Reply method, electronic equipment, program product and storage medium

Abstract

The invention discloses a reply method, electronic equipment, a program product and a storage medium. The method comprises the following steps: acquiring first evaluation information of a first user; the first evaluation information is analyzed through a first analysis model, concern information of the first user is obtained, the concern information comprises overall concern information and N sub concern information, N is a positive integer, the overall concern information comprises overall concern and target information corresponding to the overall concern, and N is a positive integer; the sub-concern information comprises a sub-concern and target information corresponding to the sub-concern; determining first reply link information according to the focus information, wherein the first reply link information comprises the focus information and sequence information of the focus information; and generating first reply information according to the first reply link information, wherein the first reply information is reply information corresponding to the first evaluation information.

Description

Technical Field

[0001] This application relates to the field of artificial intelligence technology, and in particular to a response method, electronic device, program product, and storage medium. Background Technology

[0002] Online customer feedback has become one of the main channels for customer feedback. Quick responses to online customer complaints are beneficial for improving customer experience and customer service quality. The commonly used method for responding to customer feedback is manual replying by customer service representatives, which is time-consuming, labor-intensive, and costly. Limited by the number of human customer service representatives, the processing speed and volume are extremely limited, resulting in poor timeliness. Currently, there are also some automated replies based on predefined templates, which pre-set some general reply content, such as "Hello, thank you for your feedback. We will process your complaint as soon as possible. Please feel free to contact us if needed." This type of automated reply based on predefined text is rigid and uniform, unable to provide personalized responses and rapid processing based on the customer's concerns and satisfaction levels. Therefore, how to achieve comprehensive and efficient responses has become an urgent problem to be solved. Summary of the Invention

[0003] To address the aforementioned technical problems, this application provides a response method, an electronic device, a program product, and a storage medium.

[0004] The response method provided in this application includes: Obtain the first user's initial review information; The first evaluation information is analyzed by the first analysis model to obtain the first user's focus information. The focus information includes overall focus information and N sub-focus information, where N is a positive integer. The overall focus information includes the overall focus and the target information corresponding to the overall focus. The sub-focus information includes the sub-focus and the target information corresponding to the sub-focus. The first response link information is determined based on the focus information, which includes the focus information and the order information of the focus information; The first response information is generated based on the first response link information, and the first response information is the response information corresponding to the first evaluation information.

[0005] The electronic device provided in this application includes: a processor and a memory for storing a computer program capable of running on the processor, wherein the processor is used to call and run the computer program stored in the memory to perform the above-described response method.

[0006] This application provides a computer program product, comprising: a computer program that, when executed by a processor, implements the above-described response method.

[0007] The computer-readable storage medium provided in this application is used to store a computer program that causes a computer to perform the above-described response method.

[0008] In the technical solution of this application, the first user's first evaluation information is obtained; the first evaluation information is analyzed using a first analysis model to obtain the first user's focus information, which includes overall focus information and N sub-focus information, where N is a positive integer. The overall focus information includes the overall focus and the target information corresponding to the overall focus, and the sub-focus information includes the sub-focus and the target information corresponding to the sub-focus. Based on the focus information, a first response link information is determined, which includes the focus information and its order information. Based on the first response link information, a first response information is generated, which is the response information corresponding to the first evaluation information. Thus, the first response link information can be determined based on the first user's focus, and the response information for the first evaluation information can be determined based on the first response link information, enabling responses to the first user's focus information, making the response more efficient and comprehensive, and improving the customer experience. Attached Figure Description

[0009] The accompanying drawings, which are provided to further illustrate this application and form part of this application, illustrate exemplary embodiments of this application and are used to explain this application, but do not constitute an undue limitation of this application.

[0010] Figure 1 This is a flowchart illustrating the response method provided in an embodiment of this application; Figure 2 This is a flowchart illustrating the personalized intelligent response mechanism for customer reviews provided in this application embodiment; Figure 3 This is a schematic diagram of the process for generating personalized response based on the satisfaction level of concerns provided in the embodiments of this application; Figure 4 This is a schematic diagram of the structural composition of the response device provided in the embodiments of this application; Figure 5 This is a schematic structural diagram of an electronic device provided in an embodiment of this application; Figure 6 This is a schematic structural diagram of the chip according to an embodiment of this application. Detailed Implementation

[0011] The technical solutions of the embodiments of this application will now be described with reference to the accompanying drawings. Obviously, the described embodiments are only a part of the embodiments of this application, and not all of them. All other embodiments obtained by those skilled in the art based on the embodiments of this application without creative effort are within the scope of protection of this application.

[0012] In the following description, references are made to “some embodiments,” which describe a subset of all possible embodiments. However, it is understood that “some embodiments” may be the same subset or different subsets of all possible embodiments and may be combined with each other without conflict.

[0013] It should also be noted that the terms "first," "second," and "third" used in the embodiments of this application are only used to distinguish similar objects and do not represent a specific order of objects. It is understood that "first," "second," and "third" can be interchanged in a specific order or sequence where permissible, so that the embodiments of this application described herein can be implemented in an order other than that illustrated or described herein. The term "and / or" in this document is merely a description of the relationship between related objects, indicating that three relationships can exist. For example, A and / or B can represent: A existing alone, A and B existing simultaneously, and B existing alone. Additionally, the character " / " in this document generally indicates that the preceding and following related objects have an "or" relationship. It should also be understood that the "instruction" mentioned in the embodiments of this application can be a direct instruction, an indirect instruction, or an indication of an association relationship. For example, A instructing B can mean that A directly instructs B, for example, B can be obtained through A; it can also mean that A indirectly instructs B, for example, A instructs C, and B can be obtained through C; or it can mean that there is an association relationship between A and B. It should also be understood that the term "correspondence" mentioned in the embodiments of this application may indicate a direct or indirect correspondence between the two, or an association between the two, or a relationship of instruction and being instructed, configuration and being configured, etc.

[0014] To facilitate understanding of the technical solutions of the embodiments of this application, the relevant technologies of the embodiments of this application are described below. The following relevant technologies are optional solutions and can be combined with the technical solutions of the embodiments of this application in any way, and they all fall within the protection scope of the embodiments of this application.

[0015] Online customer feedback has become one of the main channels for customer feedback. Quick responses to online customer complaints are beneficial for improving customer experience and customer service quality. The commonly used method for responding to customer feedback is manual reply by customer service representatives, which is time-consuming, labor-intensive, and costly. Limited by the number of human customer service representatives, the processing speed and volume are extremely limited, resulting in poor timeliness. Currently, there are also some automated replies based on predefined templates, which pre-set some general reply content, such as "Hello, thank you for your feedback. We will process your complaint as soon as possible. Please feel free to contact us if needed." This type of automated reply based on predefined text is rigid and monotonous, unable to provide personalized responses and rapid processing based on the customer's concerns and satisfaction levels. Therefore, how to achieve the most efficient and comprehensive replies becomes a problem that needs to be considered. To this end, the following technical solution is proposed in the embodiments of this application.

[0016] To facilitate understanding of the technical solutions of the embodiments of this application, the technical solutions of this application are described in detail below through specific embodiments. The above-mentioned related technologies are optional solutions and can be arbitrarily combined with the technical solutions of the embodiments of this application, all of which fall within the protection scope of the embodiments of this application. The embodiments of this application include at least some of the following contents.

[0017] Figure 1 This is a flowchart illustrating the response method provided in an embodiment of this application, as shown below. Figure 1 As shown, the response method includes the following steps: Step 101: Obtain the first user's first evaluation information.

[0018] Here, the first evaluation information refers to the evaluation feedback information of the first user. For example, the first evaluation information is the evaluation feedback information of the first user regarding the stay experience.

[0019] Step 102: Analyze the first evaluation information using the first analysis model to obtain the first user's focus information.

[0020] The information on points of interest includes overall information on points of interest and information on N sub-points of interest, where N is a positive integer. The overall information on points of interest includes the overall points of interest and the target information corresponding to the overall points of interest, and the sub-points of interest information includes the sub-points of interest and the target information corresponding to the sub-points of interest.

[0021] In some implementations, the first analysis model is a fine-grained user satisfaction analysis model. Specifically, the first analysis model analyzes the first user's first evaluation information to obtain the first user's focus information. This focus information includes overall focus information and N sub-focus information. The overall focus information includes the overall focus and its corresponding target information, and each sub-focus information includes the sub-focus and its corresponding target information.

[0022] In some implementations, the target information includes one or more of the following: satisfaction information, emotional information, causal information, and impact information.

[0023] In some implementations, satisfaction information refers to satisfaction with overall concerns and / or sub-concerns, such as satisfaction levels of five degrees: very satisfied, satisfied, neutral, dissatisfied, and very dissatisfied. Emotional information includes emotion type information and emotion intensity information. For example, emotion type information includes positive, negative, happy, anxious, irritable, frustrated, angry, sad, melancholic, disgusted, fearful, and indifferent emotions. Emotion intensity information is the strength of the emotion, expressed numerically, for example, on a scale of 1 to 5, where 5 represents very strong, 4 represents strong, 3 represents neutral, 2 represents weak, and 1 represents very weak. The degree, emotion type, and emotion intensity of the satisfaction information can be set according to actual circumstances; this application does not impose specific limitations on this.

[0024] It should be noted that emotional information can also be replaced by emotional index, which includes emotional type information and emotional intensity information. Emotional intensity information can also be replaced by affective intensity information.

[0025] For example, when the first evaluation information is the feedback from the first user regarding their stay experience, the overall focus is "overall," meaning the overall stay experience. Sub-focuses can be rooms, facilities, restaurants, etc. More specifically, sub-focuses can be rooms, facilities (air conditioning), restaurants (service), restaurants (breakfast), etc. For the overall focus and sub-focuses, determine the corresponding satisfaction information, emotion type information, emotion intensity information, reason information, and impact information. For example, this can be represented as: {Focus: Room; Satisfaction: Very Satisfied; Emotion Type: Happy (Emotion Intensity: 5); Reason: Cleanliness; Impact: NA}, where NA indicates that the first evaluation information does not contain impact information for the sub-focus of "Room."

[0026] It should be understood that the overall focus and sub-focuses of the first evaluation information differ depending on the application scenario. The embodiments in this application are merely examples and do not limit the specific content of the first evaluation information.

[0027] In some implementations, the first evaluation information is analyzed by a first analysis model to obtain the first user's concern information, including: determining the first user's overall concern information and N sub-concern information based on the first evaluation information according to the first analysis model.

[0028] In some implementations, when the first evaluation information includes the first user's statement text information on the overall evaluation, the overall concern information is determined based on the first evaluation information. Specifically, the first evaluation information is analyzed by a first analysis model to determine the first user's overall concern information and N sub-concern information.

[0029] For example, the first evaluation information includes statements such as "The overall experience of this stay was good, a pretty good experience" or "This product is terrible". These statements are summary evaluations, and the focus is the overall focus, that is, the overall. The corresponding satisfaction level is the first user's overall satisfaction level.

[0030] In some implementations, the first evaluation information is analyzed using a first analysis model to obtain the first user's focus information, including: determining N sub-focus information based on the first evaluation information using the first analysis model; determining the focus category corresponding to each sub-focus based on the N sub-focus information; and determining the overall focus information based on the sub-focus information of each focus category.

[0031] In some implementations, when the first evaluation information does not contain the first user's statement text information on the overall evaluation, the first evaluation information is first analyzed by the first analysis model to determine N sub-concern information, the concern category corresponding to each sub-concern information is determined based on the N sub-concern information, and the overall concern information is determined based on the sub-concern information of each concern category.

[0032] In some implementations, the average of the satisfaction information of the sub-concerns corresponding to each concern category is taken as the satisfaction information of that concern category, and the overall satisfaction information of the concerns is taken as the average of the satisfaction information of each concern category.

[0033] In some implementations, N sub-concern information is determined based on the first evaluation information according to the first analysis model, including: extracting concern text fragments and satisfaction text fragments from the first evaluation information based on the first analysis model; extracting key semantic elements from the concern text fragments and satisfaction text fragments; constructing concern feature vectors and satisfaction feature vectors based on the key semantic elements; and determining the N sub-concern information of the first user based on the concern feature vectors and satisfaction vectors using an attention mechanism.

[0034] Specifically, a fine-grained user satisfaction analysis model is constructed based on user reviews to conduct fine-grained user satisfaction analysis on user concerns. During model construction, firstly, text fragments describing concerns (i.e., concern text fragments) and satisfaction description fragments (i.e., satisfaction text fragments) are extracted. Then, key semantic elements of the concerns and satisfaction descriptions are extracted, including the lexical semantic features and semantic categories of the concern text fragments; and the lexical semantic distribution features, sentiment distribution features, and semantic dependency distribution features of the satisfaction description text fragments. Based on these key semantic elements, a semantic distribution feature vector of concerns (i.e., concern feature vector) and a semantic distribution feature vector of satisfaction description text fragments (i.e., satisfaction feature vector) are constructed. Through multi-head self-attention mechanisms and cross-text fragment attention mechanisms, a semantic association between concerns and satisfaction descriptions is established, satisfaction and sentiment indices are calculated, and the causes and impacts of satisfaction are extracted.

[0035] Step 103: Determine the first response link information based on the focus information. The first response link information includes the focus information and the order information of the focus information.

[0036] In some implementations, after obtaining the information on points of interest, a first response chain is constructed based on this information. This first response chain includes the information on the points of interest and the order of these points. Thus, the order of responses to each point of interest can be determined based on the first response chain, thereby determining the final first response.

[0037] In some implementations, determining the first response link information based on the concern information includes: determining empathic concern information based on the overall concern information and N sub-concern information; determining a first order of empathic concerns based on the empathic concern information; and determining the first response link information based on the first order and the concern information.

[0038] In some implementations, during the process of constructing the first response link information, empathic concern information is first determined from the overall concern information and N sub-concern information, that is, empathic concern is selected from the overall concern and N sub-concerns. The first order of empathic concern is determined based on the empathic concern information, and the first response link information is determined based on the first order and the concern information.

[0039] In some implementations, the empathy attention information includes satisfaction information, emotion intensity information, and emotion type information. Determining the first order of empathy attention points based on the empathy attention point information includes: obtaining weight information corresponding to the satisfaction information of each empathy attention point; obtaining emotion influence factors corresponding to the emotion type information of each empathy attention point; determining the priority value of each empathy attention point based on the weight information, emotion intensity information, and emotion influence factors; and determining the first order of empathy attention points based on the priority value of each empathy attention point.

[0040] In some implementations, the empathy concern information includes satisfaction information, emotion intensity information, and emotion type information corresponding to each empathy concern. First, the weight information corresponding to the satisfaction information of each empathy concern is obtained. This weight information is a risk weight for the concern, and different satisfaction information has different risk weights. Therefore, once the empathy concern information is determined, the corresponding risk weight can be determined based on the satisfaction information in the empathy concern information. Different emotion type information corresponds to different emotion influencing factors. Therefore, the emotion influencing factor corresponding to the emotion type information of each empathy concern can be obtained. The priority value corresponding to each empathy concern is determined based on the weight information, emotion intensity information, and emotion influencing factor. The first order of empathy concerns is determined based on the priority value of each empathy concern.

[0041] In some implementations, the priority values ​​of each empathy concern are sorted from high to low to determine the first order of empathy concerns.

[0042] In some implementations, the correspondence between satisfaction information and weight information can be as follows: the weight information corresponding to empathic concerns with satisfactory satisfaction is 1, and the risk weight for empathic concerns with unsatisfactory satisfaction is predefined based on the potential danger level, with such weight information being greater than 1. The correspondence between emotion type and emotion influencing factors can be as follows: the weight of emotion type is predefined based on the application scenario, with negative emotion types having a higher weight than positive emotion types.

[0043] It is understandable that the empathic concern information for satisfaction here includes both satisfied and very satisfied empathic concern information, that is, satisfaction information belonging to the broad category of "satisfied" in the specific classification. Similarly, the empathic concern information for dissatisfaction here includes both dissatisfied and moderately dissatisfied empathic concern information, that is, satisfaction information belonging to the broad category of "dissatisfied" in the specific classification.

[0044] In some implementations, the priority value of each empathy concern is determined based on weight information, emotion intensity information, and emotion influencing factors, including: the product of weight information, emotion intensity information, and emotion influencing factors is the priority value.

[0045] In some implementations, the first priority principle is determined as follows: unsatisfactory concerns take precedence over satisfactory concerns. Unsatisfactory concerns are sorted according to priority values, and concerns with higher priority values ​​are responded to first. Overall concerns take precedence over sub-concerns.

[0046] For example, the order of concerns in the first response chain is as follows: overall concerns, sub-concerns with higher emotional intensity among very dissatisfied concerns, sub-concerns with lower emotional intensity among dissatisfied concerns, sub-concerns with higher emotional intensity among very satisfied concerns, and sub-concerns with lower emotional intensity among satisfied concerns.

[0047] Step 104: Generate first response information based on the first response link information. The first response information is the response information corresponding to the first evaluation information.

[0048] In some implementations, the first response link information includes concern information and the order information corresponding to the concern. Therefore, based on the order information in the first response link information, response information corresponding to each concern in the first evaluation information is generated, thereby determining the first response information corresponding to the first evaluation information.

[0049] In some implementations, the empathic concerns include a first type of empathic concern and a second type of empathic concern; generating first response information based on first response link information includes: generating initial response information for the empathic concerns based on the first response link information; determining first suggestion information based on the empathic concern information corresponding to the first type of empathic concern; and determining the first response information based on the initial response information and the first suggestion information.

[0050] In some implementations, empathic concerns are categorized into two types based on satisfaction information within the empathic concern information: a first type of empathic concern and a second type of empathic concern. The satisfaction information corresponding to the first type of empathic concern falls under the category of dissatisfaction, while the satisfaction information corresponding to the second type of empathic concern falls under the category of satisfaction. Initial response information is generated for the empathic concerns in the first response chain information. First suggested information is determined for the empathic concern information corresponding to the first type of empathic concern. Finally, the first response information is determined based on the initial response information and the first suggested information.

[0051] In some implementations, generating initial response information for empathic concerns based on first response link information includes: determining a first empathy strategy based on empathy concern information for each first type of empathy concern; determining a first prompt for each first type of empathy concern based on the first empathy strategy and empathy concern information; determining a first sub-response for each first type of empathy concern based on the first prompt for each first type of empathy concern; determining a second empathy strategy based on empathy concern information for a second type of empathy concern; determining a second prompt based on the second empathy strategy and empathy concern information; determining a second sub-response for a second type of empathy concern based on the second prompt; and determining initial response information based on the first sub-response for each first type of empathy concern and the second sub-response for the second type of empathy concern.

[0052] In some implementations, firstly, for a first type of empathic concern, a first empathy strategy is determined, where the first empathy strategy corresponds to a first type of empathic concern. Based on the first empathy strategy and the first type of empathic concern information, a first prompt message is determined for each first type of empathic concern, where the first prompt message is an empathic response constraint semantic prompt. Based on the first prompt message for each first type of empathic concern, a first sub-response message is determined. Secondly, for a second type of empathic concern, a second empathy strategy is determined, where the second empathy strategy corresponds to a second type of empathic concern. Based on the second empathy strategy and the second type of empathic concern information, a second prompt message for the second type of empathic concern is determined, where the second prompt message is an empathic response constraint semantic prompt. Based on the second prompt message for the second type of empathic concern, a second sub-response message for the second type of empathic concern is determined. Finally, initial response information is determined based on the first sub-response message for each first type of empathic concern and the second sub-response message for the second type of empathic concern.

[0053] For example, the first empathy strategy can be an empathy strategy of identification and emotional soothing, and the second empathy strategy can be an empathy strategy of positive identification, gratitude and expectation.

[0054] In some implementations, the first prompt information includes a first type of empathy focus, satisfaction information, emotional information, causal information, impact information, and a first empathy strategy. The second prompt information includes a second type of empathy, satisfaction information, emotional information, causal information, impact information, and a second empathy strategy.

[0055] Understandably, for the first type of empathic concern, a first sub-response message needs to be determined for each first type of empathic concern, and for the second type of empathic concern, an overall response needs to be given for the second type of empathic concern.

[0056] For example, for the first type of empathic concern, the first empathic strategy is the empathic strategy of empathy, identification and emotional soothing. The empathic concern can be represented as: Focus_1: {Concern: Device - Air Conditioner, Satisfaction: Very dissatisfied, Emotional Index: Irritable (5), Reason: Loud noise, Impact: Did not sleep well at night}; The first prompt information can be represented as: Prompts(Focus_1)={Focus: Device - Air Conditioner, Satisfaction: Very dissatisfied, Emotional Index: Irritable (5), Reason: Loud noise, Impact: Did not sleep well at night, Strategy: Empathy - Empathy, Identification and Emotional Soothing}; The first sub-response information generated according to the first prompt information is: Response(text | Prompts)={I'm sorry, the air conditioner noise affected your rest}.

[0057] In some implementations, determining the first suggestion information based on the empathy concern information corresponding to the first type of empathy concern includes: determining the complaint category of each first type of empathy concern based on the empathy concern information corresponding to each first type of empathy concern; determining the rectification measures corresponding to the complaint category of each first type of empathy concern; determining the third prompt information based on the rectification measures and the empathy concern information corresponding to each first type of empathy concern; and determining the first suggestion information based on the third prompt information.

[0058] In some implementations, for each type of empathy concern, it is necessary to determine corresponding first suggestion information. Specifically, based on the empathy concern information corresponding to each type of empathy concern, the complaint category of each type of empathy concern is determined, and the rectification measures corresponding to the complaint category of each type of empathy concern are determined. Based on the rectification measures and the empathy concern information corresponding to each type of empathy concern, third prompt information is determined; and based on the third prompt information, first suggestion information is determined.

[0059] In some implementations, the complaint category is determined based on the semantic category of the concern and the reason for dissatisfaction, and the service manual and reference sample library are consulted to select appropriate rectification measures.

[0060] For example, the empathy concern is the equipment - air conditioner, the corresponding satisfaction information is dissatisfaction, and the reason information is: loud noise; for this empathy concern, the complaint category is equipment failure, and the rectification measure is failure repair. Then the third prompt information can be expressed as: {complaint type: equipment failure, concern: equipment - air conditioner, rectification measure: failure repair, rectification goal: reduce air conditioner noise}. The first suggestion information determined based on the third prompt information can be expressed as: {we will arrange personnel to carry out failure repair as soon as possible to reduce air conditioner noise}.

[0061] In some implementations, after obtaining the first sub-response information and the first suggestion information for each first type of empathy concern, the final response information, i.e., the first response information, is determined by combining the second sub-response information for the second type of empathy concern. Specifically, the empathy language (i.e., sub-response information) and rectification measures (i.e., the first suggestion information) for each concern are merged and spliced ​​together to generate a complete empathy response reply (i.e., the first response information).

[0062] The technical solution of this application embodiment obtains first evaluation information from a first user; analyzes the first evaluation information using a first analysis model to obtain the user's focus information, which includes overall focus information and N sub-focus information, where N is a positive integer. The overall focus information includes the overall focus and the target information corresponding to the overall focus, and the sub-focus information includes the sub-focus and the target information corresponding to the sub-focus; determines first response link information based on the focus information, which includes the focus information and its order; and generates first response information based on the first response link information, which is the response information corresponding to the first evaluation information. Thus, the first response link information can be determined based on the user's focus, and the response information for the first evaluation information can be determined based on the first response link information, enabling responses to the user's focus information more efficient and comprehensive, thereby improving customer experience.

[0063] The technical solutions of the embodiments of this application are illustrated below with specific application examples.

[0064] Online customer feedback has become one of the main channels for customer feedback. Quick responses to online customer complaints improve customer experience and enhance customer service quality. The most common method for responding to customer feedback is manual replying by customer service representatives, which is time-consuming, labor-intensive, and costly. Limited by the number of human representatives, the processing speed and volume are extremely limited, resulting in poor timeliness. Currently, there are also some automated replies based on predefined templates, which pre-set some generic response content, such as "Hello, thank you for your feedback. We will process your complaint as soon as possible. Please feel free to contact us if needed." This type of automated reply based on predefined text is rigid and monotonous, failing to provide personalized responses and rapid processing based on the customer's concerns and satisfaction levels.

[0065] Based on this, this application proposes a personalized intelligent response method for customer service evaluation. By deeply analyzing and understanding customer feedback, it extracts customer concerns and satisfaction levels, generates fine-grained personalized empathetic responses that address these concerns, and generates rectification suggestions for the parts that users are dissatisfied with and submits them to the relevant management departments. Figure 2 This is a flowchart illustrating the personalized intelligent response mechanism for customer reviews provided in this application embodiment, as shown below. Figure 2As shown, the process mainly includes the following three key steps: 1) Multi-level user satisfaction analysis: Based on customer feedback and evaluation, conduct user satisfaction analysis focusing on key areas, calculate overall user satisfaction, and based on this, determine each customer's focus and the corresponding target information; 2) Construction of an empathetic response link based on focus satisfaction: That is, determine the overall satisfaction based on the satisfaction of each focus, determine each empathetic focus from multiple focuses based on the overall satisfaction and the satisfaction of each focus, and determine the priority of each empathetic focus based on the target information of each empathetic focus; 3) Generation of personalized response based on focus satisfaction: The system generates personalized empathic responses based on satisfaction levels with key concerns, provides improvement suggestions based on satisfaction levels with key concerns, and generates personalized, fine-grained empathic responses. Specifically, it generates fine-grained, personalized empathic responses based on various target information related to the empathic concerns. When the satisfaction level for an empathic concern is unsatisfactory, improvement feedback is generated based on the reasons for the concern. The fine-grained, personalized empathic responses and improvement feedback are then merged to obtain a complete empathic response for each concern. Finally, based on the priority of each concern, the complete empathic responses for all concerns are integrated into a target empathic response and then fed back to the customer. A detailed description follows: Input: Customer Feedback (Comments, equivalent to the first review mentioned above) = {Overall, the stay was good, and the room was very clean. However, the air conditioner was extremely noisy, and I didn't sleep well last night. This morning at the restaurant, the staff were very welcoming, but the breakfast selection was limited and none of the dishes were particularly appealing.} Output: Personalized empathic responses (equivalent to the first response information mentioned above).

[0066] The key steps are as follows: (I) Multi-level user satisfaction analysis A fine-grained user satisfaction analysis model (equivalent to the first analysis model mentioned above) is constructed based on user reviews to conduct fine-grained user satisfaction analysis on user concerns. During model construction, firstly, text fragments describing concerns and satisfaction are extracted; then, key semantic elements of the concerns and satisfaction descriptions are extracted, including the lexical semantic features and semantic categories of the concern text fragments; and the lexical semantic distribution features, sentiment distribution features, and semantic dependency relationship distribution features of the satisfaction text fragments. Based on these key semantic elements, semantic distribution feature vectors of concern and satisfaction text fragments are constructed. Through multi-head self-attention and cross-text fragment attention mechanisms, the semantic association between concerns and satisfaction descriptions is established, satisfaction and sentiment indices are calculated, and the causes and impacts of satisfaction are extracted.

[0067] The focus-satisfaction ratio can be expressed as: Satisfaction (focus) := {focus, satisfaction, emotional index, reason / impact}, which is equivalent to the aforementioned focus information; Satisfaction level includes five levels, which can be expressed as: Satisfaction = {Very Satisfied | Satisfied | Neutral | Dissatisfied | Very Dissatisfied}; The emotion index (equivalent to the aforementioned emotion information) includes emotion type (equivalent to the aforementioned emotion type information) and emotion intensity (equivalent to the aforementioned emotion intensity information).

[0068] For example, emotion types = {positive | negative | happy | anxious | irritable | frustrated | angry | sad | depressed | disgusted | fearful | indifferent | ...} etc.

[0069] Emotional intensity: The degree of intensity of the emotion (1-5 points, where 5 points is very intense, 4 points is intense, 3 points is moderate, 2 points is weak, and 1 point is very weak). 1. Fine-grained satisfaction analysis based on focus points Extract text snippets of user concerns and satisfaction descriptions, use a fine-grained user satisfaction analysis model to obtain the correlation between concerns and satisfaction descriptions, calculate user satisfaction and sentiment index for each concern, and identify the reasons for satisfaction / dissatisfaction and their impact on users.

[0070] For example, based on the aforementioned input, the relationship between the focus and satisfaction description in this application embodiment can be expressed as: Satisfy(comments) = { SAT 1: {Focus: Room; Satisfaction: Very Satisfied; Emotional Index: Happy (Emotional Intensity: 5); Reason: Cleanliness; Impact: NA} SAT 2: {Focus: Equipment - Air Conditioning, Satisfaction: Very Dissatisfied, Emotional Index: Irritable (5), Reason: Loud Noise, Impact: Did Not Sleep Well at Night} SAT 3: {Focus: Restaurant-service; Satisfaction: Very satisfied; Emotional index: Happy (emotional intensity: 5); Reason: The waiters were very enthusiastic; Impact: NA} SAT 4: {Focus: Restaurant - Breakfast, Satisfaction: Unsatisfied, Mood Index: Frustrated (Emotional Intensity: 3), Reason: Limited variety; Impact: Lack of palatable food} } 2. Calculation of overall user satisfaction Calculate overall customer satisfaction based on the correlation between concerns and satisfaction levels.

[0071] Scenario 1) Analyze the entire text to extract the user's explicit overall evaluation statements and determine the user's overall satisfaction. For example, "The overall experience of this stay was good, a pretty good experience," or "This product is terrible." These statements are summary evaluations, with the focus being "overall," and the corresponding satisfaction level is taken as the user's overall satisfaction.

[0072] Scenario 2) If there is no explicit overall evaluation, based on the results of fine-grained user satisfaction analysis, determine the corresponding concern category according to the semantic feature vector of the concern. The satisfaction of each concern category is the average of the satisfaction of each concern under that category, and the overall satisfaction is the average of the satisfaction of each concern category.

[0073] It is understandable that the focus information determined by fine-grained satisfaction analysis based on focus points is sub-focus information, while the overall user satisfaction calculation determines the overall focus information.

[0074] (II) Construction of an Empathic Response Link Based on Satisfaction with Focus Points Based on the overall user satisfaction and the fine-grained satisfaction of the focus points {focus points, satisfaction, emotion index, related evaluations}, the focus points with satisfaction values ​​of {very satisfied, satisfied, dissatisfied, very dissatisfied} are extracted as empathy focus points, and the focus points are sorted according to their satisfaction levels to construct an empathy response chain (equivalent to the first response chain information mentioned above).

[0075] The specific empathic response chain can be represented as follows: Emotion-Chain(sentiments):={ Focus_1, Focus_2,….., Focus_i,…} Focus_i = {Focus points, satisfaction levels, sentiment index, reasons and impacts} Among them, the principle for calculating the priority of empathy concerns is as follows: 1) Dissatisfaction-related concerns > Satisfaction-related concerns 2) Dissatisfaction points are prioritized according to their emotional index, and those with higher emotional indices are given priority for empathetic responses.

[0076] The priority of empathy concerns is calculated as follows: ResponsePriority(focus_i, satisfy)= Risk(focus_i, satisfy)*Risk(emotion) Risk(emotion)= Factor(emotion)*Strength(emotion) Wherein, Risk(focus_i, satisfaction) represents the risk weight of the focus point, with a risk weight of 1 for a satisfactory focus point. The risk weight of each unsatisfactory focus point is predefined based on the degree of potential danger it may cause. In the specific implementation, the risk weight of an unsatisfactory focus point is greater than 1. Risk(emotion) represents the emotional risk, describing the potential risk of the current emotion. It is calculated based on the predefined emotion type influencing factor Factor(emotion) and emotion intensity Strength(emotion). Factor(emotion) is the emotion influencing factor, describing the potential danger of the emotion type. The weight of the emotion type is predefined according to the application scenario, with the weight of negative emotion types being higher than that of positive emotion types. Strength(emotion) represents the emotion intensity, ranging from 1 to 5 points.

[0077] Based on this, given the above satisfaction analysis results (Sats), the following empathic response chain can be constructed using ResponsePriority calculation: Emotion-Chain (Sats) = { Overall: {Focus: overall, Satisfaction: satisfied, Emotional Index: happy (4), Reason: overall not bad, Impact: will come again next time} Focus_1: {Focus: Equipment - Air Conditioner, Satisfaction: Very Dissatisfied, Emotional Index: Irritable (Emotional Intensity: 5), Reason: Loud Noise, Impact: Didn't Sleep Well at Night} Focus_2: {Focus point: Restaurant - Breakfast, Satisfaction: Unsatisfied, Emotional index: Frustrated (emotional intensity: 3), Reason: Limited variety; Impact: Lack of palatable food} Focus_3: {Focus point: Restaurant service, Satisfaction level: Very satisfied, Emotional index: Happy (emotional intensity: 5), Reason: The waiters were very enthusiastic; Impact: NA} Focus_4: {Focus: Room; Satisfaction: Very Satisfied; Emotional Index: Happy (Emotional Intensity: 5); Reason: Cleanliness; Impact: NA} } (III) Generation of personalized responses based on empathy and satisfaction Given an empathic response chain (Emotion-chain), based on the satisfaction level, sentiment index, reasons, and impact of each empathic concern, a fine-grained personalized empathic response is generated, focusing on the satisfaction level of each concern. For areas of user dissatisfaction, appropriate rectification strategies are selected from the service manual and reference sample library, generating personalized rectification feedback. Figure 3 This is a schematic diagram of the process for generating personalized response based on satisfaction with key concerns, as provided in the embodiments of this application. Figure 3As shown, it mainly includes three steps: 1) generating personalized empathetic responses based on satisfaction with the focus; 2) generating improvement suggestions based on satisfaction with the focus; and 3) generating fine-grained personalized empathetic responses. A detailed description follows: 1. Personalized empathic dialogue generation based on satisfaction with key concerns Based on the results of fine-grained user satisfaction analysis, personalized empathetic responses are generated. Based on multi-dimensional semantic constraints including user concerns, satisfaction levels, sentiment indices, and reasons / impacts, a semantic constraint mechanism for generating empathetic text based on user concerns and satisfaction levels is constructed. Around these user concerns, and based on user satisfaction and sentiment indices, personalized empathetic responses are generated addressing the reasons for and potential impacts of user satisfaction. The specific steps are as follows: 1.1 Selection of Empathy Strategies Based on User Satisfaction Based on the satisfaction and emotional index of the empathy focus, an empathy strategy is determined. If the user is dissatisfied, an empathy strategy of empathy, identification, and emotional soothing is adopted (equivalent to the first empathy strategy mentioned above). If the user is satisfied, an empathy strategy of positive identification, gratitude, and expectation is adopted (equivalent to the second empathy strategy mentioned above).

[0078] 1.2 Generation of semantic prompts based on empathy constraints and satisfaction with attention points This involves constructing semantic prompts to constrain empathic text generation. Based on the satisfaction level of the focus and the empathy strategy, semantic prompts constraining empathic responses are generated, which mainly include the focus, satisfaction level, sentiment index, reason, impact, and empathy strategy. Specifically, this can be represented as: Prompts(focus) = {Focus, Satisfaction, Sentiment Index, Reasons, Impact, Strategy} Strategy = {Empathy - Compassion, Identification, Emotional Soothing | Empathy - Positive Identification, Gratitude, Expectation | ...} Specifically, for unsatisfactory empathy focuses, Prompts(focus) = {focus, satisfaction, emotional index, reason, impact, empathy strategy: empathy, identification, emotional soothing}; for satisfactory empathy focuses, Prompts(focus) = {focus, satisfaction, emotional index, reason, impact, empathy strategy: identification, gratitude}.

[0079] 1.3 Generation of Empathic Scripts Targeting Satisfaction with Key Concerns Generate an empathic response response based on the semantic prompts of the empathic response constraint.

[0080] Response(focus) = TextGeneration(Text|Prompts) = {Empathic Response Script} This includes two scenarios: one is the generation of empathetic and compassionate responses addressing dissatisfaction with a focus, and the other is the generation of positive and grateful empathetic responses addressing satisfaction with a focus. Therefore, the process for generating empathetic response responses based on different levels of user satisfaction is as follows: User satisfaction focus: Empathetic responses that emphasize positive recognition, gratitude, and expectations are generated to address all user satisfaction concerns. First, all satisfaction concerns are extracted, and empathetic constraint semantic prompts (Prompts) are constructed based on satisfaction level, sentiment index, and reasons / impacts. Then, an empathetic response summary emphasizing positive recognition, gratitude, and expectations (equivalent to the aforementioned second response sub-information) is generated based on the Prompts.

[0081] For example, the process of generating personalized empathetic dialogues focusing on user satisfaction in the emotion_chain is as follows: Summary of overall satisfaction descriptions and analysis of all satisfaction concerns: SatisfiedSet = { Overall: {Focus: overall, Satisfaction: satisfied, Emotional Index: happy (4), Reason: overall not bad, Impact: will come again next time} Focus_3: {Focus point: Restaurant service, Satisfaction level: Very satisfied, Emotional index: Happy (emotional intensity: 5), Reason: The waiters were very enthusiastic; Impact: NA} Focus_4: {Focus: Room; Satisfaction: Very Satisfied; Emotional Index: Happy (Emotional Intensity: 5); Reason: Cleanliness; Impact: NA} 1) Constructing empathic constraint semantic prompts: Prompts ( SatisfiedSet ) = {Focus: Overall | Room | Service, Satisfaction: Satisfied | Very Satisfied | Very Satisfied, Mood: Happy (5), Reason / Impact: Will come again, Strategy: Empathy - Positive Recognition, Gratitude, Expectation} 2) Generate an empathic response summary based on the empathy constraints (Prompts): Response (text | Prompts) = {Thank you very much for your recognition. We are glad you liked this room type. Thank you for your appreciation of our service. Welcome to visit us again.} User dissatisfaction focuses on: Empathic and emotionally resonant responses are provided for each point of user dissatisfaction using a focus-oriented approach. Given a point of dissatisfaction, empathic semantic constraints (Prompts) are constructed based on the point of dissatisfaction, satisfaction level, emotional index, and cause / impact. Then, an empathic and resonant response is generated based on these Prompts.

[0082] For example, the process of generating personalized empathic messages for the dissatisfaction focus_1 is as follows.

[0083] Focus_1: {Focus: Equipment - Air Conditioner, Satisfaction: Very dissatisfied, Emotional Index: Irritable (5), Reason: Loud noise, Impact: Did not sleep well at night} 1) Constructing empathic constraint semantic prompts: Prompts (Focus_1) = {Focus: Equipment - Air Conditioner, Satisfaction: Very Dissatisfied, Emotional Index: Irritable (5), Reason: Loud Noise, Impact: Didn't Sleep Well at Night, Strategy: Empathy - Empathy, Identification, Emotional Soothing} 2) Generate personalized empathic messages based on empathy constraints (Prompts) and focus on key concerns. Response(text | Prompts) = {Sorry, the air conditioning noise disturbed your rest} For example, the process of generating personalized empathic dialogue for the Focus_2 dissatisfaction focus is as follows: Focus_2: {Focus point: catering - breakfast, satisfaction: unsatisfied, mood index: frustrated (3), reason: limited variety; impact: lack of delicious food} 1) Constructing empathic constraint semantic prompts: Prompts (Focus_2) = {Focus: Catering - Breakfast, Satisfaction: Unsatisfied, Emotional Index: Frustrated (3), Reason: Limited variety; Impact: No palatable food, Strategy: Empathy - Empathy, Identification, Emotional Soothing} 2) Generate personalized reassurance and empathy messages based on empathy constraints (Prompts) and address specific concerns. Response (text | Prompts) = {Sorry, we didn't provide a decent breakfast, which may have disappointed you.} 2. Generation of rectification suggestions based on satisfaction with key concerns Based on empathetic concerns regarding user dissatisfaction, generate corrective action feedback according to the reasons for user dissatisfaction. The main steps include: 1) Construct semantic prompts for rectification requests: Based on the semantic categories of concerns and the reasons for dissatisfaction, the complaint classification is determined. The service manual and reference sample library are consulted to select appropriate rectification measures. Based on the reasons for user dissatisfaction, semantic prompts for rectification requests are constructed through reverse engineering. ReqPrompts={Complaint type, concerns, corrective measures, corrective objectives} 2) Personalized rectification feedback generation based on concerns: Generate personalized rectification suggestions (Request) based on semantic prompts for rectification requests (such as complaint type, rectification measures and rectification goals).

[0084] Example: Focus_1 = {Focus: Equipment - Air Conditioner, Satisfaction: Unsatisfied, Reason: Loud Noise} The process of generating a personalized rectification feedback request is as follows: 1) Construct semantic prompts for rectification requests (ReqPrompts): Complaint type (Focus_1) = Equipment malfunction; Corrective action: Fault repair. ReqPrompts_1 = ReqPrompts(Focus_1, Complaint Type: Equipment Failure) = {Complaint Type: Equipment Failure, Focus: Equipment - Air Conditioner, Corrective Measures: Failure Repair, Corrective Goal: Reduce Air Conditioner Noise} 2) Generate personalized rectification feedback based on the semantic prompts in the rectification request (ReqPrompts): Request(Focus_1) = Request(text | ReqPrompts_1) = {We will arrange for personnel to repair the fault and reduce the air conditioner noise as soon as possible} For example, Focus_2 = {Focus: Food & Beverage - Breakfast, Satisfaction: Unsatisfied, Reason: Limited variety; Impact: Lack of palatable options}, the personalized improvement suggestion request generation process is as follows: 1) Construct semantic prompts for rectification requests (ReqPrompts): Complaint type (Focus_2) = Food quality; Corrective measures: Improve food quality. ReqPrompts_2 = ReqPrompts(Focus_2, Complaint Type: Food and Beverage Quality) = {Complaint Type: Food and Beverage Quality, Focus: Food and Beverage - Breakfast, Corrective Measures: Improve Food and Beverage Quality, Goal: Increase Breakfast Items} 2) Generate personalized rectification feedback suggestions based on the rectification semantic constraints ReqPrompts: Request(Focus_2) = Request(text | Req Prompts_2) = {We will improve the quality of our food and beverage offerings and increase the variety of breakfast items} 3. Fine-grained personalized empathic responses are integrated and generated. By integrating and combining empathetic language and corrective measures that address specific concerns, a complete empathetic response can be generated.

[0085] Response={Response_i, Request_j}={ Thank you so much for your recognition. We're so glad you liked the room type. Thank you for your appreciation of our service. We welcome you to come again next time.

[0086] We apologize that the air conditioner noise is disturbing your rest. We will arrange for someone to repair the air conditioner immediately to reduce the noise.

[0087] We apologize for the disappointment of not having a decent breakfast. We will improve the quality of our meals and increase the variety of breakfast options. The technical solution of this application addresses the technical problems of existing technologies that use manual customer responses or automatic responses based on predefined templates to reply to user complaints. However, manual responses are inefficient and costly, and automatic responses based on predefined templates have limited content and cannot provide personalized responses based on the customer's concerns and satisfaction levels. This method provides a personalized intelligent response method for customer feedback. This method analyzes customer feedback to determine each customer's concerns and corresponding target information (such as satisfaction, sentiment index, cause / impact). It determines the overall satisfaction level based on the satisfaction level of each concern, identifies empathetic concerns from multiple concerns based on the overall satisfaction and the satisfaction levels of each concern, generates fine-grained personalized empathetic responses based on the target information of each empathetic concern, and generates rectification feedback based on the cause of the empathetic concern when the satisfaction level of the empathetic concern is unsatisfactory. The fine-grained personalized empathetic responses and rectification feedback are merged to obtain a complete empathetic response for the empathetic concern, and then integrated into a target empathetic response before being fed back to the customer. Specifically, this method is manifested in: 1. Feedback to customers is integrated into a target empathic response by combining complete empathic responses for each empathic concern. By extracting each empathic concern, responses are made separately for each empathic concern, addressing each of the customer's specific questions in turn, thereby improving the comprehensiveness of the response.

[0088] 2. The specific method for determining the target empathic response is as follows: By analyzing customer feedback, we identify each customer's concern and the corresponding target information. Based on the satisfaction level of each concern, we determine the overall satisfaction. Then, based on the overall satisfaction and the satisfaction level of each concern, we identify individual empathic concerns from among multiple concerns. Based on the target information of each empathic concern, we generate fine-grained, personalized empathic responses tailored to the empathic concerns. When the satisfaction level of an empathic concern is unsatisfactory, we generate rectification feedback based on the reasons for the empathic concern. We then integrate the fine-grained, personalized empathic responses and the rectification feedback to obtain a complete empathic response for each empathic concern. Finally, we integrate all the complete empathic response responses for each empathic concern into a target empathic response and feed it back to the customer, improving the efficiency and relevance of the response.

[0089] 3. Determine the priority of each empathy concern based on the target information of each empathy concern, and integrate the complete empathy response responses of each empathy concern into the target empathy response response according to the priority of each empathy concern. Then, feed the response of the concern that the customer cares about most is placed at the beginning of the description of the target empathy response response to improve the customer experience.

[0090] This application proposes a personalized intelligent response mechanism for customer feedback, providing personalized, fine-grained, empathetic responses. This mechanism enhances the personalized response and empathetic capabilities of intelligent customer service, and can quickly generate rectification suggestion requests for submission to relevant departments. This method can significantly reduce customer service management costs for various companies, improve the efficiency of customer feedback processing, enhance emotional resonance with customers, and improve customer experience. The technical solution of this application can be applied to digital human intelligent customer service scenarios. This method can greatly improve the intelligent interaction capabilities of various digital human customer service systems, improve the efficiency of customer feedback processing, enhance emotional resonance with customers, and improve customer experience. This solution is highly efficient and stable, and can be extended to other intelligent interaction scenarios besides digital human customer service.

[0091] Online customer feedback has become one of the main channels for customer feedback, and how to respond and reply quickly is a significant challenge in intelligent customer service applications. This application provides an automated intelligent customer service response solution. Through in-depth analysis and understanding of customer feedback, it extracts customer concerns, satisfaction levels, reasons, and impacts, generating fine-grained, personalized, and empathetic responses tailored to these concerns. For customer dissatisfaction, it generates suggestions for improvement for relevant departments to handle further. This proposal can automatically generate fine-grained, personalized, and empathetic responses, significantly improving the efficiency and quality of automated customer feedback responses, greatly enhancing customer experience and satisfaction, strengthening the competitiveness of intelligent customer service products, and possessing broad application prospects and huge market demand.

[0092] Figure 4This is a schematic diagram of the structural composition of the response device provided in the embodiments of this application, as shown below. Figure 4 As shown, the response device includes: Acquisition unit 401 is used to acquire the first evaluation information of the first user; The processing unit 402 is configured to analyze the first evaluation information using a first analysis model to obtain the first user's focus information, which includes overall focus information and N sub-focus information, where N is a positive integer. The focus information includes the focus and the target information corresponding to the focus. Based on the focus information, the processing unit 402 determines first response link information, which includes the focus information and the order information of the focus information. Based on the first response link information, the processing unit 402 generates first response information, which is the response information corresponding to the first evaluation information.

[0093] In some implementations, the processing unit 402 is used to determine the overall concern information and N sub-concern information of the first user based on the first evaluation information according to the first analysis model.

[0094] In some implementations, the processing unit 402 is configured to determine N sub-concern information based on the first evaluation information according to the first analysis model; determine the concern category corresponding to each sub-concern information according to the N sub-concern information; and determine the overall concern information based on the sub-concern information of each concern category.

[0095] In some implementations, the processing unit 402 is configured to determine empathic concern information based on overall concern information and N sub-concern information; determine a first order of empathic concerns based on the empathic concern information; and determine a first response link information based on the first order and the concern information.

[0096] In some implementations, the empathy attention information includes satisfaction information, emotion intensity information, and emotion type information; the processing unit 402 is used to obtain weight information corresponding to the satisfaction information of each empathy attention point; obtain emotion influence factors corresponding to the emotion type information of each empathy attention point; determine the priority value of each empathy attention point based on the weight information, emotion intensity information, and emotion influence factors; and determine the first order of empathy attention points based on the priority value of each empathy attention point.

[0097] In some implementations, the empathic concerns include a first type of empathic concern and a second type of empathic concern; the processing unit 402 is configured to generate initial response information for the empathic concerns based on the first response link information; determine first suggestion information based on the empathic concern information corresponding to the first type of empathic concern; and determine the first response information based on the initial response information and the first suggestion information.

[0098] In some embodiments, the processing unit 402 is configured to: determine a first empathy strategy based on the empathy concern information of each first type of empathy concern; determine a first prompt message for each first type of empathy concern based on the first empathy strategy and the empathy concern information; determine a first sub-response message for each first type of empathy concern based on the first prompt message for each first type of empathy concern; determine a second empathy strategy based on the empathy concern information of a second type of empathy concern; determine a second prompt message based on the second empathy strategy and the empathy concern information; determine a second sub-response message for a second type of empathy concern based on the second prompt message; and determine initial response information based on the first sub-response message for each first type of empathy concern and the second sub-response message for the second type of empathy concern.

[0099] In some embodiments, the processing unit 402 is configured to determine the complaint category of each first type of empathy concern based on the empathy concern information corresponding to each first type of empathy concern; determine the rectification measures corresponding to the complaint category of each first type of empathy concern; determine third prompt information based on the rectification measures and the empathy concern information corresponding to each first type of empathy concern; and determine first suggestion information based on the third prompt information.

[0100] In some implementations, the processing unit 402 is configured to extract text fragments of concern and text fragments of satisfaction from the first evaluation information based on a first analysis model; extract key semantic elements from the text fragments of concern and text fragments of satisfaction; construct feature vectors of concern and feature vectors of satisfaction based on the key semantic elements; and determine N sub-concern information of the first user based on the feature vectors of concern and the satisfaction vectors using an attention mechanism.

[0101] In some implementations, the target information is characterized by including one or more of the following: satisfaction information, emotional information, causal information, and impact information.

[0102] Those skilled in the art should understand that Figure 4 The functions of each unit in the recovery device shown can be understood by referring to the relevant description of the aforementioned method. Figure 4 The functions of each unit in the recovery device shown can be implemented by a program running on a processor or by specific logic circuits.

[0103] Figure 5 This is a schematic structural diagram of an electronic device 500 provided in an embodiment of this application. Figure 5 The illustrated electronic device 500 includes a processor 510, which can call and run computer programs from memory to implement the methods in the embodiments of this application.

[0104] Optionally, such as Figure 5As shown, the electronic device 500 may further include a memory 520. The processor 510 can retrieve and run computer programs from the memory 520 to implement the methods described in the embodiments of this application.

[0105] The memory 520 can be a separate device independent of the processor 510, or it can be integrated into the processor 510.

[0106] Optionally, such as Figure 5 As shown, the electronic device 500 may also include a transceiver 530, which the processor 510 can control to communicate with other devices. Specifically, it can send information or data to other devices or receive information or data sent by other devices.

[0107] The transceiver 530 may include a transmitter and a receiver. The transceiver 530 may further include antennas, and the number of antennas may be one or more.

[0108] The electronic device 500 can implement the corresponding processes implemented by the response device in the various methods of the embodiments of this application, which will not be described in detail here for the sake of brevity.

[0109] Figure 6 This is a schematic structural diagram of the chip according to an embodiment of this application. Figure 6 The chip 600 shown includes a processor 610, which can call and run computer programs from memory to implement the methods in the embodiments of this application.

[0110] Optionally, such as Figure 6 As shown, chip 600 may further include memory 620. Processor 610 can retrieve and run computer programs from memory 620 to implement the methods described in this embodiment.

[0111] The memory 620 can be a separate device independent of the processor 610, or it can be integrated into the processor 610.

[0112] Optionally, the chip 600 may also include an input interface 630. The processor 610 can control the input interface 630 to communicate with other devices or chips; specifically, it can acquire information or data sent by other devices or chips.

[0113] Optionally, the chip 600 may also include an output interface 640. The processor 610 can control the output interface 640 to communicate with other devices or chips, specifically, to output information or data to other devices or chips.

[0114] This chip can implement the corresponding processes implemented by the response device in the various methods of the embodiments of this application, which will not be described in detail here for the sake of brevity.

[0115] It should be understood that the chip mentioned in the embodiments of this application may also be referred to as a system-on-a-chip, system chip, chip system, or system-on-a-chip, etc.

[0116] It should be understood that the processor in the embodiments of this application may be an integrated circuit chip with signal processing capabilities. In implementation, the steps of the above method embodiments can be completed by integrated logic circuits in the processor's hardware or by software instructions.

[0117] It is understood that the memory in the embodiments of this application can be volatile memory or non-volatile memory, or may include both volatile and non-volatile memory. The non-volatile memory can be read-only memory (ROM), programmable read-only memory (PROM), erasable programmable read-only memory (EPROM), electrically erasable programmable read-only memory (EEPROM), or flash memory. The volatile memory can be random access memory (RAM), which is used as an external cache. By way of example, but not limitation, many forms of RAM are available, such as Static Random Access Memory (SRAM), Dynamic Random Access Memory (DRAM), Synchronous DRAM (SDRAM), Double Data Rate SDRAM (DDR SDRAM), Enhanced Synchronous DRAM (ESDRAM), Synchlink DRAM (SLDRAM), and Direct Rambus RAM (DR RAM). It should be noted that the memory used in the systems and methods described herein is intended to include, but is not limited to, these and any other suitable types of memory.

[0118] It should be understood that the above-described memory is exemplary and not a limiting description. For example, the memory in the embodiments of this application may also be static random access memory (SRAM), dynamic random access memory (DRAM), synchronous dynamic random access memory (SDRAM), double data rate synchronous dynamic random access memory (DDR SDRAM), enhanced synchronous dynamic random access memory (ESDRAM), synchronous link dynamic random access memory (SLDRAM), and direct memory bus RAM (DR RAM), etc. That is to say, the memory in the embodiments of this application is intended to include, but is not limited to, these and any other suitable types of memory.

[0119] This application also provides a computer program product, including a computer program.

[0120] When executed by a processor, the computer program implements the corresponding processes implemented by the response device in the various methods of the embodiments of this application, which will not be described in detail here for the sake of brevity.

[0121] This application also provides a computer-readable storage medium for storing computer programs.

[0122] The computer program causes the computer to execute the corresponding processes implemented by the response device in the various methods of the embodiments of this application, which will not be described in detail here for the sake of brevity.

[0123] Those skilled in the art will recognize that the units and algorithm steps of the various examples described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are implemented in hardware or software depends on the specific application and design constraints of the technical solution. Those skilled in the art can use different methods to implement the described functions for each specific application, but such implementation should not be considered beyond the scope of this application.

[0124] Those skilled in the art will understand that, for the sake of convenience and brevity, the specific working processes of the systems, devices, and units described above can be referred to the corresponding processes in the foregoing method embodiments, and will not be repeated here.

[0125] In the several embodiments provided in this application, it should be understood that the disclosed systems, apparatuses, and methods can be implemented in other ways. For example, the apparatus embodiments described above are merely illustrative; for instance, the division of units is only a logical functional division, and in actual implementation, there may be other division methods. For example, multiple units or components may be combined or integrated into another system, or some features may be ignored or not executed. Furthermore, the coupling or direct coupling or communication connection shown or discussed may be through some interfaces; the indirect coupling or communication connection between apparatuses or units may be electrical, mechanical, or other forms.

[0126] The units described as separate components may or may not be physically separate. The components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the units can be selected to achieve the purpose of this embodiment according to actual needs.

[0127] In addition, the functional units in the various embodiments of this application can be integrated into one processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit.

[0128] If the aforementioned functions are implemented as software functional units and sold or used as independent products, they can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of this application, in essence, or the part that contributes to the prior art, or a portion of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods described in the various embodiments of this application. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.

[0129] The above description is merely a specific embodiment of this application, but the scope of protection of this application is not limited thereto. Any changes or substitutions that can be easily conceived by those skilled in the art within the scope of the technology disclosed in this application should be included within the scope of protection of this application.

Claims

1. A response method, characterized in that, The method includes: Obtain the first user's initial review information; The first evaluation information is analyzed by the first analysis model to obtain the first user's attention information. The attention information includes overall attention information and N sub-attention information, where N is a positive integer. The overall attention information includes the overall attention and the target information corresponding to the overall attention. The sub-attention information includes the sub-attention and the target information corresponding to the sub-attention. The first response link information is determined based on the concern information, and the first response link information includes the concern information and the order information of the concern information; First response information is generated based on the first response link information, and the first response information is the response information corresponding to the first evaluation information.

2. The method according to claim 1, characterized in that, The step of analyzing the first evaluation information using a first analysis model to obtain the first user's focus information includes: Based on the first analysis model and the first evaluation information, determine the overall focus information and N sub-focus information of the first user; or... Based on the first analysis model, N sub-concern information is determined according to the first evaluation information; based on the N sub-concern information, the concern category corresponding to each sub-concern is determined, and the overall concern information is determined based on the sub-concern information of each concern category.

3. The method according to claim 1, characterized in that, Determining the first response link information based on the concern information includes: Empathic concern information is determined based on the overall concern information and the N sub-concern information; The first order of empathic concerns is determined based on the aforementioned empathic concern information; The first response link information is determined based on the first sequence and the information of concern.

4. The method according to claim 3, characterized in that, The empathy-focused information includes satisfaction information, emotion intensity information, and emotion type information; The step of determining the first order of empathy concerns based on empathy concern information includes: Obtain the weight information corresponding to the satisfaction information of each empathy concern; Obtain the emotion influencing factors corresponding to the emotion type information of each empathic concern; The priority value of each empathy focus is determined based on the weight information, the emotion intensity information, and the emotion influencing factors; The first order of the empathic concerns is determined based on the priority value of each empathic concern.

5. The method according to claim 3, characterized in that, The empathy concerns include a first type of empathy concerns and a second type of empathy concerns; generating the first response information based on the first response link information includes: Initial response information on empathic concerns is generated based on the first response link information; The first suggestion information is determined based on the empathy concern information corresponding to the first type of empathy concern. The first response information is determined based on the initial response information and the first suggestion information.

6. The method according to claim 5, characterized in that, The process of generating initial response information based on the first response link information, including: A first empathy strategy is determined based on the empathy concern information of each first type of empathy concern; a first prompt message is determined based on the first empathy strategy and the empathy concern information; a first sub-response message is determined based on the first prompt message of each first type of empathy concern; A second empathy strategy is determined based on the empathy concern information of the second type of empathy concern; a second prompt message is determined based on the second empathy strategy and the empathy concern information; and a second sub-response message of the second type of empathy concern is determined based on the second prompt message. The initial response information is determined based on the first sub-response information of each first type of empathy concern and the second sub-response information of each second type of empathy concern.

7. The method according to claim 5, characterized in that, The step of determining the first suggestion information based on the empathy concern information corresponding to the first type of empathy concern includes: The complaint category for each type of empathy concern is determined based on the empathy concern information corresponding to each type of empathy concern; Determine the corrective measures corresponding to each complaint category of the first type of empathy concern; The third prompt information is determined based on the rectification measures and the empathy concern information corresponding to each first type of empathy concern. The first suggestion information is determined based on the third prompt information.

8. An electronic device, characterized in that, include: The processor and the memory used to store computer programs that can run on the processor. When the processor is used to run the computer program, it performs the steps of the method as described in any one of claims 1 to 7.

9. A storage medium having a computer program stored thereon, characterized in that, When the computer program is executed by a processor, it implements the steps of the method according to any one of claims 1 to 7.

10. A computer program product, comprising a computer program, characterized in that, When the computer program is executed by a processor, it implements the steps of the method according to any one of claims 1 to 7.

Images