Training pair interaction data generation method, device and equipment and storage medium
By building a customer profile and objection topic database, the system automatically recalls and integrates target objection topics to generate training practice interaction data. This solves the problems of large number of prompt words and high maintenance costs in existing technologies, and achieves efficient and comprehensive training data generation.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2025-12-25
- Publication Date
- 2026-04-07
AI Technical Summary
Existing technologies suffer from problems such as a large number of prompts, high maintenance costs, and poor scalability when generating training and practice interaction data. This is especially true in fields with complex business processes and rules, such as insurance, where it is difficult to achieve efficient generation and full coverage of practice data.
By acquiring audio and video call data, a customer profile and objection topic database are constructed. Based on preset target business scenarios and communication stages, basic objection topics and attribute objection topics are automatically recalled and integrated to generate training and practice interaction data, thereby achieving structured management and systematic generation.
It enables the efficient generation of training practice interaction data, covering objection topics corresponding to various business scenarios, communication stages, and customer attributes, solving the problems of incomplete coverage and high maintenance costs, and improving the scalability and accuracy of training data.
Smart Images
Figure CN121808074A_ABST
Abstract
Description
TECHNICAL FIELD
[0001] The present application relates to the technical field of computer, and particularly relates to a training interactive data generation method and device, equipment and a storage medium. BACKGROUND
[0002] With the development of business skill training in the direction of refinement and scale, the interactive data of training that can truly reflect the interaction process of customers and agents in the business scene has become an important basis for supporting agent skill training and business capability improvement. Especially in the field of insurance and other business processes and rules, the accuracy, consistency and coverage of training interactive data generation in different business scenarios are put forward with higher requirements.
[0003] In recent years, with the rapid development of large language models in the field of natural language understanding and generation, the use of large language models to build intelligent role-playing training has gradually been applied to the training scene. The existing technology mainly prewrites prompt words for the large language model to guide the large language model to simulate the customer role or the agent role, thereby generating interactive data. However, since the prompt words corresponding to different business scenarios need to be customized, there are problems of large number of prompt words, high maintenance cost and poor scalability in the case of a large number of business scenarios, thereby limiting the efficient generation of training interactive data. SUMMARY
[0004] The embodiments of the present application provide a training interactive data generation method, device, equipment and storage medium, which can realize efficient generation of training interactive data.
[0005] In a first aspect, the embodiments of the present application provide a training interactive data generation method, which comprises:
[0006] Obtaining audio and video traffic data, processing the audio and video traffic data to obtain a customer portrait, a basic objection topic database and an attribute objection topic database, wherein the customer portrait includes customer attribute labels, the basic objection topic database includes a business scenario, a communication stage and a basic objection topic, and the attribute objection topic database includes the customer attribute labels and attribute objection topics;
[0007] Based on a preset target business scenario and a preset target communication stage, recalling a basic objection topic corresponding to the target business scenario and the target communication stage from the basic objection topic database;
[0008] Based on the customer attribute labels in the customer portrait, recalling an attribute objection topic corresponding to the customer attribute labels from the attribute objection topic database;
[0009] The basic objection topic and the attribute objection topic are merged to obtain the target objection topic;
[0010] Based on the customer profile and the target objection topics, training and practice interaction data is generated.
[0011] One feasible implementation includes acquiring audio and video call data, processing the audio and video call data to obtain customer profiles, a basic objection topic database, and an attribute objection topic database, including:
[0012] Obtain a customer tag database, which includes multiple customer attribute tags;
[0013] The customer profile is constructed based on the customer attribute tags.
[0014] The audio and video call data are processed by speech-to-text transcription to obtain text call data;
[0015] The text call data is subjected to knowledge extraction processing to obtain the basic objection topic database;
[0016] Based on the customer attribute tags and the basic objection topic database, an attribute objection topic database is constructed.
[0017] One feasible implementation involves acquiring a customer tag database, the customer tag database including multiple customer attribute tags, including:
[0018] Obtain customer policy information data;
[0019] Based on the text call data and the customer policy information data, the customer tag database is obtained.
[0020] One feasible implementation is that the basic objection topic database includes business scenarios, communication stages, objection categories, procedural guidelines, as well as the basic objection topics and basic reference scripts;
[0021] The attribute objection topic database includes the business scenario, communication stage, objection category, procedure guidance, as well as the customer attribute tags, attribute objection topics, and attribute reference scripts.
[0022] One feasible implementation includes recalling basic objection topics corresponding to the target business scenario and the target communication stage from the basic objection topic database based on a preset target business scenario and a preset target communication stage, comprising:
[0023] Based on the preset target business scenario and preset target communication stage, the target business scenario and target communication stage are matched with the business scenarios and communication stages in the basic objection topic database to obtain matching results;
[0024] Based on the matching results, the corresponding basic objection topics are retrieved from the basic objection topic database.
[0025] One feasible implementation, wherein generating training practice interaction data based on the customer profile and the target objection topic, includes:
[0026] Based on the customer profile, the target objection topics, and the target objection reference scripts, training practice interaction data is generated. The target objection reference scripts are obtained by merging the basic reference scripts and the attribute reference scripts.
[0027] One feasible implementation of the method further includes:
[0028] The training practice interaction data is evaluated and scored to obtain the evaluation results of the training practice interaction data.
[0029] Based on the feedback results, the target objection reference script will be updated.
[0030] Secondly, embodiments of this application provide a training practice interactive data generation device, comprising:
[0031] The data acquisition module is used to acquire audio and video call data, process the audio and video call data, and obtain a customer profile, a basic objection topic database, and an attribute objection topic database. The customer profile includes customer attribute tags, the basic objection topic database includes business scenarios, communication stages, and basic objection topics, and the attribute objection topic database includes customer attribute tags and attribute objection topics.
[0032] The first recall module is used to recall basic objection topics corresponding to the target business scenario and the target communication stage from the basic objection topic database based on the preset target business scenario and the preset target communication stage.
[0033] The second recall module is used to recall attribute objection topics corresponding to the customer attribute tags from the attribute objection topic database based on the customer attribute tags in the customer profile.
[0034] The topic fusion module is used to merge the basic objection topic and the attribute objection topic to obtain the target objection topic;
[0035] The data generation module is used to generate training and practice interaction data based on the customer profile and the target objection topics.
[0036] Thirdly, embodiments of this application provide an electronic device, the device including: a processor, a memory, and a system bus;
[0037] The processor and the memory are connected via the system bus;
[0038] The memory is used to store a program, the program including instructions that, when executed by the processor, cause the processor to perform any of the implementation steps of the training practice data generation method described above.
[0039] Fourthly, embodiments of this application provide a computer-readable storage medium for storing a computer program, which, when executed by a terminal device, performs any of the implementation steps of the above-described training data generation method.
[0040] As can be seen from the above technical solutions, the embodiments of this application have the following advantages:
[0041] In this embodiment, firstly, audio and video call data are acquired and processed to obtain a customer profile, a basic objection topic database, and an attribute objection topic database. The customer profile includes customer attribute tags, the basic objection topic database includes business scenarios, communication stages, and basic objection topics, and the attribute objection topic database includes customer attribute tags and attribute objection topics. Secondly, based on a preset target business scenario and target communication stage, basic objection topics corresponding to the target business scenario and target communication stage are retrieved from the basic objection topic database, and attribute objection topics corresponding to the customer attribute tags are retrieved from the attribute objection topic database based on the customer attribute tags in the customer profile. Next, the basic objection topics and attribute objection topics are merged to obtain the target objection topic. Finally, based on the customer profile and the target objection topic, training and practice interaction data is generated.
[0042] As can be seen, this solution processes audio and video call data to construct customer profiles, a basic objection topic database, and an attribute objection topic database. Based on this, it automatically recalls corresponding basic and attribute objection topics according to preset target business scenarios, target communication stages, and customer attribute tags, and merges them to generate target objection topics. By combining customer profiles and target objection topics, this solution not only achieves efficient generation of training and practice interaction data, but also ensures that the generated training and practice interaction data covers basic / attribute objection topics corresponding to various business scenarios, communication stages, and customer attributes, realizing structured management and systematic generation of training and practice interaction data. Compared with existing technologies, this solution effectively solves the problems of incomplete coverage, high maintenance costs, and poor scalability of training and practice interaction data. Attached Figure Description
[0043] Figure 1 A flowchart illustrating a training practice interactive data generation method provided in this application embodiment;
[0044] Figure 2 A schematic diagram of a knowledge extraction process provided in an embodiment of this application;
[0045] Figure 3 A statistical distribution chart of basic objection topics provided for embodiments of this application;
[0046] Figure 4 A recall diagram illustrating a basic objection topic provided for an embodiment of this application;
[0047] Figure 5 A schematic diagram illustrating the recall of an attribute objection topic provided in an embodiment of this application;
[0048] Figure 6 This application provides a schematic diagram of a review process for interactive training data.
[0049] Figure 7 This is a schematic diagram of a training and practice interactive data generation device provided in an embodiment of this application. Detailed Implementation
[0050] As mentioned earlier, with the rapid development of large language models in the fields of natural language understanding and generation in recent years, the use of large language models to build role-playing-based intelligent training partners has been increasingly applied in training scenarios. Existing technologies mainly generate training data by pre-writing prompts for the large language model, guiding it to simulate customer or agent roles. However, because prompts for different business scenarios require customization, the large number of prompts, high maintenance costs, and poor scalability become problematic when there are many business scenarios, thus limiting the efficient generation of training interaction data.
[0051] Furthermore, in the insurance sector, the examination systems for different business scenarios vary significantly. Existing technologies typically generate training practice data on a single training objective basis, lacking unified organization and structured management of the target business scenarios. This makes it difficult to achieve full coverage and orderly presentation of the examination knowledge corresponding to each business scenario, easily leading to fragmented and unfocused training content, which in turn affects agents' systematic mastery of key skills. At the same time, general-purpose large language models are mainly trained based on open-domain data, lacking sufficient understanding of the domain knowledge related to insurance business scenarios, process specifications, and procedural guidelines. This makes the training practice data they generate difficult to adapt to the actual training needs of different business scenarios.
[0052] To address the aforementioned issues, this application provides a method, apparatus, device, and storage medium for generating training and practice interaction data. First, audio and video call data is acquired and processed to obtain a customer profile, a basic objection topic database, and an attribute objection topic database. The customer profile includes customer attribute tags, the basic objection topic database includes business scenarios, communication stages, and basic objection topics, and the attribute objection topic database includes customer attribute tags and attribute objection topics. Second, based on a preset target business scenario and target communication stage, basic objection topics corresponding to the target business scenario and target communication stage are retrieved from the basic objection topic database, and attribute objection topics corresponding to the customer attribute tags are retrieved from the attribute objection topic database based on the customer attribute tags in the customer profile. Next, the basic objection topics and attribute objection topics are merged to obtain target objection topics. Finally, training and practice interaction data is generated based on the customer profile and target objection topics.
[0053] As can be seen, this solution processes audio and video call data to construct customer profiles, a basic objection topic database, and an attribute objection topic database. Based on this, it automatically recalls corresponding basic and attribute objection topics according to preset target business scenarios, target communication stages, and customer attribute tags, and merges them to generate target objection topics. By combining customer profiles and target objection topics, this solution not only achieves efficient generation of training and practice interaction data, but also ensures that the generated training and practice interaction data covers basic / attribute objection topics corresponding to various business scenarios, communication stages, and customer attributes, realizing structured management and systematic generation of training and practice interaction data. Compared with existing technologies, this solution effectively solves the problems of incomplete coverage, high maintenance costs, and poor scalability of training and practice interaction data.
[0054] It should be noted that the embodiments of this application do not limit the executing entity of the training and practice interactive data generation method. For example, the training and practice interactive data generation method of this application embodiment can be applied to information processing devices such as servers or terminal devices. The server can be a standalone server, a cluster server, or a cloud server. The terminal device can be an electronic device such as a smartphone, computer, personal digital assistant (PDA), or tablet computer.
[0055] To make the objectives, technical solutions, and advantages of the embodiments of this application clearer, the technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this application, and not all embodiments. Based on the embodiments of this application, all other embodiments obtained by those of ordinary skill in the art without creative effort are within the scope of protection of this application.
[0056] Figure 1 This is a flowchart illustrating a method for generating interactive training data as provided in an embodiment of this application. (In conjunction with...) Figure 1 As shown, it may include steps S101-S105.
[0057] S101: Obtain audio and video call data, process the audio and video call data to obtain customer profiles, basic objection topic databases, and attribute objection topic databases. The customer profile includes customer attribute tags, the basic objection topic database includes business scenarios, communication stages, and basic objection topics, and the attribute objection topic database includes customer attribute tags and attribute objection topics.
[0058] Figure 2 This is a schematic diagram of a knowledge extraction process provided in an embodiment of this application, combined with... Figure 2 As can be seen, the embodiments of this application first acquire audio and video call data, and then perform automatic speech recognition (ASR) processing on the audio and video call data to transcribe the audio and video call data into initial text data. Subsequently, data governance processing is performed on the initial text data based on a large language model. The data governance processing includes at least role restoration, content optimization, and invalid content filtering to obtain text call data with clear structure and accurate semantics.
[0059] After obtaining text-based call data, a large language model is used to extract knowledge from the data, yielding structured information such as business scenarios, communication stages, objection categories, and basic objection topics. Based on this, the large language model summarizes and derives this structured information according to pre-defined rules and logic, generating corresponding procedural guidelines and basic reference scripts. Simultaneously, an expert verification mechanism is introduced, where business experts review and correct the business scenarios, communication stages, objection categories, basic objection topics, procedural guidelines, and basic reference scripts to ensure the accuracy and business usability of the resulting basic objection topic database.
[0060] Furthermore, by acquiring customer policy information data, which includes at least basic customer information, vehicle information, historical insurance information, and other insurance-related information, a multi-dimensional analysis of customers is conducted based on this policy information data, combined with text call data, to construct a customer tag database. This database includes multiple customer attribute tags, which at least include basic attribute information such as name, age, and gender.
[0061] It should be noted that the basic objection topics in this application embodiment include price-related objections, guarantee-related objections, service-related objections, objections of refusal, public opinion-related objections, information-sensitive objections, and other objection types. Figure 3 A statistical distribution chart of basic objection topics provided for embodiments of this application, combined with Figure 3 It can be seen that price-related objections account for the highest proportion among basic objection topics, exceeding 40%, mainly focusing on issues such as insurance plan prices being higher than last year, disputes over cost-effectiveness, and inconsistent understanding of promotional activity rules. Objections related to refusal and public opinion are also relatively concentrated, totaling 34%. Refusal-related objections mainly reflect customers' refusal or avoidance of further communication, while public opinion-related objections are mostly related to cognitive feedback related to external public information or social opinion. Service-related objections account for 11%, mainly involving questions or dissatisfaction with the service process and experience, such as the timeliness of service response and the complexity of procedures. Objections related to protection, sensitive information, and other objections have a relatively low overall proportion, totaling less than 20%, but they exhibit certain concentration characteristics in specific business scenarios. Sensitive information objections are usually related to concerns about the use of personal information or data security.
[0062] In this embodiment, business scenarios, communication stages, objection categories, basic objection topics, procedural guidelines, and basic reference scripts reviewed and revised by business experts are compiled to construct a basic objection topic database. For example, in the basic objection topic database, the business scenario can be "early warm-up," the communication stage can be "opening remarks," the objection category can be "decline-type objections," the basic objection topic can be "I'm not in a hurry now, I want to wait a bit longer," the procedural guidelines can be "downplay sales intent and attract customers with discounts," and the basic reference script can be "Sir / Madam, it's okay. We called you in advance to give you more options. Locking in your spot now will result in a greater discount later. I'm helping you save money. Let's learn about our promotional activities in advance."
[0063] Furthermore, based on customer attribute tags in the customer tag database and combined with the basic objection topic database, the basic objection topics are extended with attributes to construct an attribute objection topic database. For example, in the attribute objection topic database, the customer attribute tag can be "female," the business scenario can be "early warm-up," the communication stage can be "opening remarks," the objection category can be "decline-type objection," the attribute objection topic can be "My husband is responsible for car insurance, I don't understand it," the procedure guidance can be "1. Express agreement and understanding; 2. Emphasize service upgrades and limited-time offers to attract customers," and the attribute reference script can be "I completely understand what you said. In many families, the husband is responsible for car insurance. The main reason for this contact is that your car is about to renew its insurance. The company has prepared a service upgrade plan for high-quality customers like you, including special value-added services and a fast claims channel. Would you mind providing your husband's mobile phone number? I can call him later to introduce the specific car insurance upgrade plan and lock in this limited-time offer in advance."
[0064] In addition, customer profiles are constructed based on customer attribute tags in the customer tag database. These profiles are used to represent the characteristics of different customer types. Each customer profile includes required customer attribute tags and optional customer attribute tags.
[0065] The required customer attribute tags include name, age, gender (1. Male, 2. Female), vehicle type (1. Gasoline vehicle, 2. New energy vehicle), vehicle brand, renewal method (1. First year renewal, 2. Multi-year renewal, 3. Nearly new vehicle, 4. Agreed-upon renewal, 5. Coverage), customer group (1. Service-sensitive, 2. Price-sensitive, 3. Service user, 4. High intent, 5. Family single-claim extension), and claims history (1. Yes, 2. No). Among these, the customer group is a multi-select tag, while the other required customer attribute tags are single-select tags.
[0066] Optional customer attribute tags include vehicle age (1.1-3 years, 2.3 years and above), vehicle price (below 13,000 RMB, above 23,000 RMB), product combination (1. Single delivery, 2. Delivery to dealership, 3. Vehicle insurance, 4. Vehicle health insurance, 5. Vehicle health insurance, 6. Vehicle health insurance service, 7. Two-wheeled vehicle), and communication goals (1. Full coverage for vehicle price, 2. Additional insurance, 3. Upgrade). Among these, communication goals are multiple-selectable tags, while the other optional customer attribute tags are single-selectable tags.
[0067] Therefore, based on the above processing flow, the customer profile, customer tag database, basic objection topic database, and attribute objection topic database have been constructed, thus laying the foundation for subsequent objection topic recall.
[0068] S102: Based on the preset target business scenario and preset target communication stage, recall the basic objection topics corresponding to the target business scenario and target communication stage from the basic objection topic database.
[0069] In this embodiment of the application, based on the preset target business scenario and target communication stage, the target business scenario and target communication stage are matched with the business scenario and communication stage in the basic objection topic database to obtain the matching result. Based on the matching result, the corresponding basic objection topic is recalled from the basic objection topic database.
[0070] Specifically, Figure 4 This application provides a recall diagram of a basic objection topic, combined with... Figure 4 It can be seen that when the preset target business scenario is "early warm-up" and the preset target communication stage is "opening remarks", "early warm-up" and "opening remarks" are matched and retrieved with the business scenario and communication stage in the basic objection topic database. When the match is successful, the corresponding basic objection topic can be recalled, such as the basic objection topic "I'm not in a hurry now, I want to wait a little longer".
[0071] When the preset target communication stage is "value linking", "predicted hot topics" and "value linking" are matched and searched in the basic objection topic database. When a match is successful, multiple basic objection topics related to the target business scenario and target communication stage can be recalled, such as "Are there regional restrictions on roadside assistance services? How many times can it be used throughout the year?" and "How is the claims service?"
[0072] It should be noted that when the preset target business scenario and preset target communication stage fail to match in the basic objection topic database, this application embodiment can mark the target business scenario and target communication stage as business scenarios and communication stages to be supplemented and trigger the update process of the basic objection topic database to supplement and improve the basic objection topics under the corresponding business scenario and communication stage.
[0073] S103: Based on the customer attribute tags in the customer profile, retrieve the attribute objection topics corresponding to the customer attribute tags from the attribute objection topic database.
[0074] Figure 5 This application provides a schematic diagram illustrating the recall of an attribute objection topic, in conjunction with... Figure 5It can be seen that, based on the customer attribute tags in the customer profile, attribute objection topics corresponding to the customer attributes are retrieved from the attribute objection topic database. Specifically, if the customer attribute tags in the customer profile include [Name = "Zhang San"], [Gender = "Female"], [Age = "25"], [Vehicle Type = "Gasoline Car"], [Vehicle Brand = "BMW X3"], [Renewal Method = "Multi-Year Renewal"], [Customer Group = "Service Usage"], and [Claims Record = "None"], then each of the above customer attribute tags is matched and retrieved with the customer attribute tags in the attribute objection topic database to obtain the attribute objection topics corresponding to each customer attribute tag.
[0075] For example, when the identified business scenario is "early warm-up", the communication stage is "opening remarks", and the objection type is "decline objection", matching the customer attribute tag [gender = "female"] in the customer profile with the customer attribute tags in the attribute objection topic database can recall the corresponding attribute objection topic "My husband is in charge of car insurance, I don't understand it".
[0076] It should be noted that when no attribute objection topic corresponding to a certain customer attribute tag is found in the attribute objection topic database, the customer attribute tag can be marked as an attribute tag to be expanded, and the attribute objection topic database update process can be triggered. Through this method, while ensuring the stability of attribute objection topic retrieval, the coverage and adaptability of the attribute objection topic database to different customer attribute tags can be continuously improved.
[0077] S104: Merge the basic objection topic and the attribute objection topic to obtain the target objection topic.
[0078] In this embodiment of the application, the basic objection topics and attribute objection topics recalled in the aforementioned steps are merged to obtain target objection topics that match the current business scenario, communication stage, and customer attribute tags.
[0079] For example, in the current business scenario of "early warm-up," when the communication stage is "value linking," the basic objections recalled are "Are there regional restrictions on roadside assistance? How many times can it be used throughout the year?" and "How is the claims service?" Simultaneously, when the communication stage is "opening remarks," the attribute objections recalled based on customer attribute tags are "My husband is in charge of car insurance, I don't understand it." By combining and integrating these basic and attribute objections, the target objections can be obtained, for example: [Opening remarks: My husband is in charge of car insurance, I don't understand it; Value linking: Are there regional restrictions on roadside assistance? How many times can it be used throughout the year? How is the claims service?"]
[0080] By integrating the above methods, basic objection topics and attribute objection topics can be combined at different communication stages to generate target objection topics that match the business scenario, communication stage, and customer attribute tags.
[0081] S105: Generate training and practice interaction data based on customer profiles and target objection topics.
[0082] In this embodiment, training and practice interaction data is generated based on customer profiles, target objection topics, and target objection reference scripts. It should be noted that the target objection reference scripts correspond one-to-one with the target objection topics, reflecting complete response examples under the corresponding business scenario, communication stage, objection category, and customer attribute tag. The target objection reference scripts may reference the basic reference scripts corresponding to the basic objection topics and the attribute reference scripts corresponding to the attribute objection topics during the generation process, but their content remains consistent with the target objection topics to ensure the accuracy and relevance of the training and practice interaction data.
[0083] Interactive training data is used to simulate real conversations between customers and agents, enabling agent training and skills enhancement. Specifically, customer profiles are used to simulate the behavioral characteristics of different customers, and target objection topics are used to simulate objection questions that customers may raise in different business scenarios, communication stages, and objection categories.
[0084] In addition, it supports simulation training for various business scenarios and different customer profiles. This application embodiment combines the needs of the insurance and property insurance telemarketing training field and constructs a simulation system based on 6 major business scenarios and 28 communication stages. At the same time, it combines 2 types of vehicles, 3 major channels, 5 types of customer groups, and 7 types of product factors to achieve coverage of customer communication simulation in more than 4,000 different scenarios.
[0085] Specifically, business scenarios, communication stages, and customer attribute tags are arranged and combined to generate corresponding training and practice interaction data. Alternatively, based on the target objection topics obtained from the above integration, and combined with customer profiles, corresponding training and practice interaction data is generated, thus forming training and practice interaction data covering all business scenarios and the entire process of interaction.
[0086] Furthermore, this application embodiment can evaluate and score training practice interaction data to obtain evaluation results, and update the target objection reference scripts based on these evaluation results. Specifically, the training practice interaction data generated based on customer profiles records the multi-round dialogue process between agents and simulated customers. Based on this multi-round dialogue process, this application embodiment performs single-round dialogue evaluation and overall dialogue evaluation on the agent's performance during the dialogue process, so as to realize online feedback and real-time improvement for the agent.
[0087] Figure 6This application provides a schematic diagram of a review process for training and practice interaction data, combined with... Figure 6 It is understood that, based on the recalled target objection topics and target objection reference scripts, a single-round dialogue review is conducted on the agent's response. Specifically, based on the assessment points corresponding to the Standard Operating Procedure (SOP), the agent's response in the single-round dialogue is analyzed and evaluated, generating a single-round dialogue review, including the single-round dialogue score, performance strengths, improvement suggestions, and single-round script recommendations.
[0088] Next, after completing the feedback for each round of dialogue, the results of all rounds of dialogue feedback are summarized and analyzed to form an overall dialogue feedback, thus obtaining the final feedback result. Based on this feedback result, the target objection reference scripts corresponding to the target objection topic can be updated or optimized to ensure that the subsequently generated training practice interaction data can better reflect high-quality and standardized response methods, and continuously improve the training effectiveness and communication skills of the agents.
[0089] It should be noted that, taking outbound call business scenarios as an example, the specific assessment criteria and scoring details of the SOPs corresponding to different communication stages are as follows:
[0090] When the communication phase is the "opening remarks", the assessment criteria are "when the agent introduces himself, his or her speech should include at least the company name, personal position, employee number and agent name". The scoring details are "the opening remarks phase is worth 10 points, the company name is worth 3 points, the personal position is worth 2 points, the employee number is worth 2 points, the agent name is worth 2 points and the fluency of expression is worth 1 point".
[0091] When the communication stage is "before quoting", the assessment criteria are "whether the agent takes the initiative to communicate and interact with the customer, and whether they take the initiative to ask the customer for basic information such as the date of purchase and driving habits to complete the necessary information collection". The scoring details are "the full score for the pre-quoting stage is 10 points. Please make a comprehensive judgment based on the content of the agent's speech in the historical dialogue to give the score".
[0092] When the communication stage is "in the process of quoting", the assessment criteria are "whether the agent provides detailed information on the names of compulsory traffic accident liability insurance, vehicle damage insurance, third-party liability insurance, seat insurance, driving insurance, and health insurance when providing car insurance plans, and whether the corresponding coverage amount and premium for each type of insurance are clearly stated. Among them, insurance products involving medical protection (such as Tai Health Insurance, million-dollar medical insurance, hospitalization medical insurance, etc.) are all classified as health insurance". The scoring details are "the full score for the quoting stage is 20 points. If any type of insurance is missing, or the corresponding coverage amount and premium are not fully explained, points will be deducted according to the preset rules".
[0093] When the communication stage is "post-quotation", the assessment criteria are: "whether the agent promptly closes the deal after completing the quotation; when the customer has not explicitly agreed to the deal in the past conversation, whether the agent proactively provides a reasonable reason for follow-up and schedules a specific follow-up time, while expressing gratitude and saying goodbye politely; when the customer has explicitly agreed to the deal in the past conversation, whether the agent directly expresses gratitude and politely ends the conversation." The scoring details are: "The post-quotation stage is worth 20 points, and the score is determined based on a comprehensive assessment of the agent's remarks."
[0094] Based on the aforementioned steps S101-S105, in this embodiment, firstly, audio and video call data is acquired and processed to obtain a customer profile, a basic objection topic database, and an attribute objection topic database. The customer profile includes customer attribute tags, the basic objection topic database includes business scenarios, communication stages, and basic objection topics, and the attribute objection topic database includes customer attribute tags and attribute objection topics. Secondly, based on preset target business scenarios and target communication stages, basic objection topics corresponding to the target business scenarios and target communication stages are retrieved from the basic objection topic database, and attribute objection topics corresponding to the customer attribute tags are retrieved from the attribute objection topic database based on the customer attribute tags in the customer profile. Next, the basic objection topics and attribute objection topics are merged to obtain target objection topics. Finally, based on the customer profile and target objection topics, training and practice interaction data is generated. As can be seen, this solution processes audio and video call data to construct customer profiles, a basic objection topic database, and an attribute objection topic database. Based on this, it automatically recalls corresponding basic and attribute objection topics according to preset target business scenarios, target communication stages, and customer attribute tags, and merges them to generate target objection topics. By combining customer profiles and target objection topics, this solution not only achieves efficient generation of training and practice interaction data, but also ensures that the generated training and practice interaction data covers basic / attribute objection topics corresponding to various business scenarios, communication stages, and customer attributes, realizing structured management and systematic generation of training and practice interaction data. Compared with existing technologies, this solution effectively solves the problems of incomplete coverage, high maintenance costs, and poor scalability of training and practice interaction data.
[0095] It should be noted that the customer information (including but not limited to customer profiles, customer attributes, etc.) and data (including but not limited to data used for analysis, data stored, data displayed, etc.) involved in this application are all information and data authorized by the user or fully authorized by all parties, and the collection, use and processing of related data must comply with the relevant laws, regulations and standards of the relevant countries and regions.
[0096] Furthermore, Figure 7This is a schematic diagram of a training and practice interactive data generation device provided in an embodiment of this application. (Combined with...) Figure 7 As shown, the training and practice interactive data generation device 700 provided in this application embodiment may include:
[0097] The data acquisition module 701 is used to acquire audio and video call data, process the audio and video call data, and obtain a customer profile, a basic objection topic database, and an attribute objection topic database. The customer profile includes customer attribute tags, the basic objection topic database includes business scenarios, communication stages, and basic objection topics, and the attribute objection topic database includes customer attribute tags and attribute objection topics.
[0098] The first recall module 702 is used to recall basic objection topics corresponding to the target business scenario and the target communication stage from the basic objection topic database based on the preset target business scenario and the preset target communication stage.
[0099] The second recall module 703 is used to recall attribute objection topics corresponding to the customer attribute tags from the attribute objection topic database based on the customer attribute tags in the customer profile.
[0100] The topic fusion module 704 is used to fuse the basic objection topic and the attribute objection topic to obtain the target objection topic;
[0101] The data generation module 705 is used to generate training and practice interaction data based on the customer profile and the target objection topic.
[0102] Optionally, the data acquisition module 701 may include:
[0103] The database acquisition module is used to acquire a customer tag database, which includes multiple customer attribute tags.
[0104] A profile building module is used to build the customer profile based on the customer attribute tags;
[0105] The data transcription module is used to perform speech transcription processing on the audio and video call data to obtain text call data;
[0106] The data extraction module is used to perform knowledge extraction processing on the text call data to obtain the basic objection topic database;
[0107] The attribute library construction module is used to construct an attribute objection topic database based on the customer attribute tags and the basic objection topic database.
[0108] Optionally, the database acquisition module is specifically used for:
[0109] Obtain customer policy information data;
[0110] Based on the text call data and the customer policy information data, the customer tag database is obtained.
[0111] Optionally, the basic objection topic database includes business scenarios, communication stages, objection categories, procedural guidelines, as well as the basic objection topics and basic reference scripts;
[0112] The attribute objection topic database includes the business scenario, communication stage, objection category, procedure guidance, as well as the customer attribute tags, attribute objection topics, and attribute reference scripts.
[0113] Optionally, the first recall module 702 is specifically used for:
[0114] Based on the preset target business scenario and preset target communication stage, the target business scenario and target communication stage are matched with the business scenarios and communication stages in the basic objection topic database to obtain matching results;
[0115] Based on the matching results, the corresponding basic objection topics are retrieved from the basic objection topic database.
[0116] Optionally, the data generation module 704 is specifically used for:
[0117] Based on the customer profile, the target objection topics, and the target objection reference scripts, training practice interaction data is generated. The target objection reference scripts are obtained by merging the basic reference scripts and the attribute reference scripts.
[0118] Optionally, the training and practice interactive data generation device 700 may include:
[0119] The data review module is used to review and score the training practice interaction data to obtain the review results of the training practice interaction data;
[0120] The script update module is used to update the target objection reference script based on the feedback results.
[0121] Furthermore, embodiments of this application also provide an electronic device, including: a processor, a memory, and a system bus;
[0122] The processor and the memory are connected via the system bus;
[0123] The memory is used to store one or more programs, the one or more programs including instructions that, when executed by the processor, cause the processor to perform any of the implementation steps of the training and practice interactive data generation method described above.
[0124] Furthermore, embodiments of this application also provide a computer-readable storage medium for storing a computer program, which, when executed by a terminal device, implements any of the implementation steps of the above-described training and practice interactive data generation method.
[0125] As can be seen from the above description of the embodiments, those skilled in the art can clearly understand that all or part of the steps in the methods of the above embodiments can be implemented by means of software plus necessary general-purpose hardware platforms. Based on this understanding, the technical solution of this application, in essence, or the part that contributes to the prior art, can be embodied in the form of a software product. This computer software product can be stored in a storage medium, such as ROM / RAM, magnetic disk, optical disk, etc., and includes several instructions to cause a computer device (which may be a personal computer, a server, or a network communication device such as a media gateway, etc.) to execute the methods described in various embodiments or some parts of the embodiments of this application. It should be noted that the various embodiments in this specification are described in a progressive manner, and each embodiment focuses on describing the differences from other embodiments. The same or similar parts between the various embodiments can be referred to mutually.
[0126] The system disclosed in the embodiments is described in a relatively simple manner because it corresponds to the method disclosed in the embodiments. For relevant details, please refer to the method section.
[0127] It should also be noted that, in this document, relational terms such as "first" and "second" are used only to distinguish one entity or operation from another, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Furthermore, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such a process, method, article, or apparatus. Without further limitations, an element defined by the phrase "comprising one..." does not exclude the presence of other identical elements in the process, method, article, or apparatus that includes said element.
[0128] The above description of the disclosed embodiments enables those skilled in the art to make or use this application. Various modifications to these embodiments will be readily 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 this application. Therefore, this application is not to be limited to the embodiments shown herein, but is to be accorded the widest scope consistent with the principles and novel features disclosed herein.
Claims
1. A method for generating interactive training data, characterized in that, The method includes: Acquire audio and video call data, process the audio and video call data to obtain customer profiles, basic objection topic databases and attribute objection topic databases, wherein the customer profiles include customer attribute tags, the basic objection topic databases include business scenarios, communication stages and basic objection topics, and the attribute objection topic databases include customer attribute tags and attribute objection topics; Based on the preset target business scenario and preset target communication stage, basic objection topics corresponding to the target business scenario and the target communication stage are recalled from the basic objection topic database; Based on the customer attribute tags in the customer profile, retrieve the attribute objection topics corresponding to the customer attribute tags from the attribute objection topic database; The basic objection topic and the attribute objection topic are merged to obtain the target objection topic; Based on the customer profile and the target objection topics, training and practice interaction data is generated.
2. The method according to claim 1, characterized in that, The process of acquiring audio and video call data, processing the audio and video call data to obtain customer profiles, a basic objection topic database, and an attribute objection topic database includes: Obtain a customer tag database, which includes multiple customer attribute tags; The customer profile is constructed based on the customer attribute tags. The audio and video call data are processed by speech-to-text transcription to obtain text call data; The text call data is subjected to knowledge extraction processing to obtain the basic objection topic database; Based on the customer attribute tags and the basic objection topic database, an attribute objection topic database is constructed.
3. The method according to claim 2, characterized in that, The process of obtaining a customer tag database, which includes multiple customer attribute tags, including: Obtain customer policy information data; Based on the text call data and the customer policy information data, the customer tag database is obtained.
4. The method according to claim 1, characterized in that, The basic objection topic database includes business scenarios, communication stages, objection categories, procedural guidelines, as well as the basic objection topics and basic reference scripts; The attribute objection topic database includes the business scenario, communication stage, objection category, procedure guidance, as well as the customer attribute tags, attribute objection topics, and attribute reference scripts.
5. The method according to claim 1, characterized in that, The step of retrieving basic objection topics corresponding to the preset target business scenario and the preset target communication stage from the basic objection topic database includes: Based on the preset target business scenario and preset target communication stage, the target business scenario and target communication stage are matched with the business scenarios and communication stages in the basic objection topic database to obtain matching results; Based on the matching results, the corresponding basic objection topics are retrieved from the basic objection topic database.
6. The method according to any one of claims 1 to 5, characterized in that, The step of generating training and practice interaction data based on the customer profile and the target objection topics includes: Based on the customer profile, the target objection topics, and the target objection reference scripts, training practice interaction data is generated. The target objection reference scripts are obtained by merging the basic reference scripts and the attribute reference scripts.
7. The method according to claim 6, characterized in that, The method further includes: The training practice interaction data is evaluated and scored to obtain the evaluation results of the training practice interaction data. Based on the feedback results, the target objection reference script will be updated.
8. A training and practice interactive data generation device, characterized in that, include: The data acquisition module is used to acquire audio and video call data, process the audio and video call data, and obtain a customer profile, a basic objection topic database, and an attribute objection topic database. The customer profile includes customer attribute tags, the basic objection topic database includes business scenarios, communication stages, and basic objection topics, and the attribute objection topic database includes customer attribute tags and attribute objection topics. The first recall module is used to recall basic objection topics corresponding to the target business scenario and the target communication stage from the basic objection topic database based on the preset target business scenario and the preset target communication stage. The second recall module is used to recall attribute objection topics corresponding to the customer attribute tags from the attribute objection topic database based on the customer attribute tags in the customer profile. The topic fusion module is used to merge the basic objection topic and the attribute objection topic to obtain the target objection topic; The data generation module is used to generate training and practice interaction data based on the customer profile and the target objection topics.
9. An electronic device, characterized in that, The device includes: a processor, a memory, and a system bus; The processor and the memory are connected via the system bus; The memory is used to store a program, the program including instructions that, when executed by the processor, cause the processor to perform the steps of the training and practice interactive data generation method according to any one of claims 1 to 7.
10. A computer-readable storage medium, characterized in that, The computer-readable storage medium is used to store a computer program, which, when executed by a terminal device, implements the steps of the training and practice interactive data generation method according to any one of claims 1 to 7.