Welcome speech and first contact script intelligent recommendation method, medium and electronic device

Abstract

This application provides an intelligent recommendation method, medium, and electronic device for welcome messages and initial contact scripts. The method includes: acquiring historical chat data, including historical welcome message data and historical initial contact script data from different channels; responding to accessing a chat channel, acquiring high-rated welcome messages based on the historical welcome message data, performing language structure analysis on the high-rated welcome messages to obtain their expression patterns, and generating a welcome message corresponding to the current channel based on these expression patterns; upon receiving user reply chat data regarding the welcome message, acquiring the category and content corresponding to the high-rated initial contact script based on the historical initial contact script data, and recommending the category and content of the high-rated initial contact script to the user for editing the initial contact script corresponding to the current channel. This application can achieve efficient and accurate intelligent recommendation of welcome messages and initial contact scripts.

Description

Technical Field

[0001] This application belongs to the field of artificial intelligence technology, particularly the field of intelligent information recommendation technology related to intelligent investment advisory. Background Technology

[0002] In the investment advisory service sector, sending a welcome message and initial contact script via chat tools is a crucial customer service step when employees first contact clients. The quality of this initial interaction directly impacts the client's trust in the service and the potential for subsequent investment services. Because client needs vary significantly across different channels and chat tools, precise welcome messages and initial contact scripts need to be designed based on channel characteristics and client profiles to effectively drive client conversion. However, in industrial practice, evaluating welcome messages and initial contact scripts faces challenges due to the diversity of domains and tasks. Chat scenarios and data characteristics differ across domains, making it difficult to adapt general methods to the specific needs of the investment advisory field.

[0003] In the existing technology, there are mainly two methods, but both have obvious limitations:

[0004] A single evaluation method based on customer responses judges the effectiveness of a welcome message by the content of the customer's reply; for example, a customer who responds positively is considered a successful welcome. However, this method is limited, ignoring more information that may affect the quality of customer interaction, such as the language composition of the welcome message itself, the time interval between customer responses (e.g., a quick reply may indicate high interest), and whether the customer subsequently purchases investment advisory services (a key indicator). This one-sidedness prevents a comprehensive evaluation of the welcome message's effectiveness, thus limiting optimization potential. Another approach is a segmentation and matching method based on basic customer information. This method segments customers according to their basic information (such as WeChat origin, nickname, wealth value, age group, etc.) and then matches different welcome messages accordingly. This method is more comprehensive than a single evaluation, but it still has problems, ignoring the potential semantic information in chat data and lacking in-depth analysis of the long-term effects of the welcome message (such as service conversion rate).

[0005] The existing patent CN 115630206 A proposes a welcome message matching method that can adaptively match welcome messages based on customer origin. However, this patent's technical solution ignores the impact of the welcome message content itself on customer response behavior, making it difficult to comprehensively evaluate the effectiveness of the welcome message content. It does not quantify the impact of different welcome messages on customer response rates and cannot guide businesses to optimize welcome messages to improve customer engagement. Furthermore, this patent does not incorporate service conversion rate analysis, fails to establish a connection between welcome messages, customer interaction behavior, and the final service conversion rate, and cannot comprehensively assess the actual contribution of welcome messages to business objectives.

[0006] Therefore, it is necessary to design a recommendation method that can adapt to investment advisory service scenarios in order to support efficient and accurate communication with clients. Summary of the Invention

[0007] This application provides a method, medium, and electronic device for intelligently recommending welcome messages and initial contact scripts, which are used to efficiently and accurately recommend welcome messages and initial contact scripts to users.

[0008] To achieve the above and other related objectives, a first aspect of this application provides an intelligent recommendation method for welcome messages and initial contact scripts, comprising: acquiring historical chat data, the historical chat data including historical welcome message data and historical initial contact script data from different channels; responding to accessing a chat channel, acquiring the average response time, response rate, and conversion rate of each welcome message based on the historical welcome message data, and acquiring a comprehensive score for each welcome message based on the average response time, the response rate, and the conversion rate; acquiring high-scoring welcome messages based on the comprehensive score of each welcome message, performing language structure analysis on the high-scoring welcome messages to obtain the expression patterns of the high-scoring welcome messages, and based on the high-scoring welcome messages... The greeting expression pattern generates a welcome message corresponding to the current channel. Upon receiving user reply chat data for the welcome message, the system obtains the average response time, response rate, and conversion rate for each type of initial contact script based on the historical initial contact script data. A comprehensive score is then obtained for each type of initial contact script based on these scores. The system further identifies the category and script content corresponding to high-scoring initial contact scripts based on their comprehensive scores, and recommends these categories and content to the user, allowing the user to edit an initial contact script corresponding to the current channel based on these categories and content.

[0009] In some embodiments of the first aspect of this application, the intelligent recommendation method for welcome messages and initial contact scripts according to claim 1, wherein the average response time for each welcome message is obtained by means of:

[0010]

[0011] Among them, T i Let T be the average response time for each welcome message, T be the set of non-empty response times, min(T) be the shortest response time, and max(T) be the longest response time.

[0012] The response rate for each welcome message is obtained through the following methods:

[0013]

[0014] Among them, R i C represents the response rate for each welcome message. i For the total count of each welcome message, N_C i The number of welcome messages that are not empty in the reply;

[0015] The conversion rate for each welcome message is obtained through the following methods:

[0016]

[0017] Among them, O i For the conversion rate of each welcome message, O_C i The number of service transactions for each welcome message is the number of service transactions where the transaction time is not empty.

[0018] In some embodiments of the first aspect of this application, obtaining the comprehensive score for each welcome message includes: normalizing the average response time, response rate, and transaction rate of each welcome message to obtain normalized average response time, normalized response rate, and normalized transaction rate; configuring corresponding weight coefficients for each of the normalized average response time, normalized response rate, and normalized transaction rate; wherein the sum of each weight coefficient is 1; and the comprehensive score for each welcome message is the sum of the normalized average response time, normalized response rate, and normalized transaction rate multiplied by their respective weight coefficients.

[0019] In some embodiments of the first aspect of this application, the step of performing language structure analysis processing on the high-scoring welcome message to obtain the expression pattern of the high-scoring welcome message includes: accepting input welcome message analysis prompts; calling an AI big model, wherein the AI ​​big model performs language structure analysis processing on the high-scoring welcome message according to the analysis requirements of the welcome message analysis prompts to obtain the expression pattern of the high-scoring welcome message.

[0020] In some embodiments of the first aspect of this application, the step of obtaining the average response time, response rate, and conversion rate of each type of initial contact script based on the historical initial contact script data includes: accepting input initial contact script analysis prompts; calling an AI big model, wherein the AI ​​big model classifies the historical initial contact script data according to the analysis requirements of the initial contact script analysis prompts to obtain the category of the initial contact script; and obtaining the average response time, response rate, and conversion rate of each type of initial contact script.

[0021] In some embodiments of the first aspect of this application, the average response time for each type of initial contact script is obtained by means of:

[0022]

[0023] Among them, T i Let T be the average response time for each type of initial contact script, where T is the set of non-empty response times, min(T) is the shortest response time, and max(T) is the longest response time.

[0024] The methods for obtaining the response rate for each type of initial contact script include:

[0025]

[0026] Among them, R i C represents the response rate for each type of initial contact script. i N_C is the total count of the initial contact scripts for each type. i The number of initial contact scripts for each type of response that is not empty;

[0027] The methods for obtaining the conversion rate for each type of initial contact script include:

[0028]

[0029] Among them, O i For the conversion rate of each type of first-contact script, O_C i The number of service transactions for each type of initial contact script is the number of service transactions where the transaction time is not empty.

[0030] In some embodiments of the first aspect of this application, obtaining the comprehensive score of each type of initial contact script includes: normalizing the average response time, response rate, and conversion rate of each type of initial contact script to obtain normalized average response time, normalized response rate, and normalized conversion rate; configuring corresponding weight coefficients for the normalized average response time, normalized response rate, and normalized conversion rate respectively; wherein the sum of each weight coefficient is 1; the comprehensive score of each type of initial contact script is the sum of the normalized average response time, normalized response rate, and normalized conversion rate multiplied by their respective weight coefficients.

[0031] To achieve the above and other related objectives, a second aspect of this application provides a computer-readable storage medium having a computer program stored thereon that, when executed by a processor, implements the method.

[0032] To achieve the above and other related objectives, a third aspect of this application provides a computer program product comprising computer program code that, when executed on a computer, causes the computer to implement the method.

[0033] To achieve the above and other related objectives, a fourth aspect of this application provides an electronic terminal, including a memory, a processor, and a computer program stored in the memory; the processor executes the computer program to implement the method.

[0034] The intelligent recommendation method for welcome messages and initial contact phrases provided in this application has the following beneficial effects:

[0035] This application enables efficient and accurate intelligent recommendation of welcome messages and initial contact scripts, improving the overall service level and effectively avoiding the problems of traditional script optimization relying on subjective experience and being difficult to quantify. Attached Figure Description

[0036] Figure 1 The flowchart shown is a method for intelligently recommending welcome messages and initial contact phrases according to an embodiment of this application.

[0037] Figure 2 The flowchart shown is a process for obtaining a comprehensive score for each welcome message in an intelligent recommendation method for welcome messages and initial contact phrases, as described in an embodiment of this application.

[0038] Figure 3 The diagram shows a schematic representation of the principle of obtaining the welcome expression pattern in an intelligent recommendation method for welcome messages and initial contact scripts according to an embodiment of this application.

[0039] Figure 4 The flowchart shown is a process for obtaining parameters of the initial contact script in an intelligent recommendation method for welcome messages and initial contact scripts according to an embodiment of this application.

[0040] Figure 5 The diagram shown illustrates the principle of obtaining the category of the initial contact script in the intelligent recommendation method for welcoming messages and initial contact scripts according to an embodiment of this application.

[0041] Figure 6 The flowchart shown is a process for obtaining a comprehensive score of the initial contact speech in an intelligent recommendation method for welcome messages and initial contact speech according to an embodiment of this application.

[0042] Figure 7 This diagram illustrates the overall implementation principle of an intelligent recommendation method for welcome messages and initial contact phrases, as shown in an embodiment of this application.

[0043] Figure 8 The diagram shown is a structural schematic of an electronic device according to an embodiment of this application.

[0044] Component designation explanation

[0045] 100 Electronic devices

[0046] 101 Memory

[0047] 102 processor

[0048] 103 Monitor

[0049] S100~S500 Steps

[0050] Steps S210~S230

[0051] S410~S430 Steps

[0052] Steps S441~S443 Detailed Implementation

[0053] The following specific examples illustrate the implementation of this application. Those skilled in the art can easily understand other advantages and effects of this application from the content disclosed in this specification. This application can also be implemented or applied through other different specific embodiments, and various details in this specification can also be modified or changed based on different viewpoints and applications without departing from the spirit of this application. It should be noted that, unless otherwise specified, the following embodiments and features in the embodiments can be combined with each other.

[0054] Before providing a further detailed description of the present invention, the nouns and terms used in the embodiments of the present invention are explained, and the nouns and terms used in the embodiments of the present invention are subject to the following interpretations:

[0055] <1> The welcome message is the first message a user sends to a customer after the customer joins the chat channel (such as WeChat), and it is usually a mass message script.

[0056] <2> The initial contact script is the first sentence a customer sends after responding with a welcome message; it is usually a one-on-one conversation script.

[0057] This embodiment provides an intelligent recommendation method, medium, and electronic device for welcome messages and initial contact scripts, applied to communication and chat scenarios. By analyzing the effectiveness of welcome messages and initial contact scripts from different channels, the system automatically identifies high-scoring language structures and expression patterns, guiding employees to send more attractive and convertible scripts, thereby improving customer response rates and service conversion rates, and achieving refined operations.

[0058] This embodiment provides an intelligent recommendation method, medium, and electronic device for welcome messages and initial contact scripts. First, it acquires and structures nearly a month's worth of chat data between company employees and customers, focusing on interactions within the first 24 hours after a customer's call. Then, for welcome messages from different channels, it statistically analyzes welcome message frequency, customer response rate, average response time, and conversion rate. This data is normalized and weighted to obtain a comprehensive score index, which is then ranked to identify high-performing welcome messages. Next, a large language model is used to analyze the structure of high-scoring scripts and extract expression patterns. These high-scoring expression patterns and scripts are provided for business use. Then, the large model's capabilities are used to classify the language content of initial contact scripts, identifying script types (such as download guidance, account opening guidance, etc.). The effectiveness indicators for each script type—response rate, average response time, and conversion rate—are analyzed, normalized, and weighted averages are calculated as the final score. Finally, the high-scoring contact script categories and related script content are output for business personnel to use as a targeted reference based on the channel, improving service efficiency and conversion rates. By combining the above-mentioned technical features, this embodiment effectively avoids the problems of traditional script optimization relying on subjective experience and being difficult to quantify, guiding business personnel to send more attractive and convertible scripts, thereby improving customer response rate and service conversion rate, and achieving refined operation.

[0059] The following will refer to the appendices in the embodiments of this application. Figure 1 To be continued Figure 8 This application provides a detailed description of the technical solutions in its embodiments. It aims to provide a smart recommendation method for welcome messages and initial contact scripts that can be understood and implemented by those skilled in the art without creative effort.

[0060] Figure 1 This is a flowchart illustrating the intelligent recommendation method for welcome messages and initial contact phrases in an embodiment of this application. For example... Figure 1 As shown, the intelligent recommendation method for welcome messages and initial contact scripts provided in this application embodiment includes the following steps S100 to S500.

[0061] Step S100: Obtain historical chat data, which includes historical welcome message data and historical first contact script data from different channels.

[0062] Step S200: In response to accessing the chat channel, obtain the average reply time, reply rate and conversion rate of each welcome message based on the historical welcome message data, and obtain a comprehensive score for each welcome message based on the average reply time, the reply rate and the conversion rate;

[0063] Step S300: Based on the comprehensive score of each welcome message, obtain the high-scoring welcome message, perform language structure analysis on the high-scoring welcome message to obtain the expression pattern of the high-scoring welcome message, and generate a welcome message corresponding to the current channel based on the expression pattern of the high-scoring welcome message;

[0064] Step S400: Upon receiving the user's reply chat data to the welcome message, the average reply time, reply rate, and conversion rate for each type of initial contact script are obtained based on the historical initial contact script data. A comprehensive score for each type of initial contact script is then obtained based on the average reply time, reply rate, and conversion rate. Step S500: Based on the comprehensive score of each type of initial contact script, the category and script content corresponding to high-scoring initial contact scripts are obtained. The category and script content corresponding to the high-scoring initial contact scripts are then recommended to the user, allowing the user to edit an initial contact script corresponding to the current channel based on the category and script content corresponding to the high-scoring initial contact scripts.

[0065] This embodiment's intelligent recommendation method for welcome messages and initial contact scripts assists users in accurately selecting or adjusting script content based on customers from different channels. From customer access, response, and transaction to script optimization, it constructs a complete closed loop from data collection, analysis, evaluation to execution, achieving efficient and accurate intelligent recommendation of welcome messages and initial contact scripts. This improves communication efficiency and customer experience, providing reliable support for intelligent services in the investment advisory industry and effectively avoiding the problems of traditional script optimization relying on subjective experience and being difficult to quantify.

[0066] The following provides a detailed description of steps S100 to S500 of the intelligent recommendation method for welcome messages and initial contact scripts in this embodiment.

[0067] Step S100: Obtain historical chat data, which includes historical welcome messages from different channels and historical initial contact scripts.

[0068] The historical chat data is preferably user-customer chat data from a recent period. For example, filtering historical chat data of customers who have called in the past month. That is, in this embodiment, the chat data between company employees and customers over the past month is acquired and structured, focusing on the interaction content within 24 hours after the customer calls in.

[0069] In this embodiment, the historical chat data is further processed into structured data. For example, for the contact situation in the 24 hours after the call is initiated, the historical welcome message data and historical first contact script data of different channels are organized into structured data. The structured data includes fields including but not limited to: employee name, customer name, call time, service transaction time (can be empty), channel, welcome message, customer's reply to the welcome message (empty if the customer does not reply), customer's reply time interval to the welcome message (empty if the customer does not reply, in seconds), first contact script, customer's reply to the first contact (empty if the customer does not reply), customer's reply time interval to the first contact (empty if the customer does not reply, in seconds), etc.

[0070] Step S200: In response to accessing the chat channel, obtain the average reply time, reply rate, and conversion rate of each welcome message based on the historical welcome message data, and obtain a comprehensive score for each welcome message based on the average reply time, the reply rate, and the conversion rate.

[0071] In this embodiment, after obtaining the welcome message data from different channels, the frequency of the welcome message, customer response rate, average response time, and conversion rate are statistically analyzed for each channel's welcome message, and finally, a comprehensive score is obtained for each welcome message.

[0072] In one specific implementation of this embodiment, the intelligent recommendation method for welcome messages and initial contact scripts includes obtaining the average response time for each welcome message through the following methods:

[0073]

[0074] Among them, T i Let T be the average response time for each welcome message, T be the set of non-empty response times, min(T) be the shortest response time, and max(T) be the longest response time.

[0075] In this embodiment, two new fields are added: welcome message count and average welcome message response time. The welcome message count is determined by clustering repeated welcome messages, with each welcome message counted as C. i The average response time for the welcome message is T. i The calculation method is to remove null values ​​for the customer's response time in the welcome message, and then take the average response time after removing the shortest and longest times (in minutes, 1 minute is taken if it is less than 1 minute, and rounded up if it exceeds 1 minute).

[0076] The clustering and counting of repeated welcome messages includes: converting the welcome message text into a vector representation using TF-IDF (Term Frequency-Inverse Document Frequency), Word2Vec, or BERT methods to facilitate similarity calculation. For example, converting the welcome message text into a sparse matrix, or averaging or pooling the word vectors of each sentence to obtain sentence-level vectors. Cosine similarity or Jaccard similarity is then used to calculate the similarity of the welcome message texts. Finally, a similarity threshold is set, and clustering algorithms (K-Means, DBSCAN, or hierarchical clustering) are used to cluster and count identical welcome messages.

[0077] For each welcome message, let its non-empty response time set be:

[0078] T = {t1, t2, ..., t} n (Length is N), shortest time: min(T), longest time: max(T), average response time calculation method:

[0079]

[0080] In this embodiment, a response rate field is added to calculate the response rate R for each welcome message. i The number of customer responses that are not empty is N_C. i The response rate for each welcome message is obtained through the following methods:

[0081]

[0082] Among them, R i C represents the response rate for each welcome message. i For the total count of each welcome message, N_C i This represents the number of welcome messages that are not empty.

[0083] In this embodiment, a new conversion rate field is added to calculate the conversion rate O for each welcome message. i The number of service transactions for each welcome message is the number of service transactions with non-empty times, assumed to be O_C. i The conversion rate for each welcome message is obtained through the following methods:

[0084]

[0085] Among them, O i For the conversion rate of each welcome message, O_C i The number of service transactions for each welcome message is the number of service transactions where the transaction time is not empty.

[0086] In this embodiment, a new welcome message comprehensive score field is added to calculate the comprehensive score S for each welcome message. i . Figure 2This is a flowchart illustrating the intelligent recommendation method for welcome messages and initial contact phrases, as shown in an embodiment of this application, for obtaining a comprehensive score for each welcome message. (See flowchart for example.) Figure 2 As shown, in one specific implementation of this embodiment, obtaining the comprehensive score of each welcome message includes the following steps S210 to S230.

[0087] Step S210: Normalize the average response time, response rate, and transaction rate for each welcome message to obtain normalized average response time, normalized response rate, and normalized transaction rate.

[0088] Step S220: Configure corresponding weight coefficients for the normalized average response time, the normalized response rate, and the normalized transaction rate; wherein the sum of all the weight coefficients is 1.

[0089] Step S230: The overall score for each welcome message is the sum of the normalized average response time, the normalized response rate, and the normalized conversion rate multiplied by their respective weighting coefficients.

[0090] In this embodiment, the average customer response time, response rate, and conversion rate for each welcome message are first normalized. The normalized average response time T norm The calculation method is as follows:

[0091]

[0092] Among them, T max T is the customer with the longest average response time for a welcome message. min T is the shortest average customer response time for a welcome message. i The average time for customers to respond to this welcome message.

[0093] In this embodiment, the normalized recovery rate R norm The calculation method is as follows:

[0094]

[0095] Among them, R max R is the highest response rate for a welcome message. min R represents the minimum response rate for a welcome message. i The response rate to this welcome message.

[0096] In this embodiment, the normalized transaction rate O norm The calculation method is as follows:

[0097]

[0098] Among them, O max The highest conversion rate for a welcome message is O.min The minimum conversion rate for a welcome message is O i This represents the conversion rate of the welcome message.

[0099] Each welcome message received an overall score of S. i Calculation method:

[0100] S i =W T *T norm +W R *R norm +W O *O norm

[0101] Among them: W T +W R +W O =1, W T W R W O It can be automatically adjusted according to actual business needs, for example, taking W. R =0.5,W O =0.4,W T =0.1, business focuses on response rate, so W can be used. R Increasing and focusing on service conversion rates can improve W O Adjusting W to prioritize user experience and response time T Increase. In this embodiment, by flexibly setting scoring weight parameters, such as focusing on response rate, conversion rate, or response experience, the scoring system can be automatically optimized according to actual business scenarios to meet the needs of word choice optimization under different operational strategies.

[0102] Step S300: Based on the overall score of each welcome message, obtain the high-scoring welcome message, perform language structure analysis on the high-scoring welcome message to obtain the expression pattern of the high-scoring welcome message, and generate a welcome message corresponding to the current channel based on the expression pattern of the high-scoring welcome message.

[0103] The language structure analysis of the high-scoring welcome message was performed to obtain its expression pattern, including:

[0104] 1) Perform syntactic analysis on high-scoring welcome messages: Use natural language processing tools (such as SpaCy, NLTK or BERT) to perform syntactic analysis on the welcome messages and extract the subject, predicate, object, modifiers and other components of the sentence.

[0105] 2) Pattern recognition of high-scoring welcome messages: Extract expression patterns by analyzing the commonalities of high-scoring welcome messages. For example: greetings, emotional expressions, and channel-related vocabulary.

[0106] 3) Pattern extraction from high-scoring welcome messages: Use machine learning algorithms (such as Hidden Markov Models and Conditional Random Fields) to automatically extract expression patterns from high-scoring welcome messages.

[0107] 4) Generate a pattern representation template based on the extracted pattern.

[0108] Then, based on the current channel and target audience, fill in the variables in the edit mode expression template to generate a welcome message corresponding to the channel.

[0109] Figure 3 This diagram illustrates the principle of obtaining the welcome expression pattern in an intelligent recommendation method for welcome messages and initial contact phrases, as shown in an embodiment of this application. Figure 3 As shown, in one specific implementation of this embodiment, the step of performing language structure analysis on the high-scoring welcome message to obtain its expression pattern includes: receiving input welcome message analysis prompts; calling an AI big model, wherein the AI ​​big model performs language structure analysis on the high-scoring welcome message according to the analysis requirements of the welcome message analysis prompts to obtain its expression pattern.

[0110] In this embodiment, based on the comprehensive score of each welcome message, they are ranked from highest to lowest to obtain the welcome messages with the highest comprehensive scores. Language structure analysis is then performed on one or more of the top-ranked welcome messages. The number of welcome messages is determined based on the actual number; for example, language structure analysis is performed on the top 10 welcome messages, and prompts are written such as: "You are a professional key point extraction expert, capable of extracting key points from the message, requiring conciseness." Specific prompts can be adjusted as needed, utilizing large AI models, including but not limited to deepseek-R1, Qwen-max, and Qwen2.5-32B models, to obtain the expression patterns of the welcome messages. For example, the expression pattern of a high-scoring welcome message is as follows:

[0111] (1) Use polite language, (2) Introduce the company, a long-established investment institution, (3) Introduce the employees (employee number, professional license number), (4) Introduce the services (materials, etc.), (5) Explain that you are not a robot.

[0112] When sending welcome messages through different channels, users can refer to the above-mentioned high-scoring welcome message expression models and content, modify them, and send them to customers.

[0113] Step S400: Upon receiving the user's reply chat data to the welcome message, the average reply time, reply rate, and conversion rate for each type of first-contact script are obtained based on the historical first-contact script data. A comprehensive score for each type of first-contact script is obtained based on the average reply time, reply rate, and conversion rate for each type of first-contact script.

[0114] Figure 4 This is a flowchart illustrating the process of obtaining parameters for the initial contact script in an intelligent recommendation method for welcome messages and initial contact scripts, as shown in an embodiment of this application. Figure 4 As shown, in one specific implementation of this embodiment, the step of obtaining the average response time, response rate and conversion rate of each type of first contact script based on the historical first contact script data includes the following steps S410 to S430.

[0115] Step S410: Accept the input initial contact dialogue analysis prompts;

[0116] Step S420: Call the AI ​​big model. The AI ​​big model classifies the historical first contact dialogue data according to the analysis requirements of the first contact dialogue analysis prompts to obtain the category of the first contact dialogue.

[0117] Step S430: Obtain the average response time, response rate, and conversion rate for each type of initial contact script.

[0118] This embodiment utilizes large language models (such as Qwen, DeepSeek, etc.) to extract language structure and classify dialogue, which can extract core expressive elements and potential categories from unstructured employee language, thereby promoting the intelligent classification and continuous optimization of communication dialogue.

[0119] Figure 5 This diagram illustrates the principle of obtaining the category of the initial contact script in an intelligent recommendation method for welcoming greetings and initial contact scripts, as shown in an embodiment of this application. Figure 5 As shown, the user-inputted initial contact dialogue analysis prompts are fed into the AI ​​model. The AI ​​model then categorizes the historical initial contact dialogue data according to the analysis requirements of the initial contact dialogue analysis prompts, and outputs the category of the initial contact dialogue. In this embodiment, the initial contact dialogue category field can be a single category or a combination of multiple categories.

[0120] In this embodiment, a new field for the category of initial contact dialogue is added. The initial contact dialogue is processed, and its language structure is analyzed. Prompt words are written, and large AI models, including but not limited to qwen-plus and qwen-turbo, are used to classify the dialogue. Examples of input prompt words for the initial contact dialogue analysis are as follows:

[0121] You are a professional classification expert, classifying text. Current categories: greetings, software download instructions.

[0122] We may need to explore more categories, requiring extraction from the client's actual problems, and keeping them brief with no unnecessary explanations.

[0123] The specific prompts for the initial contact script can be adjusted based on the actual content of the script.

[0124] The AI ​​model categorizes historical initial contact dialogue data based on the analysis requirements of the initial contact dialogue prompts, obtaining the categories of the initial contact dialogues. For example, it may obtain all categories such as: guidance on downloading and using software (indicator mini-program), free stocks, stock analysis guidance, cooperation guidance, inquiries about opening an account, and information-related. Initial contact dialogues may be a single category or a combination of multiple categories, such as: information-related / guidance on downloading and using software, indicator mini-program.

[0125] In one specific implementation of this embodiment, the method for obtaining the average response time for each type of initial contact script includes:

[0126]

[0127] Among them, T i Let T be the average response time for each type of initial contact script, T be the set of non-empty response times, min(T) be the shortest response time, and max(T) be the longest response time.

[0128] In this embodiment, clustering is performed based on the type of initial contact script. New fields for category count and average response time for each contact script are added, with a count of C for each category. i The average response time for the contact script is T. i The calculation method involves removing null values ​​for the initial customer response time from the initial contact script, and then taking the average response time after removing the shortest and longest response times. For example, in minutes, data within 1 minute is rounded up, and data exceeding 1 minute is rounded to the nearest minute.

[0129] For each type of initial contact script, let its non-empty response time set be:

[0130] T = {t1, t2, ..., t} n (Length is N), shortest time: min(T), longest time: max(T), calculation method for average response time of the first contact:

[0131]

[0132] In this embodiment, a response rate field is added to calculate the response rate R for each type of contact script. i The number of customers whose responses are not empty during the first contact is N_C. i The methods for obtaining the response rate for each type of initial contact script include:

[0133]

[0134] Among them, Ri C represents the response rate for each type of initial contact script. i N_C is the total count of the initial contact scripts for each type. i The number of initial contact scripts for each type of response that is not empty;

[0135] In this embodiment, a new conversion rate field is added to calculate the conversion rate O for each type of initial contact script. i The number of successful transactions for each type of initial contact script is the number of transactions where the transaction time is not empty, assuming it is O_C. i The methods for obtaining the conversion rate for each type of initial contact script include:

[0136]

[0137] Among them, O i For the conversion rate of each type of first-contact script, O_C i The number of service transactions for each type of initial contact script is the number of service transactions where the transaction time is not empty.

[0138] In this embodiment, a new comprehensive score field for the initial contact script is added to calculate the comprehensive score S for each type of contact script. i , Figure 6 This is a flowchart illustrating the process of obtaining a comprehensive score for the initial contact script in an intelligent recommendation method for welcoming greetings and initial contact scripts, as described in an embodiment of this application. Figure 6 As shown, in one specific implementation of this embodiment, obtaining the comprehensive score of each type of initial contact script includes the following steps S441 to S443.

[0139] Step S441: Normalize the average response time, response rate, and conversion rate for each type of initial contact script to obtain normalized average response time, normalized response rate, and normalized conversion rate.

[0140] Step S442: Configure corresponding weight coefficients for the normalized average response time, the normalized response rate, and the normalized transaction rate; wherein the sum of all the weight coefficients is 1.

[0141] Step S443: The comprehensive score for each type of initial contact script is the sum of the normalized average response time, the normalized response rate, and the normalized conversion rate multiplied by their respective weighting coefficients.

[0142] In this embodiment, the average customer response time, response rate, and conversion rate for each type of initial contact script are normalized. The normalized average response time T for each type of initial contact script is... norm The calculation method is as follows:

[0143]

[0144] Among them, T max For initial contact, the average response time to customers is the longest. (T) min T is the shortest average response time for initial customer contact. i The average response time for customers using this type of initial contact script.

[0145] In this embodiment, the normalized response rate R of the initial contact script is... norm The calculation method is as follows:

[0146]

[0147] Among them, R max To achieve the highest response rate for initial contact scripts, R min R is the minimum response rate for the initial contact script. i This refers to the response rate for this type of initial contact script.

[0148] In this embodiment, the normalized conversion rate of the first-contact script is O. norm The calculation method is as follows:

[0149]

[0150] Among them, O max To achieve the highest conversion rate for the first contact, O min To achieve the minimum closing rate for the initial contact script, O i This represents the conversion rate for this type of initial contact script.

[0151] In this embodiment, the comprehensive score S for each type of initial contact script is... i The calculation method is as follows:

[0152] S i =W T *T norm +W R *R norm +W O *O norm

[0153] W T +W R +W O =1

[0154] Among them, W T W R W O This solution can be automatically adjusted according to actual business needs; W is the optimal value. R =0.5,W O =0.4,W T =0.1, business focuses on response rate, so W can be used. RIncreasing and focusing on service conversion rates can improve W O Adjusting W to prioritize user experience and response time T Increase. In this embodiment, by flexibly setting scoring weight parameters, such as focusing on response rate, conversion rate, or response experience, the scoring system can be automatically optimized according to actual business scenarios to meet the needs of word choice optimization under different operational strategies.

[0155] Step S500: Based on the comprehensive score of each type of first contact script, obtain the category and script content corresponding to the high-scoring first contact script, and recommend the category and script content corresponding to the high-scoring first contact script to the user so that the user can edit the first contact script corresponding to the current channel based on the category and script content corresponding to the high-scoring first contact script.

[0156] Based on the overall score of the initial contact scripts, they are ranked from highest to lowest to determine the category of initial contact scripts with the highest overall scores. These high-scoring initial contact script categories and related content are provided for reference by business system users. Users, depending on the channel, can refer to the high-scoring initial contact script categories and content after receiving a welcome message from the customer, edit or modify the content, and then send it to the customer.

[0157] Figure 7 This diagram illustrates the overall implementation principle of an intelligent recommendation method for welcome messages and initial contact phrases, as shown in an embodiment of this application. Figure 7 As shown, the implementation process of the intelligent recommendation method for welcome messages and initial contact scripts in this embodiment is as follows:

[0158] Acquire historical chat data: Acquire and structure chat data between company employees and customers over the past month, focusing on interactions within 24 hours of a customer's call.

[0159] The historical welcome message data and historical first contact script data from different channels will be organized into structured data, including but not limited to: employee name, customer name, call time, service completion time, channel, welcome message, customer's response to the welcome message, time interval between customer's response to the welcome message, first contact script, customer's response to the first contact, and time interval between customer's response to the first contact, etc.

[0160] For welcome messages from different channels, statistical analysis is performed on the welcome message data, including the frequency of welcome messages, customer response rate, average response time, and conversion rate. Then, the average response time, response rate, and conversion rate of each welcome message are normalized, and a comprehensive score is calculated and ranked. In other words, the above data is normalized, weighted, and averaged to obtain a comprehensive score index and ranking, thus identifying high-performing welcome messages.

[0161] Then, the large language model is used to extract the key points of the high-scoring phrases and output the expression patterns and content of the high-scoring welcome messages for users to refer to and edit.

[0162] After sending a welcome message to the customer, if a reply is received, continue to recommend the initial contact script to the user.

[0163] Statistical analysis is performed on the initial contact script data. Analysis prompts for the initial contact scripts are input into an AI model, which categorizes the language content, identifies script types, and generates category fields, such as guidance on downloading or account opening. Then, for each script type, performance metrics are calculated: frequency, average customer response time, response rate, and conversion rate. The average response time, response rate, and conversion rate of the initial contact scripts are then normalized. A comprehensive score for the initial contact scripts is calculated and ranked, and high-scoring script categories and related script content are output for business personnel to use for targeted reference based on channel, improving service efficiency and conversion rates.

[0164] As can be seen from the above, this embodiment performs structured processing and quantitative analysis of customer contact data within 24 hours of customer inquiries, introducing key indicators such as response rate, conversion rate, and average response time. It establishes a comprehensive evaluation system for welcome messages and initial contact scripts through normalization and weighted scoring, effectively avoiding the problems of traditional script optimization relying on subjective experience and being difficult to quantify. By analyzing the effectiveness of welcome messages and contact scripts from different channels, it automatically identifies high-scoring language structures and expression patterns, guiding users to send more attractive and convertible scripts, thereby improving customer response rates and service conversion rates, achieving refined operations. This embodiment utilizes large language models (such as Qwen and DeepSeek) for language structure extraction and script classification, enabling the extraction of core expressive elements and potential categories from unstructured employee language, thereby promoting intelligent classification and continuous optimization of contact scripts.

[0165] This embodiment regularly analyzes and updates incoming call data, dynamically adjusting evaluation criteria based on performance across different channels. This enables continuous updates and intelligent recommendations of welcome messages and initial contact scripts, improving overall service levels. By flexibly setting scoring weight parameters, such as focusing on response rate, conversion rate, or response experience, the scoring system can be automatically optimized based on actual business scenarios, meeting the script optimization needs under different operational strategies. The business system automatically prompts high-scoring script expression models and contact categories, assisting employees in accurately selecting or adjusting script content based on customers from different channels, improving communication efficiency and customer experience. From customer access, response, and conversion to script optimization, a complete closed loop from data collection, analysis, evaluation to execution is constructed, providing reliable support for intelligent services in the investment advisory industry.

[0166] In summary, the intelligent recommendation method for welcome messages and initial contact scripts described in this application supports automatic optimization of the scoring system based on actual business scenarios, meeting the script optimization needs under different operational strategies. The system automatically prompts high-scoring script expression models and contact categories, assisting employees in accurately selecting or adjusting script content based on customers from different channels, thereby improving communication efficiency and customer experience. From customer access, response, and transaction to script optimization, a complete closed loop from data collection, analysis, evaluation to execution is constructed, providing reliable support for intelligent services in the investment advisory industry.

[0167] The scope of protection for the intelligent recommendation method for welcome messages and initial contact scripts described in this application is not limited to the execution order of the steps listed in this embodiment. Any solution implemented by adding, subtracting, or replacing steps in the prior art based on the principles of this application is included within the scope of protection of this application.

[0168] According to the method provided in the embodiments of this application, the embodiments of this application also provide a computer program product, which includes: computer program code, which, when the computer program code is run on a computer, causes the computer to execute the intelligent recommendation method for welcome messages and initial contact scripts provided in any embodiment of this application.

[0169] This application also provides a computer-readable storage medium storing a computer program thereon, which, when executed by a processor, implements the intelligent recommendation method for welcome messages and initial contact scripts provided in any embodiment of this application.

[0170] In the embodiments of this application, any combination of one or more storage media can be used. The storage medium can be a computer-readable signal medium or a computer-readable storage medium. A computer-readable storage medium can be, for example—but not limited to—an electrical, magnetic, optical, electromagnetic, infrared, or semiconductor system, apparatus, or device, or any combination thereof. More specific examples of computer-readable storage media (a non-exhaustive list) include: an electrical connection having one or more wires, a portable computer disk, a hard disk, RAM, ROM, erasable programmable read-only memory (EPROM or flash memory), optical fiber, portable compact disk read-only memory (CD-ROM), optical storage device, magnetic storage device, or any suitable combination thereof. In this document, a computer-readable storage medium can be any tangible medium that contains or stores a program that can be used by or in conjunction with an instruction execution system, apparatus, or device.

[0171] This application also provides an electronic device. Figure 8The diagram shown is a structural schematic of the electronic device 100 provided in an embodiment of this application. In some embodiments, the electronic device may be a mobile phone, tablet computer, wearable device, in-vehicle device, augmented reality (AR) / virtual reality (VR) device, laptop computer, ultra-mobile personal computer (UMPC), netbook, personal digital assistant (PDA), or other terminal device. Furthermore, the intelligent recommendation method for welcome messages and initial contact scripts provided in this application can also be applied to databases, servers, and service response systems based on terminal artificial intelligence. This application embodiment does not limit the specific application scenarios of the intelligent recommendation method for welcome messages and initial contact scripts.

[0172] like Figure 8 As shown, the electronic device 100 provided in this application embodiment includes a memory 101 and a processor 102.

[0173] The memory 101 is used to store computer programs; preferably, the memory 101 includes various media that can store program code, such as ROM, RAM, magnetic disk, USB flash drive, memory card or optical disk.

[0174] Specifically, memory 101 may include computer system readable media in the form of volatile memory, such as random access memory (RAM) and / or cache memory. Electronic device 100 may further include other removable / non-removable, volatile / non-volatile computer system storage media. Memory 101 may include at least one program product having a set (e.g., at least one) of program modules configured to perform the functions of the embodiments of this application.

[0175] The processor 102 is connected to the memory 101 and is used to execute the computer program stored in the memory 101 so that the electronic device 100 executes the intelligent recommendation method for welcome messages and initial contact scripts provided in any embodiment of this application.

[0176] Optionally, the processor 102 may be a general-purpose processor, including a central processing unit (CPU), a network processor (NP), etc.; it may also be a digital signal processor (DSP), an application-specific integrated circuit (ASIC), a field-programmable gate array (FPGA), or other programmable logic devices, discrete gate or transistor logic devices, or discrete hardware components.

[0177] Optionally, in this embodiment, the electronic device 100 may further include a display 103. The display 103 is communicatively connected to the memory 101 and the processor 102, and is used to display the relevant GUI interactive interface of the intelligent recommendation method for the welcome message and the initial contact script.

[0178] In summary, this application achieves efficient and accurate intelligent recommendation of welcome messages and initial contact scripts, improving overall service levels and effectively avoiding the problems of traditional script optimization relying on subjective experience and being difficult to quantify. Therefore, this application effectively overcomes the various shortcomings of existing technologies and has high industrial application value.

[0179] The above embodiments are merely illustrative of the principles and effects of this application and are not intended to limit this application. Any person skilled in the art can modify or alter the above embodiments without departing from the spirit and scope of this application. Therefore, all equivalent modifications or alterations made by those skilled in the art without departing from the spirit and technical concept disclosed in this application should still be covered by the claims of this application.

Claims

1. A method for intelligent recommendation of welcome speech and first contact script, characterized in that, include: Acquire historical chat data, which includes historical welcome messages from different channels and historical initial contact scripts. In response to accessing the chat channel, the average response time, response rate, and conversion rate of each welcome message are obtained based on the historical welcome message data, and a comprehensive score for each welcome message is obtained based on the average response time, the response rate, and the conversion rate. Based on the overall score of each welcome message, a high-scoring welcome message is obtained. The language structure of the high-scoring welcome message is analyzed to obtain its expression pattern. Based on the expression pattern of the high-scoring welcome message, a welcome message corresponding to the current channel is generated. Upon receiving user responses to the welcome message, the system obtains the average response time, response rate, and conversion rate for each type of initial contact script based on the historical initial contact script data. Furthermore, it obtains a comprehensive score for each type of initial contact script based on the average response time, response rate, and conversion rate. Based on the overall score of each type of first-contact script, the category and script content corresponding to the high-scoring first-contact script are obtained, and the category and script content corresponding to the high-scoring first-contact script are recommended to the user so that the user can edit the first-contact script corresponding to the current channel based on the category and script content corresponding to the high-scoring first-contact script. The process of performing linguistic structure analysis on the high-scoring welcome message to obtain its expression pattern includes: Analyze the welcome message prompts received from the input; The AI ​​model is invoked, and it performs language structure analysis on the high-scoring welcome message based on the analysis requirements of the welcome message analysis prompts to obtain the expression pattern of the high-scoring welcome message. 2.The method of claim 1, wherein the welcome message and first contact script recommendation method is characterized by, The average response time for each welcome message was obtained through the following methods: ； in, The average response time for each welcome message. For a non-empty set of response times, To achieve the shortest response time, This is the longest possible response time. The response rate for each welcome message is obtained through the following methods: ； in, The response rate for each welcome message. For the total count of each welcome message, The number of welcome messages that are not empty in the reply; The conversion rate for each welcome message is obtained through the following methods: ； wherein, is the conversion rate of each welcome message, is the service conversion number of each welcome message, which is the number of service conversion time that is not empty.

3. The intelligent recommendation method for welcome messages and initial contact scripts according to claim 2, characterized in that, The overall score for each welcome message includes: For each welcome message, the average response time, the response rate, and the conversion rate are normalized to obtain the normalized average response time, normalized response rate, and normalized conversion rate. Each of the normalized average response time, the normalized response rate, and the normalized transaction rate is assigned a corresponding weight coefficient; wherein the sum of all the weight coefficients is 1. The overall score for each welcome message is the sum of the normalized average response time, the normalized response rate, and the normalized conversion rate multiplied by their respective weighting coefficients.

4. The intelligent recommendation method for welcome messages and initial contact scripts according to claim 1, characterized in that, The process of obtaining the average response time, response rate, and conversion rate for each type of initial contact script based on the historical initial contact script data includes: Analyze prompts for the initial contact dialogue received from the user; The AI ​​model is invoked, and the AI ​​model classifies the historical first-contact dialogue data according to the analysis requirements of the first-contact dialogue analysis prompts to obtain the category of the first-contact dialogue. Obtain the average response time, response rate, and conversion rate for each type of initial contact script.

5. The intelligent recommendation method for welcome messages and initial contact scripts according to claim 1, characterized in that, The methods for obtaining the average response time for each type of initial contact script include: ； in, The average response time for each type of initial contact script. For a non-empty set of response times, To achieve the shortest response time, This is the longest possible response time. The methods for obtaining the response rate for each type of initial contact script include: ； in, The response rate for each type of initial contact script. The total count of the initial contact scripts for each category. The number of initial contact scripts for each type of response that is not empty; The methods for obtaining the conversion rate for each type of initial contact script include: ； in, The conversion rate for each type of initial contact script. The number of service transactions for each type of initial contact script is the number of service transactions where the transaction time is not empty. 6.The method of claim 5, wherein the welcome message and first contact script recommendation method is characterized by, The overall score for each type of initial contact script includes: For each type of initial contact script, the average response time, the response rate, and the conversion rate are normalized to obtain normalized average response time, normalized response rate, and normalized conversion rate. Each of the normalized average response time, the normalized response rate, and the normalized transaction rate is assigned a corresponding weight coefficient; wherein the sum of all the weight coefficients is 1. The overall score for each type of initial contact script is the sum of the normalized average response time, the normalized response rate, and the normalized conversion rate multiplied by their respective weighting coefficients.

7. A computer-readable storage medium having stored thereon a computer program, characterized in that When executed by a processor, the computer program implements the intelligent recommendation method for the welcome message and initial contact script as described in any one of claims 1 to 6.

8. A computer program product, characterised in that, The computer program product includes computer program code, which, when run on a computer, enables the computer to implement the intelligent recommendation method for welcome messages and initial contact scripts as described in any one of claims 1 to 6.

9. An electronic device, comprising: The electronic device includes: Processor and memory; The memory stores program instructions; The processor is configured to run the program instructions to execute the intelligent recommendation method for welcome messages and initial contact scripts as described in any one of claims 1 to 6.

Images