Method for generating dialogue text based on customer behaviors and related product
By obtaining and analyzing the dialogue text and setting of target characters, and using psychological models to predict the reply content, the problem of insufficient accuracy of dialogue text in the existing technology is solved, and the accuracy of dialogue text and the practical ability of sales personnel is improved.
Patent Information
- Application Number
- CN202411726142.6
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2024-11-27
- Publication Date
- 2025-03-18
AI Technical Summary
The existing software for generating dialogue text uses preset speech templates and cannot dynamically respond to the specific situation of sales personnel, resulting in insufficient accuracy of generating dialogue texts, affecting the actual training effect of sales personnel.
By obtaining the current dialogue text of the target character, analyzing its character settings and historical dialogue, generating input features, and analyzing it based on psychological models, predicting the reply content of the target character, thereby improving the accuracy of the dialogue text.
Through psychological modeling, we ensure that the simulated customer behavior is consistent with the psychological research results and is close to the response mode of real users, which improves the accuracy of generating dialogue texts and enhances the practical communication skills of sales personnel.
Smart Images
Figure CN119646153A_ABST
Abstract
Description
Technical Field
[0001] The present application relates to the field of artificial intelligence technology, and in particular to a method for generating conversation text based on customer behavior and related products. Background Art
[0002] Life insurance sales are highly dependent on the performance of agents. Many agents often find it difficult to speak in actual operations due to lack of practical experience. Therefore, it is necessary to build a life insurance sales simulation training system to enhance the agents' practical experience and communication skills so that they can speak to customers.
[0003] Existing software for generating conversation texts usually uses preset speech templates, which are very standardized and cannot dynamically respond to the specific situations of sales personnel and simulate the feedback of real customers, resulting in poor accuracy when generating conversation texts. Sales personnel mechanically memorize the conversation texts with insufficient accuracy, and such training has little effect on actual combat.
[0004] Therefore, how to improve the accuracy of generated dialogue text is an issue that technicians in this field need to solve urgently. Summary of the invention
[0005] Based on the above problems, this application provides a method and related products for generating dialogue text based on customer behavior. Through psychological model modeling, it ensures that the simulated customer behavior is consistent with the results of psychological research and close to the reaction patterns of real users, thereby improving the accuracy of generated dialogue text.
[0006] In a first aspect, an embodiment of the present application provides a method for generating a dialogue text based on customer behavior, comprising:
[0007] Get the current conversation text initiated to the target person;
[0008] Analyze the character settings and historical dialogues corresponding to the target character to generate input features;
[0009] The input features are analyzed based on a psychological model to predict a reply content text of the target person corresponding to the current conversation text.
[0010] Optionally, the method further includes:
[0011] Based on the reply content text, using the big model technology to generate a future conversation text in response to the reply content text;
[0012] A dialogue text list is generated based on the future dialogue text for sales staff training.
[0013] Optionally, before obtaining the current conversation text initiated to the target person, the method further includes:
[0014] Build a character database;
[0015] The character database includes a target character, and character settings and historical dialogues corresponding to the target character;
[0016] The character setting includes basic information, background experience, values and internal motivations.
[0017] Optionally, the constructing of a character database includes:
[0018] Obtain the character settings and historical conversation records set by technicians for different characters;
[0019] The historical dialogue records corresponding to each character are structured and stored together with the corresponding character and character settings to form a character database.
[0020] Optionally, analyzing the character settings and historical dialogues corresponding to the target character to generate input features includes:
[0021] Based on the character settings and historical dialogues corresponding to the target character, analyzing the relevant factors of the psychological model and extracting input features;
[0022] The input characteristics include audience characteristics, information quality, information sources, and situational factors.
[0023] Optionally, the method further includes:
[0024] The character database is updated based on the dialogue text list.
[0025] In a second aspect, an embodiment of the present application provides a device for generating a dialogue text based on customer behavior, including:
[0026] An acquisition module is used to obtain the current conversation text initiated to the target person;
[0027] An observation module, used to analyze the character settings and historical dialogues corresponding to the target character to generate input features;
[0028] The planning module is used to analyze the input features based on a psychological model and predict the reply content text of the target person corresponding to the current dialogue text.
[0029] Optionally, the device further comprises: an action module;
[0030] The action module is used to generate a future dialogue text in response to the reply content text by using a large model technology based on the reply content text;
[0031] A dialogue text list is generated based on the future dialogue text for sales staff training.
[0032] In a third aspect, an embodiment of the present application provides a device for generating a conversation text based on customer behavior, including:
[0033] Memory for storing computer programs;
[0034] A processor is used to implement the steps of the method for generating a dialogue text based on customer behavior as described above when executing the computer program.
[0035] In a fourth aspect, an embodiment of the present application provides a readable storage medium, characterized in that a computer program is stored on the readable storage medium, and when the computer program is executed by a processor, the steps of the method for generating a conversation text based on customer behavior as described above are implemented.
[0036] It can be seen from the above technical solutions that compared with the prior art, the present application has the following advantages:
[0037] This application first obtains the current conversation text initiated to the target person. Then analyzes the character settings and historical conversations corresponding to the target person to generate input features. Finally, the input features are analyzed based on the psychological model to predict the reply content text corresponding to the current conversation text replied by the target person. In this way, through psychological modeling, it is ensured that the simulated customer behavior is consistent with the results of psychological research and close to the reaction pattern of real users, thereby improving the accuracy of the generated conversation text. BRIEF DESCRIPTION OF THE DRAWINGS
[0038] Figure 1 A flowchart of a method for generating a conversation text based on customer behavior provided in an embodiment of the present application;
[0039] Figure 2 A schematic diagram of an overall solution flow chart provided for an embodiment of the present application;
[0040] Figure 3 A schematic diagram of the structure of a device for generating a conversation text based on customer behavior provided in an embodiment of the present application. DETAILED DESCRIPTION
[0041] As mentioned above, the existing methods for generating dialogue texts have the problem of insufficient accuracy. Specifically, the existing software for generating dialogue texts usually uses preset speech templates, which are very standardized and cannot dynamically respond to the specific situations of sales personnel and simulate the feedback of real customers, resulting in poor accuracy when generating dialogue texts. Sales personnel mechanically memorize dialogue texts with insufficient accuracy, and such training has little effect on actual combat.
[0042] To solve the above problems, the embodiment of the present application provides a method for generating a dialogue text based on customer behavior, comprising: first obtaining a current dialogue text initiated to a target person. Then analyzing the character settings and historical dialogues corresponding to the target person to generate input features. Finally, analyzing the input features based on a psychological model to predict a reply content text corresponding to the current dialogue text replied by the target person.
[0043] In this way, by modeling the psychological model, we can ensure that the simulated customer behavior is consistent with the results of psychological research and close to the reaction patterns of real users, thereby improving the accuracy of the generated dialogue text.
[0044] It should be noted that the method for generating a dialogue text based on customer behavior and related products provided in this application can be applied to the field of artificial intelligence technology. The above is only an example and does not limit the application field of the method for generating a dialogue text based on customer behavior and related products provided in this application.
[0045] In order to make the purpose, technical solution and advantages of the embodiments of the present application clearer, the technical solution in the embodiments of the present application will be clearly and completely described below in conjunction with the drawings in the embodiments of the present application. Obviously, the described embodiments are only part of the embodiments of the present application, not all of the embodiments. Based on the embodiments in the present application, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of this application.
[0046] Figure 1 A flowchart of a method for generating a conversation text based on customer behavior provided in an embodiment of the present application. Figure 1 As shown, a method for generating a dialogue text based on customer behavior provided by an embodiment of the present application may include:
[0047] S101: Acquire the current conversation text initiated to the target person.
[0048] In actual applications, life insurance sales are highly dependent on the performance of agents, and the performance of agents directly affects premium income. Many agents often encounter the problem of difficulty in speaking in actual operations due to lack of practical experience. Therefore, before life insurance sales agents face real customers for sales promotion, they need to practice dialogue to enhance the agents' practical experience and communication skills so that they have something to say to customers. The method provided in the embodiment of the present application can be performed based on training software, and the agent selects the target (training) person in the software. For example, the agent chooses to click on the person A, person B or person C marked on the list, or the software randomly assigns a person to the agent for training. The agent then initiates a dialogue with the target person by typing or voice, and the software obtains the current dialogue text initiated by the agent to the target person.
[0049] S102: Analyze the character settings and historical dialogues corresponding to the target character to generate input features.
[0050] In actual applications, after determining the target person for training with the agent, the software combines the preset character settings and historical conversations corresponding to the target person to generate input features for input into the psychological model for prediction.
[0051] In addition, since the methods of generating input features are not the same, the embodiment of the present application can illustrate a possible generation method.
[0052] In one case, S2: analyzing the character settings and historical dialogues corresponding to the target character to generate input features may specifically include:
[0053] Based on the character settings and historical dialogues corresponding to the target character, analyzing the relevant factors of the psychological model and extracting input features;
[0054] The input characteristics include audience characteristics, information quality, information sources, and situational factors.
[0055] In practical applications, the training software can analyze the relevant factors of the psychological model based on the historical dialogue and character settings of the target person, and then extract the input features of the psychological model. Among them, the input features include but are not limited to audience characteristics, information quality, information sources and situational factors. Audience characteristics include cognitive ability, motivation, personality and emotions; information quality includes information volume, complexity, relevance and presentation methods (positive, negative, two-way discussion); information sources include attractiveness, credibility, authority, professionalism, etc.; situational factors include media (face-to-face, telephone, online chat), environmental noise, time pressure and social influence (peer opinions, social identity).
[0056] S103: Analyze the input features based on a psychological model to predict a reply text corresponding to the current conversation text replied by the target person.
[0057] In actual applications, the software uses a variety of pre-integrated psychological models to analyze the multi-dimensional features of the input, and predicts the target person's reply content text corresponding to the current dialogue text. In an embodiment of the present application, the psychological models integrated in the software may include but are not limited to persuasion models, cognitive models, and theory of mind models. Among them, the persuasion model includes, for example, the elaboration possibility model, which can divide the path for individuals to receive information and influence attitude changes into central paths and peripheral paths; cognitive models include, for example, theories about cognitive consistency and cognitive dissonance; and theory of mind models include models that enable individuals to understand the beliefs, intentions, desires, etc. of others.
[0058] In addition, since the methods of generating dialogue text are different, the embodiment of the present application can illustrate a possible generation method.
[0059] In one embodiment, the method further comprises:
[0060] Based on the reply content text, using the big model technology to generate a future conversation text in response to the reply content text;
[0061] A dialogue text list is generated based on the future dialogue text for sales staff training.
[0062] In actual applications, the training software can also analyze the target person's corresponding input features and the predicted target person's reply to the current conversation text, and then use advanced big model technology to generate future conversation texts that conform to the context and character settings for the agent to reply to the target person. In this way, a list of generated conversation texts can be generated for sales staff to conduct memory training. It is understandable that the big model here can not only include existing general models such as ChatGPT, Claude, LLaMA, Tongyi Qianwen, Wenxin Yiyan, etc., but also consider developing customized models specifically optimized for sales scenarios. The generated content must be natural and fluent, while accurately reflecting the strategic direction formulated in the planning stage to ensure that each interaction is close to the behavior patterns of real customers.
[0063] In addition, since data storage methods are not the same, the embodiments of the present application can illustrate a possible storage method.
[0064] In one case, before obtaining the current dialogue text of the target person, the method further includes:
[0065] Build a character database;
[0066] The character database includes a target character, and character settings and historical dialogues corresponding to the target character;
[0067] The character setting includes basic information, background experience, values and internal motivations.
[0068] In actual applications, technicians need to set up each virtual sparring character in advance and enter it into the sparring software to build a character database. It is understandable that the character database includes multiple virtual sparring characters including the target character, and each virtual sparring character is pre-associated with character settings and historical dialogues by the technicians. When the agent chooses to train a virtual sparring character, the software can synchronously call the character settings and historical dialogues corresponding to the virtual sparring task, and then extract the input features of the psychological model for subsequent prediction of the reply content text. The character setting is the character role setting for the virtual sparring character, including some basic information (name, age, gender, occupation, family status, etc.), background experience (education experience, work experience, hobbies, etc.), values (risk tolerance, consumption habits, etc.) and intrinsic motivation (purchase motivation, concerns, expectations, etc.). In addition, the character setting is not limited to surface attributes, but can also explore the customer's psychological characteristics and social relationships in depth, so as to build a comprehensive character image, which is conducive to more accurate dialogue simulation.
[0069] In addition, since there are different ways to construct a character database, the embodiment of the present application can illustrate a possible construction method.
[0070] In one embodiment, the constructing of the character database includes:
[0071] Obtain the character settings and historical conversation records set by technicians for different characters;
[0072] The historical dialogue records corresponding to each character are structured and stored together with the corresponding character and character settings to form a character database.
[0073] In actual applications, among all virtual sparring characters including the target character, each character has a corresponding character setting and historical dialogue. The character setting corresponding to each character needs to be pre-edited by the technician and input into the training software for recording. The corresponding historical dialogue can also be pre-set by the technician, and can also be obtained through all historical dialogue records between the agent and the character generated in the sparring software. The sparring software can structure these historical dialogue records and merge and store them with the corresponding character and character setting to form a character database.
[0074] In addition, since the updating method of the character database is not the same, the embodiment of the present application can illustrate a possible updating method.
[0075] In one embodiment, the method further comprises:
[0076] The character database is updated based on the dialogue text list.
[0077] In actual applications, each time the agent chooses to train with the target person, a new list of dialogue texts may be generated. The training software can process these dialogue text lists to generate new historical dialogues corresponding to the target person, and merge and store them with the corresponding target person and character settings, so as to update the character database and provide more comprehensive analysis data for subsequent dialogue text predictions.
[0078] Figure 2 A schematic diagram of an overall solution flow chart provided in an embodiment of the present application. Figure 2 As shown, in order to more accurately predict the next round of customer responses, the embodiment of the present application defines three core steps, including observation, planning and action. The training software observes the character settings and historical conversations of the target person, and analyzes its psychological model-related factors (such as audience characteristics, information quality, information sources, and situational factors, etc.). Then use psychological models (such as persuasion models, cognitive models, and mental theory models) to plan the next action strategy. The action strategy refers to the action mode taken by the target person in the next step. Since it is a sales dialogue scenario, the action mode is usually verbal behavior, such as acceptance, rebuttal, clarification, or questioning. Then take action, use large model technology to generate dialogue content (reply text content) based on context and strategy and reply.
[0079] In summary, this application first obtains the current dialogue text of the target person. Then, the character settings and historical dialogues corresponding to the target person are analyzed to generate input features. Finally, the input features are analyzed based on the psychological model, and the reply content text corresponding to the current dialogue text is obtained. In this way, through psychological modeling, it is ensured that the simulated customer behavior is consistent with the results of psychological research and close to the reaction pattern of real users, thereby improving the accuracy of the generated dialogue text.
[0080] Figure 3 A schematic diagram of the structure of a device for generating a conversation text based on customer behavior provided in an embodiment of the present application. Figure 3 As shown, the device 300 for generating a dialogue text based on customer behavior provided in an embodiment of the present application includes:
[0081] The acquisition module 301 is used to acquire the current conversation text initiated to the target person;
[0082] An observation module 302 is used to analyze the character settings and historical dialogues corresponding to the target character to generate input features;
[0083] The planning module 303 is used to analyze the input features based on a psychological model and predict the reply content text of the target person corresponding to the current dialogue text.
[0084] As an implementation method, regarding how to generate a conversation text list, the device 300 for generating a conversation text based on customer behavior further includes: an action module;
[0085] An action module, used to generate a future dialogue text in response to the reply content text by using a large model technology based on the reply content text;
[0086] A dialogue text list is generated based on the future dialogue text for sales staff training.
[0087] As an implementation method, with respect to how to store data, the device 300 for generating a conversation text based on customer behavior further includes: a construction module;
[0088] A construction module, used to construct a character database;
[0089] The character database includes a target character, and character settings and historical dialogues corresponding to the target character;
[0090] The character setting includes basic information, background experience, values and internal motivations.
[0091] As an implementation method, regarding how to construct a character database, the above-mentioned construction modules are specifically used for:
[0092] Obtain the character settings and historical conversation records set by technicians for different characters;
[0093] The historical dialogue records corresponding to each character are structured and stored together with the corresponding character and character settings to form a character database.
[0094] As an implementation method, regarding how to generate input features, the observation module 302 is specifically used to: analyze psychological model-related factors based on the character settings and historical dialogues corresponding to the target character, and extract input features;
[0095] The input characteristics include audience characteristics, information quality, information sources, and situational factors.
[0096] As an implementation method, with respect to how to update the character database, the device 300 for generating a dialogue text based on customer behavior further includes: an updating module;
[0097] An updating module is used to update the character database based on the dialogue text list.
[0098] In summary, this application first obtains the current dialogue text of the target person. Then, the character settings and historical dialogues corresponding to the target person are analyzed to generate input features. Finally, the input features are analyzed based on the psychological model, and the reply content text corresponding to the current dialogue text is obtained. In this way, through psychological modeling, it is ensured that the simulated customer behavior is consistent with the results of psychological research and close to the reaction pattern of real users, thereby improving the accuracy of the generated dialogue text.
[0099] In addition, the present application also provides a device for generating a conversation text based on customer behavior, including:
[0100] Memory for storing computer programs;
[0101] A processor is used to implement the steps of the method for generating a dialogue text based on customer behavior as described above when executing the computer program.
[0102] In addition, the present application also provides a readable storage medium, on which a computer program is stored. When the computer program is executed by a processor, the steps of the method for generating a conversation text based on customer behavior as described above are implemented.
[0103] The above description of the disclosed embodiments enables those skilled in the art to implement or use the present application. Various modifications to these embodiments will be apparent to those skilled in the art, and the general principles defined herein may be implemented in other embodiments without departing from the spirit or scope of the present application. Therefore, the present application will not be limited to the embodiments shown herein, but will conform to the widest scope consistent with the principles and novel features disclosed herein.
Claims
1. A method for generating a dialogue text based on customer behavior, characterized in that: The method comprises: Get the current conversation text initiated to the target person; Analyze the character settings and historical dialogues corresponding to the target character to generate input features; The input features are analyzed based on a psychological model to predict a reply content text of the target person corresponding to the current conversation text.
2. The method according to claim 1, characterized in that The method further comprises: Based on the reply content text, using the big model technology to generate a future conversation text in response to the reply content text; A dialogue text list is generated based on the future dialogue text for sales staff training.
3. The method according to claim 1, characterized in that Before obtaining the current conversation text initiated to the target person, the method further includes: Build a character database; The character database includes a target character, and character settings and historical dialogues corresponding to the target character; The character setting includes basic information, background experience, values and internal motivations.
4. The method according to claim 3, characterized in that The constructing of the character database comprises: Obtain the character settings and historical conversation records set by technicians for different characters; The historical dialogue records corresponding to each character are structured and stored together with the corresponding character and character settings to form a character database.
5. The method according to claim 1, characterized in that The analyzing the character settings and historical dialogues corresponding to the target character to generate input features includes: Based on the character settings and historical dialogues corresponding to the target character, analyzing the relevant factors of the psychological model and extracting input features; The input characteristics include audience characteristics, information quality, information sources, and situational factors.
6. The method according to claim 2, characterized in that The method further comprises: The character database is updated based on the dialogue text list.
7. A device for generating a dialogue text based on customer behavior, characterized in that: include: An acquisition module is used to obtain the current conversation text initiated to the target person; An observation module, used to analyze the character settings and historical dialogues corresponding to the target character to generate input features; The planning module is used to analyze the input features based on a psychological model and predict the reply content text of the target person corresponding to the current dialogue text.
8. The device according to claim 7, characterized in that The device further comprises: an action module; The action module is used to generate a future dialogue text in response to the reply content text by using a large model technology based on the reply content text; A dialogue text list is generated based on the future dialogue text for sales staff training.
9. A device for generating a conversation text based on customer behavior, characterized in that: include: Memory for storing computer programs; A processor, configured to implement the steps of the method for generating a dialogue text based on customer behavior as described in any one of claims 1 to 6 when executing the computer program.
10. A readable storage medium, characterized in that: The readable storage medium stores a computer program, which, when executed by a processor, implements the steps of the method for generating a conversation text based on customer behavior as claimed in any one of claims 1 to 6.