object protects the candied haws, looks at you warily and says: Don't think of hitting my candied haws."

[0086] In one method, the reply content corresponding to the first conversation content can also be generated based on the evaluation parameters. Here, the reply content can be generated according to the evaluation parameters corresponding to the first conversation content. For example, if the evaluation parameters corresponding to the first conversation content reach the evaluation parameters corresponding to the target state, it means that the conversation object is already in the target state, and the generated reply content is: "Congratulations on clearing the game." For another example, if the first conversation content is the same as at least part of the historical conversation content, the evaluation parameters corresponding to the first conversation content are determined as the evaluation parameters before the first conversation content is sent, and the generated reply information is: "Is it fun for you to keep repeating this sentence?".

[0087] In actual implementation, the process of outputting the reply content corresponding to the first conversation content can be performed in the following manner.

[0088] Specifically, the object setting information of the dialogue object, the first dialogue content, the evaluation parameter, and the historical dialogue content before the first dialogue content sent by the first user are input into the pre-trained language model, and the reply content corresponding to the first dialogue content is output.

[0089] Here, the conversational robot also includes a pre-trained language model, which may be a large language model. The object setting information of the conversation object, the first conversation content, the evaluation parameters, and the historical conversation content before the first conversation content sent by the first user may be input into the pre-trained language model, and the language model outputs the reply content corresponding to the first conversation content.

[0090] Taking the dialogue object as an NPC in the game scene as an example, the first user dialogue sends the dialogue content to the dialogue object, and the dialogue content will first be extracted by the multimodal model, and after security filtering, it will be submitted to the game server. The game server transmits the dialogue content after security filtering to the model server, and the model server runs a dialogue robot, which is composed of multiple neural network models. The game server inputs the first dialogue content, the correspondence between the preset alternative strategy and the change value of the evaluation parameter, and the historical dialogue content before the first dialogue content sent by the first user into the pre-trained evaluation model, and the evaluation model outputs the evaluation parameter corresponding to the first dialogue content. Then, the object setting information of the dialogue object, the first dialogue content, the evaluation parameter corresponding to the first dialogue content, and the historical dialogue content before the first dialogue content sent by the first user are input into the pre-trained language model, and the reply content corresponding to the first dialogue content is output. The reply content matches the object setting information and the evaluation parameter corresponding to the first dialogue content. The reply content includes a reply dialogue and at least one of action information, appearance information, and expression information.

[0091] In order to avoid the uncontrollable large language model, which may lead to the situation that after receiving the first conversation content sent by the user, the reply content corresponding to the first conversation content generated is not equal to the evaluation parameters corresponding to the first conversation content, thereby affecting the user's gaming experience, therefore, before sending the reply content and the evaluation parameters to the first user, when the reply content contains the specified content, it is also necessary to detect whether the evaluation parameters match the specified content contained in the reply content. If they do not match, the evaluation parameters are updated to evaluation parameters that match the specified content.

[0092] Specifically, before sending the reply content and evaluation parameters to the first user, when the reply content contains the first specified content, it is necessary to detect whether the evaluation parameters match the first specified content, determine if the evaluation parameters do not match the first specified content, and update the evaluation parameters to evaluation parameters that match the first specified content.

[0093] The first designated content is a preset content, and the first designated content has a corresponding relationship with the evaluation parameter corresponding to the first dialogue content.

[0094] For example, the first designated content may be content information expressing that the first user has achieved the preset dialogue intention. For example, when the first designated content is "Congratulations on completing the game", it indicates that the first user has achieved the preset dialogue intention through the first dialogue content. At this time, the dialogue object should be in the target state, and the evaluation parameter corresponding to the first dialogue content should reach the preset parameter threshold. If the evaluation parameter corresponding to the first dialogue content is the preset parameter threshold at this time, it means that the evaluation parameter matches the first designated content. If the evaluation parameter corresponding to the first dialogue content is not the preset parameter threshold at this time, the evaluation parameter does not match the first designated content, and the evaluation parameter is updated to the preset parameter threshold.

[0095] This method can effectively avoid the situation where the reply content corresponding to the conversation content is not equal to the evaluation parameter corresponding to the conversation content, so as not to affect the user's gaming experience.

[0096] In one method, after the reply content and evaluation parameters are sent to the first user, in response to a preset conversation ending condition being triggered, the current conversation is controlled to end; wherein the conversation ending condition is determined based on at least one of the first conversation content, the reply content and the evaluation parameters.

[0097] It is understandable that the current conversation can end after the user achieves the preset conversation intention and successfully completes the game; or it can end after the user fails to complete the game.

[0098] Here, when the preset conversation ending condition is triggered, the current conversation is controlled to end, and the conversation ending condition can be determined based on at least one of the first conversation content, the reply content and the evaluation parameter.

[0099] Specifically, the above-mentioned conversation end conditions may include at least one of the following: the evaluation parameter reaches a preset parameter threshold; the first conversation content contains the second specified content; in the current conversation, the generation round of the reply content reaches a preset round threshold.

[0100] The above-mentioned preset parameter threshold can be understood as that the first user has realized the preset conversation intention through the first conversation content, and the conversation object should be in the target state at this time, and the evaluation parameter corresponding to the first conversation content is the preset parameter threshold. When the evaluation parameter corresponding to the first conversation content reaches the preset parameter threshold, the conversation end condition is triggered, and the current conversation is controlled to end.

[0101] In one case, after the user sends the first dialogue content, a dialogue effect opposite to the preset dialogue intention is achieved, resulting in the failure of the game and the end of the current dialogue. For example, the dialogue intention between the first user and the dialogue object is to persuade the dialogue object to go to school, and the first content sent by the first user is: "Go to school quickly or I will punish you corporally", then the triggered reply content is: "The dialogue object makes a face and replies: I won't go, I'll run! Congratulations on triggering the special ending, the dialogue object has escaped, and it will continue to skip school." The above-mentioned second specified content can be a dialogue strategy that includes achieving a dialogue effect opposite to the preset dialogue intention, such as the above-mentioned "corporal punishment if you don't go to school" dialogue strategy.

[0102] In another method, the conversation end condition may be that the number of reply content generation rounds in the current conversation reaches a preset round threshold. Here, each time the first user sends a conversation content, the conversation partner generates a reply content and evaluation parameters for the conversation content and sends them to the first user, which can be regarded as a round of conversation, and the number of reply content generation rounds increases by 1.

[0103] That is, if the generation round of reply content reaches a preset round threshold after the user sends the first conversation content, the current conversation is controlled to end.

[0104] It should be noted that the first conversation content includes: at least one of conversation text, conversation voice, conversation video and virtual props.

[0105] That is, the first user can have a current conversation with the conversation partner through at least one of conversation text, conversation voice, conversation video and virtual props.

[0106] In one method, a first user can have a conversation with a dialogue object in a game scene through a terminal device, control an object model of the dialogue object to be displayed on the first user's terminal device, and control the object model to perform an action matching the reply content.

[0107] In this method, the object model of the conversation object can be controlled to be displayed on the terminal device of the first user, and the object model can be controlled to perform an action that matches the reply content. For example, after the first user enters the first conversation content, the reply content is described in the form of text in the dialog box as: "The conversation object protects the candied haws, looks at you warily and says: Don't think of hitting my candied haws", and at the same time, the object model of the conversation object is displayed on the terminal device of the first user and the object model is controlled to perform the action of protecting the candied haws, which greatly improves the interactivity of the game and the game experience of the player.

[0108] In a specific implementation, the process of interactive control of the dialogue is introduced in detail.

[0109] 1) Create prompt words corresponding to the dialogue object, and use the prompt words corresponding to the dialogue object to train the dialogue robot that controls the dialogue object, so that the dialogue robot understands the game rules and the role of the dialogue object. For example, a prompt word is shown in Table 1 below.

[0110] Table 1

[0111]

[0112]

[0113]

[0114] 2) After the dialogue triggering instruction is triggered, the current dialogue between the first user and the dialogue object begins. The game system can send the game background to the dialogue robot and the first user, as well as the dialogue intention that the first user wants to achieve in the current dialogue.

[0115] 3) The first user inputs the first dialogue content in the graphical interactive interface provided by the terminal device. The first dialogue content can be a speech text, audio, picture or video. These data contain a state transition strategy, which is a dialogue strategy that can be used to control the change of the object state of the dialogue object recognized by the dialogue robot. For example, the dialogue object is a jailer NPC, and the first user's dialogue intention is to persuade the dialogue object to release him from prison. Then the first user's dialogue strategy can be that the first user tells the jailer NPC that he has a stomachache and needs to be released from prison; for another example: the dialogue object is an NPC, and the first user's dialogue intention is to persuade the dialogue object, then the first user's dialogue strategy can be: persuade the NPC to live well, etc.

[0116] 4) The first conversation content input by the first user will first be extracted by the multimodal model, and after security filtering, it will be submitted to the game server.

[0117] 5) The game server transmits the securely filtered first conversation content to a model server where a conversation robot is deployed. The conversation robot may be composed of multiple large language models, including a pre-trained evaluation model and a pre-trained language model.

[0118] 6) The first conversation content, the correspondence between the preset alternative strategies and the change value of the evaluation parameter, and the historical conversation content before the first conversation content sent by the first user are input into a pre-trained evaluation model. The evaluation model can determine the state transition strategy contained in the first conversation content and the target change value corresponding to the state transition strategy based on the first conversation content, the correspondence between the preset alternative strategies and the change value of the evaluation parameter, and the historical conversation content before the first conversation content sent by the first user, and output the evaluation parameters corresponding to the first conversation content according to the current value and the target change value of the evaluation parameters.

[0119] The object setting information of the dialogue object, the first dialogue content, the evaluation parameters corresponding to the first dialogue content, and the historical dialogue content before the first dialogue content sent by the first user are input into a pre-trained language model, and the reply content corresponding to the first dialogue content is output, and the reply content matches the object setting information and the evaluation parameters corresponding to the first dialogue content. The reply content includes a reply dialogue and at least one of action information, appearance information, and expression information.

[0120] In this process, the conversational robot needs to grasp the global information of the game and the user's behavioral intentions.

[0121] 7) The reply content generated by the dialogue robot will be subjected to security filtering again to detect whether the evaluation parameters match the first specified content contained in the reply content. If they do not match, the evaluation parameters will be updated to evaluation parameters that match the first specified content.

[0122] 8) The conversation robot transmits the reply content and evaluation parameters to the first user's terminal device through the server, controls the object model of the conversation object to be displayed on the first user's terminal device, and controls the object model to perform actions matching the reply content.

[0123] Figures 2 to 3 An example of a current conversation between a first user and a conversation object is provided, wherein the conversation content sent by the first user each time is a conversation text. After receiving the conversation content, the conversation robot generates a reply content corresponding to the conversation content and an evaluation parameter corresponding to the conversation content (i.e., the fluctuation value in the figure) as a reply. When the evaluation parameter reaches a preset parameter threshold (such as 100), or the conversation content contains a second specified content, the current conversation is controlled to end. The conversation robot controls the conversation object to describe the reply content in the form of text in the dialog box, and the reply content includes a reply conversation, and also includes at least one of action information, appearance information, and expression information.

[0124] Figure 2 The dialogue partner in the game plays the role of a jailer, and the first user's dialogue intention is to persuade the jailer to release him from prison. Figure 3 The dialogue partner in the game plays the role of Hu Zhuangzhuang, a child who wants to skip school. The first user's dialogue intention is to persuade Hu Zhuangzhuang to give up skipping school, but the dialogue content includes the second specified content, achieving a dialogue effect opposite to the preset dialogue intention, triggering a special ending, resulting in the failure of the game and the end of the current dialogue.

[0125] In this method, the large language model is applied to game interaction, which solves the limitation that the traditional large language model can only communicate with players, realizes the game behavior interaction between players and dialogue robots, and enriches the game experience. Through the evaluation parameters, the dialogue strategy contained in the dialogue content sent by the user is objectively evaluated, so as to guide the user to continue or change the dialogue strategy and realize the preset dialogue intention, which solves the limitation that the traditional dialogue robot can only communicate and chat with players, realizes the game behavior interaction between users and dialogue objects, and enhances the game experience of players. Applying the large language model to the game not only improves the fun of the game, but also expands the application field of the large language model, reduces the development cost of the game, and has broad market prospects.

[0126] For an embodiment corresponding to the above method, see Figure 4 A schematic diagram of a dialogue interactive control device is shown, the device comprising:

[0127] The first start module 402 is used to start the current conversation between the first user and the conversation object in response to the conversation trigger instruction; wherein the conversation object is preset with a target state; and the conversation object is controlled by the conversation robot;

[0128] The first generating module 404 is used to receive the first conversation content sent by the first user, and generate an evaluation parameter corresponding to the first conversation content; wherein the evaluation parameter is used to indicate: the state distance between the object state and the target state of the conversation object with respect to the first conversation content;

[0129] The first sending module 406 is used to generate reply content corresponding to the first conversation content, and send the reply content and evaluation parameters to the first user.

[0130] In this method, the dialogue object is preset with a target state, the user sends out dialogue content, and generates evaluation parameters corresponding to the dialogue content to indicate the state distance between the object state of the dialogue object with respect to the dialogue content and the target state, and generates reply content corresponding to the dialogue content. Through the evaluation parameters and the target state, the user can be guided to send out subsequent dialogue content, so that the user and the dialogue robot can have a purposeful dialogue interaction, thereby improving the interactivity and dialogue experience between the user and the dialogue robot.

[0131] The above-mentioned first generation module is also used to identify the state transition strategy contained in the first dialogue content; wherein the state transition strategy is used to: control the change of the object state of the dialogue object; and generate evaluation parameters corresponding to the state transition strategy.

[0132] The above-mentioned device also includes a second generating module, which is used to obtain the target change value corresponding to the state transition strategy from the correspondence between the preset alternative strategies and the evaluation parameter change values; and generate the evaluation parameters corresponding to the state transition strategy based on the target change value and the current value of the evaluation parameter.

[0133] The above-mentioned device also includes a first determination module, which is used to obtain the target alternative strategy with the greatest similarity to the state transition strategy from the corresponding relationship if the target change value corresponding to the state transition strategy cannot be obtained from the corresponding relationship; and determine the evaluation parameter change value corresponding to the target alternative strategy as the target change value corresponding to the state transition strategy.

[0134] The above-mentioned first generation module is also used to obtain historical conversation content before the first conversation content sent by the first user in the current conversation; determine that the first conversation content is the same as at least part of the historical conversation content, and determine the evaluation parameters before the first conversation content is sent as the evaluation parameters corresponding to the first conversation content.

[0135] The above-mentioned first generation module is also used to input the correspondence between the first conversation content, the pre-set alternative strategy and the change value of the evaluation parameter, and the historical conversation content before the first conversation content sent by the first user into the pre-trained evaluation model, and output the evaluation parameters corresponding to the first conversation content.

[0136] The above-mentioned first sending module is also used to generate reply content corresponding to the first dialogue content based on the object setting information of the dialogue object and the first dialogue content; wherein the object setting information includes: identity information and / or personality information; the reply content matches the object setting information.

[0137] The above reply content includes a reply dialogue, and also includes at least one of action information, appearance information, and expression information.

[0138] The first sending module is further configured to generate reply content corresponding to the first conversation content based on the evaluation parameter.

[0139] The above-mentioned first sending module is also used to input the object setting information of the dialogue object, the first dialogue content, the evaluation parameters, and the historical dialogue content before the first dialogue content sent by the first user into the pre-trained language model, and output the reply content corresponding to the first dialogue content.

[0140] The above device also includes a second determination module, which is used to detect whether the evaluation parameter matches the first specified content when the reply content contains the first specified content; determine that the evaluation parameter does not match the first specified content, and update the evaluation parameter to an evaluation parameter that matches the first specified content.

[0141] The above-mentioned device also includes a first control module, which is used to control the end of the current conversation in response to a preset conversation end condition being triggered; wherein the conversation end condition is determined based on at least one of the first conversation content, the reply content and the evaluation parameter.

[0142] The above-mentioned dialogue termination conditions include at least one of the following: the evaluation parameter reaches a preset parameter threshold; the first dialogue content contains the second specified content; in the current dialogue, the generation round of the reply content reaches a preset round threshold.

[0143] The first conversation content includes at least one of conversation text, conversation voice, conversation video and virtual props.

[0144] The first sending module is further used to control the object model of the dialogue object to be displayed on the terminal device of the first user, and to control the object model to execute an action matching the reply content.

[0145] This embodiment also provides an electronic device, including a processor and a memory, wherein the memory stores machine executable instructions that can be executed by the processor, and the processor executes the machine executable instructions to implement the above-mentioned interactive control method of the dialogue. The electronic device can be a server or a terminal device.

[0146] See also Figure 5 As shown, the electronic device includes a processor 100 and a memory 101, wherein the memory 101 stores machine executable instructions that can be executed by the processor 100, and the processor 100 executes the machine executable instructions to implement the above-mentioned interactive control method of the dialogue.

[0147] Further, Figure 5 The electronic device shown further includes a bus 102 and a communication interface 103 , and the processor 100 , the communication interface 103 and the memory 101 are connected via the bus 102 .

[0148] The memory 101 may include a high-speed random access memory (RAM), and may also include a non-volatile memory, such as at least one disk storage. The communication connection between the system network element and at least one other network element is realized through at least one communication interface 103 (which may be wired or wireless), and the Internet, wide area network, local area network, metropolitan area network, etc. may be used. The bus 102 may be an ISA bus, a PCI bus, or an EISA bus, etc. The bus may be divided into an address bus, a data bus, a control bus, etc. For ease of representation, Figure 5 Only one bidirectional arrow is used in the diagram, but this does not mean that there is only one bus or only one type of bus.

[0149] The processor 100 may be an integrated circuit chip with signal processing capabilities. In the implementation process, each step of the above method can be completed by the hardware integrated logic circuit or software instructions in the processor 100. The above processor 100 can be a general-purpose processor, including a central processing unit (CPU), a network processor (NP), etc.; it can also be a digital signal processor (DSP), an application-specific integrated circuit (ASIC), a field-programmable gate array (FPGA) or other programmable logic devices, discrete gates or transistor logic devices, discrete hardware components. The methods, steps and logic block diagrams disclosed in the embodiments of the present invention can be implemented or executed. The general-purpose processor can be a microprocessor or the processor can also be any conventional processor, etc. The steps of the method disclosed in conjunction with the embodiments of the present invention can be directly embodied as a hardware decoding processor for execution, or a combination of hardware and software modules in the decoding processor for execution. The software module may be located in a storage medium mature in the art, such as a random access memory, a flash memory, a read-only memory, a programmable read-only memory, or an electrically erasable programmable memory, a register, etc. The storage medium is located in the memory 101, and the processor 100 reads the information in the memory 101 and completes the steps of the method of the above embodiment in combination with its hardware.

[0150] The processor in the above-mentioned electronic device can implement the following operations of the above-mentioned interactive control method of the dialogue by executing machine-executable instructions: in response to a dialogue trigger instruction, start the current dialogue between the first user and the dialogue object; wherein the dialogue object is preset with a target state; the dialogue object is controlled by a dialogue robot; receive the first dialogue content sent by the first user, and generate an evaluation parameter corresponding to the first dialogue content; wherein the evaluation parameter is used to indicate: the state distance between the object state of the dialogue object for the first dialogue content and the target state; generate reply content corresponding to the first dialogue content, and send the reply content and the evaluation parameter to the first user.

[0151] In this method, the dialogue object is preset with a target state. After the user sends a dialogue content, an evaluation parameter corresponding to the dialogue content is generated to indicate the state distance between the object state of the dialogue object with respect to the dialogue content and the target state, and a reply content corresponding to the dialogue content is generated. Through the evaluation parameters and the target state, the user can be guided to send subsequent dialogue content, so that the user and the dialogue robot can have a purposeful dialogue interaction, thereby improving the interactivity and dialogue experience between the user and the dialogue robot.

[0152] The processor in the above-mentioned electronic device can implement the following operations of the above-mentioned interactive control method of the dialogue by executing machine-executable instructions: identifying the state transition strategy contained in the first dialogue content; wherein the state transition strategy is used to: control the change of the object state of the dialogue object; and generate evaluation parameters corresponding to the state transition strategy.

[0153] The processor in the above-mentioned electronic device can implement the following operations of the interactive control method of the above-mentioned dialogue by executing machine-executable instructions: obtaining the target change value corresponding to the state transition strategy from the correspondence between the pre-set alternative strategies and the evaluation parameter change values; and generating the evaluation parameters corresponding to the state transition strategy based on the target change value and the current value of the evaluation parameter.

[0154] The processor in the above-mentioned electronic device can implement the following operations of the interactive control method of the above-mentioned dialogue by executing machine-executable instructions: if the target change value corresponding to the state transition strategy cannot be obtained from the corresponding relationship, obtain the target alternative strategy with the greatest similarity to the state transition strategy from the corresponding relationship; determine the evaluation parameter change value corresponding to the target alternative strategy as the target change value corresponding to the state transition strategy.

[0155] The processor in the above-mentioned electronic device can implement the following operations of the above-mentioned interactive control method of the conversation by executing machine-executable instructions: obtaining the historical conversation content before the first conversation content sent by the first user in the current conversation; determining that the first conversation content is the same as at least part of the historical conversation content, and determining the evaluation parameters before the first conversation content is sent as the evaluation parameters corresponding to the first conversation content.

[0156] The processor in the above-mentioned electronic device can implement the following operations of the above-mentioned interactive control method of the conversation by executing machine-executable instructions: input the correspondence between the first conversation content, the pre-set alternative strategy and the change value of the evaluation parameter, and the historical conversation content before the first conversation content sent by the first user into the pre-trained evaluation model, and output the evaluation parameters corresponding to the first conversation content.

[0157] The processor in the above-mentioned electronic device can implement the following operations of the above-mentioned interactive control method of the dialogue by executing machine-executable instructions: based on the object setting information of the dialogue object and the first dialogue content, generate reply content corresponding to the first dialogue content; wherein the object setting information includes: identity information and / or personality information; the reply content matches the object setting information.

[0158] The above reply content includes a reply dialogue, and also includes at least one of action information, appearance information, and expression information.

[0159] The processor in the above-mentioned electronic device can implement the following operations of the above-mentioned interactive control method of the dialogue by executing machine-executable instructions: generating reply content corresponding to the first dialogue content based on the evaluation parameters.

[0160] The processor in the above-mentioned electronic device can implement the following operations of the above-mentioned interactive control method of the dialogue by executing machine-executable instructions: input the object setting information of the dialogue object, the first dialogue content, the evaluation parameters, and the historical dialogue content before the first dialogue content sent by the first user into a pre-trained language model, and output the reply content corresponding to the first dialogue content.

[0161] The processor in the above-mentioned electronic device can implement the following operations of the interactive control method of the above-mentioned dialogue by executing machine-executable instructions: when the reply content contains the first specified content, detecting whether the evaluation parameter matches the first specified content; determining that the evaluation parameter does not match the first specified content, updating the evaluation parameter to an evaluation parameter that matches the first specified content.

[0162] The processor in the above-mentioned electronic device can implement the following operations of the above-mentioned interactive control method of the conversation by executing machine-executable instructions: in response to a preset conversation end condition being triggered, controlling the end of the current conversation; wherein the conversation end condition is determined based on at least one of the first conversation content, the reply content and the evaluation parameter.

[0163] The above-mentioned dialogue termination conditions include at least one of the following: the evaluation parameter reaches a preset parameter threshold; the first dialogue content contains the second specified content; in the current dialogue, the generation round of the reply content reaches a preset round threshold.

[0164] The first conversation content includes at least one of conversation text, conversation voice, conversation video and virtual props.

[0165] The processor in the above-mentioned electronic device can implement the following operations of the above-mentioned interactive control method of the dialogue by executing machine-executable instructions: controlling the object model of the dialogue object to be displayed on the terminal device of the first user, and controlling the object model to perform actions matching the reply content.

[0166] This embodiment also provides a machine-readable storage medium, which stores machine-executable instructions. When the machine-executable instructions are called and executed by a processor, the machine-executable instructions prompt the processor to implement the above-mentioned interactive control method of the dialogue.

[0167] The machine executable instructions stored in the above-mentioned machine-readable storage medium can implement the following operations in the above-mentioned interactive control method of the dialogue by executing the machine executable instructions: in response to a dialogue trigger instruction, starting the current dialogue between the first user and the dialogue object; wherein the dialogue object is preset with a target state; the dialogue object is controlled by a dialogue robot; receiving the first dialogue content sent by the first user, and generating evaluation parameters corresponding to the first dialogue content; wherein the evaluation parameters are used to indicate: the state distance between the object state of the dialogue object for the first dialogue content and the target state; generating reply content corresponding to the first dialogue content, and sending the reply content and the evaluation parameters to the first user.

[0168] In this method, the dialogue object is preset with a target state. After the user sends a dialogue content, an evaluation parameter corresponding to the dialogue content is generated to indicate the state distance between the object state of the dialogue object with respect to the dialogue content and the target state, and a reply content corresponding to the dialogue content is generated. Through the evaluation parameters and the target state, the user can be guided to send subsequent dialogue content, so that the user and the dialogue robot can have a purposeful dialogue interaction, thereby improving the interactivity and dialogue experience between the user and the dialogue robot.

[0169] The machine executable instructions stored in the above-mentioned machine-readable storage medium can implement the following operations in the above-mentioned interactive control method of the dialogue by executing the machine executable instructions: identifying the state transition strategy contained in the first dialogue content; wherein the state transition strategy is used to: control the change of the object state of the dialogue object; and generate evaluation parameters corresponding to the state transition strategy.

[0170] The machine executable instructions stored in the above-mentioned machine-readable storage medium can implement the following operations in the above-mentioned interactive control method of the dialogue by executing the machine executable instructions: obtaining the target change value corresponding to the state transition strategy from the correspondence between the pre-set alternative strategies and the evaluation parameter change values; and generating the evaluation parameters corresponding to the state transition strategy based on the target change value and the current value of the evaluation parameter.

[0171] The machine executable instructions stored in the above-mentioned machine-readable storage medium can implement the following operations in the interactive control method of the above-mentioned dialogue by executing the machine executable instructions: if the target change value corresponding to the state transition strategy cannot be obtained from the corresponding relationship, obtain the target alternative strategy with the greatest similarity to the state transition strategy from the corresponding relationship; determine the evaluation parameter change value corresponding to the target alternative strategy as the target change value corresponding to the state transition strategy.

[0172] The machine executable instructions stored in the above-mentioned machine-readable storage medium can be executed to implement the following operations in the above-mentioned interactive control method of the conversation: obtaining the historical conversation content before the first conversation content sent by the first user in the current conversation; determining that the first conversation content is the same as at least part of the historical conversation content, and determining the evaluation parameters before the first conversation content is sent as the evaluation parameters corresponding to the first conversation content.

[0173] The machine executable instructions stored in the above-mentioned machine-readable storage medium can implement the following operations in the above-mentioned interactive control method of the dialogue by executing the machine executable instructions: input the correspondence between the first dialogue content, the pre-set alternative strategy and the change value of the evaluation parameter, and the historical dialogue content before the first dialogue content sent by the first user into the pre-trained evaluation model, and output the evaluation parameters corresponding to the first dialogue content.

[0174] The machine executable instructions stored in the above-mentioned machine-readable storage medium can implement the following operations in the above-mentioned interactive control method of the dialogue by executing the machine executable instructions: based on the object setting information of the dialogue object and the first dialogue content, generating reply content corresponding to the first dialogue content; wherein the object setting information includes: identity information and / or personality information; the reply content matches the object setting information.

[0175] The above reply content includes a reply dialogue, and also includes at least one of action information, appearance information, and expression information.

[0176] The machine executable instructions stored in the above-mentioned machine-readable storage medium can implement the following operations in the above-mentioned interactive control method of the dialogue by executing the machine executable instructions: based on the evaluation parameters, generating reply content corresponding to the first dialogue content.

[0177] The machine executable instructions stored in the above-mentioned machine-readable storage medium can implement the following operations in the above-mentioned interactive control method of the dialogue by executing the machine executable instructions: inputting the object setting information of the dialogue object, the first dialogue content, the evaluation parameters, and the historical dialogue content before the first dialogue content sent by the first user into a pre-trained language model, and outputting the reply content corresponding to the first dialogue content.

[0178] The machine executable instructions stored in the above-mentioned machine-readable storage medium can be executed to implement the following operations in the interactive control method of the above-mentioned dialogue: when the reply content contains the first specified content, detecting whether the evaluation parameter matches the first specified content; determining that the evaluation parameter does not match the first specified content, updating the evaluation parameter to an evaluation parameter that matches the first specified content.

[0179] The machine executable instructions stored in the above-mentioned machine-readable storage medium can implement the following operations in the above-mentioned interactive control method of the conversation by executing the machine executable instructions: in response to a preset conversation ending condition being triggered, controlling the end of the current conversation; wherein the conversation ending condition is determined based on at least one of the first conversation content, the reply content and the evaluation parameter.

[0180] The above-mentioned dialogue termination conditions include at least one of the following: the evaluation parameter reaches a preset parameter threshold; the first dialogue content contains the second specified content; in the current dialogue, the generation round of the reply content reaches a preset round threshold.

[0181] The first conversation content includes at least one of conversation text, conversation voice, conversation video and virtual props.

[0182] The machine executable instructions stored in the above-mentioned machine-readable storage medium can implement the following operations in the above-mentioned interactive control method of the dialogue by executing the machine executable instructions: controlling the object model of the dialogue object to be displayed on the terminal device of the first user, and controlling the object model to perform actions matching the reply content.

[0183] The computer program product of the interactive control method, device and electronic device for dialogue provided in the embodiments of the present invention includes a computer-readable storage medium storing program code. The instructions included in the program code can be used to execute the method in the previous method embodiment. The specific implementation can be found in the method embodiment, which will not be repeated here.

[0184] Those skilled in the art can clearly understand that, for the convenience and brevity of description, the specific working process of the system and device described above can refer to the corresponding process in the aforementioned method embodiment, and will not be repeated here.

[0185] In addition, in the description of the embodiments of the present invention, unless otherwise clearly specified and limited, the terms "installed", "connected", and "connected" should be understood in a broad sense, for example, it can be a fixed connection, a detachable connection, or an integral connection; it can be a mechanical connection or an electrical connection; it can be a direct connection or an indirect connection through an intermediate medium, and it can be the internal communication of two components. For those skilled in the art, the specific meanings of the above terms in the present invention can be understood according to specific circumstances.

[0186] If the function is implemented in the form of a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present invention, or the part that contributes to the prior art, or the part of the technical solution, can be embodied in the form of a software product. The computer software product is stored in a storage medium, including several instructions for a computer device (which can be a personal computer, a server, or a network device, etc.) to perform all or part of the steps of the methods of each embodiment of the present invention. The aforementioned storage medium includes: U disk, mobile hard disk, read-only memory (ROM, Read-Only Memory), random access memory (RAM, Random Access Memory), disk or optical disk, etc. Various media that can store program codes.

[0187] In the description of the present invention, it should be noted that the terms "center", "upper", "lower", "left", "right", "vertical", "horizontal", "inner", "outer", etc., indicating the orientation or positional relationship, are based on the orientation or positional relationship shown in the drawings, and are only for the convenience of describing the present invention and simplifying the description, rather than indicating or implying that the device or element referred to must have a specific orientation, be constructed and operated in a specific orientation, and therefore cannot be understood as limiting the present invention. In addition, the terms "first", "second", and "third" are used for descriptive purposes only, and cannot be understood as indicating or implying relative importance.

[0188] Finally, it should be noted that the above embodiments are only specific implementations of the present invention, which are used to illustrate the technical solutions of the present invention, rather than to limit them. The protection scope of the present invention is not limited thereto. Although the present invention is described in detail with reference to the above embodiments, those skilled in the art should understand that any person skilled in the art can still modify the technical solutions recorded in the above embodiments within the technical scope disclosed by the present invention, or can easily think of changes, or make equivalent replacements for some of the technical features therein; and these modifications, changes or replacements do not make the essence of the corresponding technical solutions deviate from the spirit and scope of the technical solutions of the embodiments of the present invention, and should be included in the protection scope of the present invention. Therefore, the protection scope of the present invention shall be based on the protection scope of the claims.

Claims

1. A method for interactive control of a dialogue, characterized in that the method include: In response to the dialogue triggering instruction, starting the current dialogue between the first user and the dialogue object; wherein the dialogue object is preset with a target state; and the dialogue object is controlled by the dialogue robot; Receiving a first conversation content sent by a first user, and generating an evaluation parameter corresponding to the first conversation content; wherein the evaluation parameter is used to indicate: a state distance between an object state of a conversation object and a target state with respect to the first conversation content; Generate reply content corresponding to the first conversation content, and send the reply content and the evaluation parameter to the first user.

2. The method according to claim 1, characterized in that The step of generating an evaluation parameter corresponding to the first conversation content includes: Identifying a state transition strategy included in the first conversation content; wherein the state transition strategy is used to: control the object state of the conversation object to change; Generate evaluation parameters corresponding to the state transition strategy.

3. The method according to claim 2, characterized in that The steps of generating evaluation parameters corresponding to the state transition strategy include: Obtaining the target change value corresponding to the state transition strategy from the correspondence between the preset alternative strategies and the change values ​​of the evaluation parameters; Based on the target change value and the current value of the evaluation parameter, the evaluation parameter corresponding to the state transition strategy is generated.

4. The method according to claim 3, characterized in that The method also includes: If the target change value corresponding to the state transition strategy cannot be obtained from the corresponding relationship, the target candidate strategy with the greatest similarity to the state transition strategy is obtained from the corresponding relationship; The evaluation parameter change value corresponding to the target alternative strategy is determined as the target change value corresponding to the state transition strategy.

5. The method according to claim 1, characterized in that The step of generating an evaluation parameter corresponding to the first conversation content includes: Obtaining historical conversation content before the first conversation content sent by the first user in the current conversation; It is determined that the first conversation content is the same as at least a portion of the historical conversation content, and the evaluation parameter before the first conversation content is sent is determined as the evaluation parameter corresponding to the first conversation content.

6. The method according to claim 1, characterized in that The step of generating an evaluation parameter corresponding to the first conversation content includes: The corresponding relationship between the first conversation content, the preset alternative strategy and the change value of the evaluation parameter, and the historical conversation content before the first conversation content sent by the first user are input into the pre-trained evaluation model, and the evaluation parameter corresponding to the first conversation content is output.

7. The method according to claim 1, characterized in that The step of generating reply content corresponding to the first conversation content includes: Based on the object setting information of the dialogue object and the first dialogue content, a reply content corresponding to the first dialogue content is generated; wherein the object setting information includes: identity information and / or personality information; and the reply content matches the object setting information.

8. The method according to claim 7, characterized in that The reply content includes a reply to the conversation and at least one of action information, appearance information, and expression information.

9. The method according to claim 1, characterized in that The step of generating reply content corresponding to the first conversation content includes: Based on the evaluation parameters, reply content corresponding to the first conversation content is generated.

10. The method according to claim 1, characterized in that The step of generating reply content corresponding to the first conversation content includes: The object setting information of the dialogue object, the first dialogue content, the evaluation parameter, and the historical dialogue content before the first dialogue content sent by the first user are input into the pre-trained language model, and the reply content corresponding to the first dialogue content is output.

11. The method according to claim 1, characterized in that Before the step of sending the reply content and the evaluation parameter to the first user, the method further includes: When the reply content includes the first specified content, detecting whether the evaluation parameter matches the first specified content; It is determined that the evaluation parameter does not match the first specified content, and the evaluation parameter is updated to an evaluation parameter that matches the first specified content.

12. The method according to claim 1, characterized in that After the step of sending the reply content and the evaluation parameter to the first user, the method further includes: In response to a preset conversation ending condition being triggered, the current conversation is controlled to end; wherein the conversation ending condition is determined based on at least one of the first conversation content, the reply content and the evaluation parameter.

13. The method according to claim 12, characterized in that The dialog ending condition includes at least one of the following: The evaluation parameter reaches the preset parameter threshold; The first conversation content includes the second specified content; In the current conversation, the number of rounds of reply content generation reaches the preset round threshold.

14. The method according to claim 1, characterized in that The first conversation content includes: at least one of conversation text, conversation voice, conversation video and virtual props.

15. The method according to claim 1, characterized in that The step of sending the reply content and the evaluation parameter to the first user includes: The object model of the dialog object is controlled to be displayed on the terminal device of the first user, and the object model is controlled to execute an action matching the reply content.

16. A conversational interactive control device, characterized in that: The device includes: A first start module is used to start the current conversation between the first user and the conversation object in response to the conversation trigger instruction; wherein the conversation object is preset with a target state; and the conversation object is controlled by the conversation robot; A first generating module is used to receive a first conversation content sent by a first user, and generate an evaluation parameter corresponding to the first conversation content; wherein the evaluation parameter is used to indicate: a state distance between an object state of a conversation object and a target state for the first conversation content; The first sending module is used to generate reply content corresponding to the first conversation content, and send the reply content and evaluation parameters to the first user.

17. An electronic device, characterized in that: The invention comprises a processor and a memory, wherein the memory stores computer executable instructions that can be executed by the processor, and the processor executes the computer executable instructions to implement the interactive control method of a dialogue according to any one of claims 1 to 15.

18. A computer-readable storage medium, characterized in that: The computer-readable storage medium stores computer-executable instructions. When the computer-executable instructions are called and executed by a processor, the computer-executable instructions prompt the processor to implement the interactive control method of a dialogue according to any one of claims 1 to 15.

Images