Recommended verbal skill generation method and device for non-vehicle product, equipment and storage medium

Abstract

The invention belongs to the technical field of artificial intelligence, and discloses a non-vehicle product recommendation verbal skill generation method and device, equipment and a storage medium, and the method comprises the steps: obtaining a customer information set and a seat information set; judging a customer evaluation grade of the customer according to the customer information set through a customer grade model; judging the seat evaluation level of the seat personnel according to the seat information set through a seat level model; obtaining a target recommended product set in a predefined non-vehicle product gear table according to the customer evaluation level and the seat evaluation level; and analyzing the customer information set to obtain a customer information analysis result, and generating a target product recommendation verbal skill according to the customer information analysis result, the seat evaluation level and the target recommended product set. The method can be applied to a financial scene, and the accuracy of non-vehicle product recommendation is improved by generating the target product recommendation verbal skill in the financial scene for the seat personnel to use.

Description

Technical Field

[0001] The present application relates to the field of artificial intelligence, and in particular to a method, device, equipment and storage medium for generating recommendation scripts for non-automotive products. Background Art

[0002] In the insurance business, the initial contact of auto insurance agents is a key step in the insurance sales process. Its efficiency and effectiveness directly affect the sales success rate and customer satisfaction of insurance products. Specifically, the initial contact of auto insurance agents refers to the process in which agents proactively contact potential customers for the first time through outbound calls and other means, introduce auto insurance products and related services to customers, and promote their purchase of insurance.

[0003] In the traditional first-call auto insurance operation by agents, the agent's decision on which non-auto insurance product to recommend depends on the agent's subjective experience when communicating with the customer. It can be seen that in the traditional first-call auto insurance operation by agents, the agent lacks a comprehensive assessment of the customer's comprehensive capabilities, such as the customer's purchasing power, feedback attitude during the service cycle, customer personality characteristics and other key factors. As a result, in the traditional first-call auto insurance operation by agents, the non-auto product recommendations are less targeted and matched, which in turn leads to lower accuracy in the non-auto product recommendations. Summary of the Invention

[0004] The purpose of the embodiments of the present application is to propose a method, device, equipment and storage medium for generating recommendation scripts for non-automotive products, the main purpose of which is to improve the accuracy of non-automotive product recommendations.

[0005] First, to solve the above technical problems, the present application provides a method for generating recommendation scripts for non-automotive products, which adopts the following technical solutions:

[0006] Get the customer's customer information set and the agent's seat information set;

[0007] Scoring the customer information set in multiple dimensions using a customer rating model to obtain multiple customer dimension scores, and determining the customer evaluation level of the customer based on predefined customer dimension weights and the multiple customer dimension scores;

[0008] Scoring the seat information set in multiple dimensions using a seat grading model to obtain multiple seat dimension scores, and determining the seat evaluation grade of the agent based on predefined seat dimension weights and the multiple seat dimension scores;

[0009] Obtaining a target recommended product set from a predefined non-vehicle product tier table based on the customer evaluation level and the seat evaluation level;

[0010] The customer information set is analyzed to obtain a customer information analysis result, and a target product recommendation script is generated based on the customer information analysis result, the agent evaluation level, and the target recommended product set.

[0011] Secondly, in order to solve the above technical problems, the embodiments of the present application further provide a device for generating recommendation scripts for non-automotive products, which adopts the following technical solutions:

[0012] The information acquisition module is used to obtain the customer information set of the customer and the seat information set of the agent;

[0013] A first grade judgment module is configured to score the customer information set in multiple dimensions using a customer grade model to obtain multiple customer dimension scores, and to determine the customer evaluation grade of the customer based on predefined customer dimension weights and the multiple customer dimension scores;

[0014] A second grade judgment module is configured to score the agent information set in multiple dimensions using an agent grade model to obtain multiple agent dimension scores, and determine the agent evaluation grade of the agent based on predefined agent dimension weights and the multiple agent dimension scores;

[0015] A product acquisition module, configured to acquire a target recommended product set from a predefined non-vehicle product level table based on the customer evaluation level and the seat evaluation level;

[0016] The speech generation module is used to analyze the customer information set to obtain the customer information analysis results, and generate target product recommendation speech based on the customer information analysis results, the seat evaluation level and the target recommended product set.

[0017] On the third aspect, in order to solve the above-mentioned technical problems, an embodiment of the present application also provides a computer device, comprising: at least one processor; and a memory communicatively connected to the at least one processor; wherein the memory stores instructions that can be executed by the at least one processor, and the instructions are executed by the at least one processor so that the at least one processor can execute the method for generating recommendation scripts for non-automotive products as described above.

[0018] Fourthly, in order to solve the above-mentioned technical problems, an embodiment of the present application further provides a computer-readable storage medium storing a computer program, which, when executed by a processor, implements the method for generating recommendation scripts for non-automotive products as described above.

[0019] Compared with the prior art, the embodiments of the present application have the following beneficial effects:

[0020] By obtaining the customer's customer information set and the agent's seat information set, we can provide a basis for determining the customer evaluation level and the agent evaluation level in the future. We can then match the target product recommendation range and sales pitch based on the level, thereby improving the accuracy of non-automotive product recommendations.

[0021] By performing a multi-dimensional scoring calculation based on the first intention type, the second intention category, the customer personality, the customer attitude, and the structured features, multiple customer dimension scores are obtained, which can provide a basis for the subsequent risk level model to determine the customer evaluation level; and the comprehensive calculation of the multi-dimensional scores can improve the accuracy of the customer evaluation level;

[0022] By obtaining the average value of each dimension in the organization and determining the indicator status of each dimension data based on the average value, the benchmark differences between different regions can be eliminated, thereby improving the accuracy of the indicator status determination; by performing a weighted summation of all the seat dimension scores, a total seat dimension score is obtained; and the seat evaluation level of the seat personnel is determined based on the total seat dimension score, thereby preventing accidental interference from a single data and improving the accuracy of the seat evaluation level;

[0023] By determining a target product set based on customer and agent ratings, we can quickly improve the relevance and matching of recommended products, thereby significantly increasing customer underwriting rates, optimizing operational processes, and enhancing customer satisfaction and loyalty.

[0024] By analyzing the customer information set, the customer information analysis results are obtained, and according to the customer information analysis results, the agent evaluation level and the target recommended product set, the target product recommendation script is generated. The script generated based on the customer, the agent and the target recommended product improves the accuracy of non-auto product recommendations and increases the customer's underwriting rate. BRIEF DESCRIPTION OF THE DRAWINGS

[0025] In order to more clearly illustrate the solutions in this application, a brief introduction will be given below to the drawings required for use in the description of the embodiments of this application. Obviously, the drawings described below are some embodiments of this application. For ordinary technicians in this field, other drawings can be obtained based on these drawings without any creative work.

[0026] Figure 1 is an exemplary system architecture diagram to which the present application may be applied;

[0027] Figure 2 A flowchart of an embodiment of a method for generating recommendation words for non-car products according to the present application;

[0028] Figure 3This is a structural diagram of an embodiment of a device for generating recommendation phrases for non-automotive products according to the present application;

[0029] Figure 4 It is a structural diagram of an embodiment of a device according to the present application. DETAILED DESCRIPTION

[0030] Unless otherwise defined, all technical and scientific terms used herein have the same meanings as commonly understood by those skilled in the art to which this application belongs. The terms used in the specification of the application are for the purpose of describing specific embodiments only and are not intended to limit this application. The terms "including" and "having" and any variations thereof in the specification and claims of this application and the above-mentioned drawings are intended to cover non-exclusive inclusions. The terms "first", "second", etc. in the specification and claims of this application or the above-mentioned drawings are used to distinguish different objects, not to describe a specific order.

[0031] References herein to "embodiments" mean that a particular feature, structure, or characteristic described in connection with the embodiments may be included in at least one embodiment of the present application. The appearance of this phrase in various places in the specification does not necessarily refer to the same embodiment, nor does it constitute an independent or alternative embodiment that is mutually exclusive of other embodiments. It is understood, both explicitly and implicitly, by those skilled in the art that the embodiments described herein may be combined with other embodiments.

[0032] In order to enable those skilled in the art to better understand the solution of the present application, the technical solution in the embodiments of the present application will be clearly and completely described below in conjunction with the accompanying drawings.

[0033] like Figure 1 As shown, system architecture 100 may include a terminal device 101, a network 102, and a server 103. Terminal device 101 may be a laptop computer 1011, a tablet computer 1012, or a mobile phone 1013. Network 102 is a medium for providing a communication link between terminal device 101 and server 103. Network 102 may include various connection types, such as wired or wireless communication links or fiber optic cables.

[0034] The user can use the terminal device 101 to interact with the server 103 via the network 102 to receive or send messages, etc. Various communication client applications can be installed on the terminal device 101, such as web browser applications, shopping applications, search applications, instant messaging tools, email clients, social platform software, etc.

[0035] The terminal device 101 can be various electronic devices with a display screen and supporting web browsing. In addition to the laptop computer 1011, tablet computer 1012 or mobile phone 1013, the terminal device 101 can also be an e-book reader, an MP3 player (Moving Picture Experts Group Audio Layer III), an MP4 (Moving Picture Experts Group Audio Layer IV) player, a laptop computer and a desktop computer, etc.

[0036] The server 103 may be a server that provides various services, such as a background server that provides support for web pages displayed on the terminal device 101 .

[0037] It should be noted that the method for generating recommendation words for non-automotive products provided in the embodiments of the present application is generally executed by a server / terminal device. Accordingly, the device for generating recommendation words for non-automotive products is generally set in the server / terminal device.

[0038] It should be understood that Figure 1 The number of terminal devices, networks and servers in the embodiment is merely illustrative. Any number of terminal devices, networks and servers may be provided as required.

[0039] Continue to refer Figure 2 , shows a flowchart of an embodiment of the method for generating recommendation words for non-automotive products according to the present application. According to different needs, the order of the steps in the flowchart can be changed, and some steps can be omitted. The method for generating recommendation words for non-automotive products provided in the embodiment of the present application can be applied to any scenario where product recommendation words need to be generated, and the method for generating recommendation words for non-automotive products can be applied to products in these scenarios. The method for generating recommendation words for non-automotive products includes the following steps:

[0040] Step S201: Acquire a customer information set of a customer and an agent information set of an agent.

[0041] In this embodiment, the method for generating recommendation words for non-car products is executed on the electronic device (eg Figure 1The server / terminal device shown in the figure) can obtain the customer information set and the agent information set through a wired connection or a wireless connection. It should be noted that the above-mentioned wireless connection method may include but is not limited to 3G / 4G / 5G connection, WiFi connection, Bluetooth connection, WiMAX connection, Zigbee connection, UWB (Ultra Wi-Fi Broadband) connection, and other wireless connection methods currently known or to be developed in the future.

[0042] In this embodiment, the customer information set of the above-mentioned customer includes structured data (customer information tags) and text data (call text information and platform text information) during the service period (such as 60 days before the customer's expiration date), and the customer information tags include: affiliated institution, vehicle invoice price, non-auto premium in the previous year, whether the customer has a company, etc.; the agent information set of the above-mentioned agent includes: 1. Expiration renewal rate: statistics on the expiration renewal rate of customers under the name of the agent in the most recent preset months (such as the most recent March); 2. Planning achievement progress: statistics on the progress of the non-auto premium planning achievement under the name of the agent in the most recent preset months; 3. Non-auto co-sales rate: statistics on the co-sales rate of auto insurance products and non-auto products under the name of the agent in the most recent preset months; 4. Non-auto premium share: statistics on the proportion of non-auto premium income under the name of the agent in the total premium income in the most recent preset months.

[0043] In this embodiment, the customer's initial customer information set and the agent's seat information set are obtained, and the initial customer information set is preprocessed to obtain the customer information set, wherein the preprocessing method includes text cleaning, data deduplication, etc., and the basic information tags in the initial customer information set are verified to check whether the data has missing values, outliers and logical errors, such as checking whether the vehicle invoice price is negative, whether the company logo under the customer's name is correct, etc.; for missing values, they can be filled according to the mean filling method, etc.; for outliers, they are deleted.

[0044] In one embodiment, the step of obtaining the customer information set of the customer and the agent information set of the agent includes:

[0045] Obtain the customer information tag of the customer and the call recordings, platform voice and platform text in the server;

[0046] Converting the call recording and the platform voice into text format using speech recognition technology to obtain a call text and a platform converted text, and collecting the platform text, the call text, the platform converted text, and the customer information tag to obtain the customer information set;

[0047] The business data of the agent is obtained from a preset database, and the business data is calculated to obtain the agent information set.

[0048] In this embodiment, customer information tags are extracted from a customer relationship management system (CRM) using tools such as an API interface to determine a target customer group. Based on the target customer group, an SQL query statement is used to retrieve the customer information tags of each target customer in the target customer group from the CRM database, including the affiliation, vehicle invoice price, previous year's non-vehicle insurance premium, whether the customer has a company, etc. For a telephone communication system, call recordings of the customer during the service period (e.g., 60 days before the customer's expiration date) are obtained and converted into text to obtain call text information. For platform text and platform voice, chat records between the customer and the agent on the platform are obtained through the platform API or a third-party data collection tool to obtain platform text and platform voice. The platform voice is converted into text to obtain platform converted text. The platform text, platform converted text, call text, and customer information tags obtained above are collected to obtain a customer information set for the customer.

[0049] In this embodiment, the business data of the agent is obtained by accessing the company's sales management system (i.e., the preset database). For example, an SQL query statement is written to extract the agent's renewal rate, plan achievement progress, non-car sales cooperation rate, and non-car premium ratio from the sales management system database, and the above-mentioned data are collected to obtain the agent's seat information set; wherein, the calculation of the renewal rate includes: according to the customer policy information recorded in the sales management system, the number of renewals and the number of policies to be renewed under the name of the agent in the last three months are counted, and the renewal rate is calculated = the number of renewals / the number to be renewed × 100%; the calculation of the progress of the plan achievement includes obtaining the non-car premiums of the agent in the last three months from the sales management system. The planning target and the actual completion amount are used to calculate the planning achievement progress = actual completion amount / planning target amount × 100%; the calculation of the non-auto co-sales rate includes: counting the number of co-sales of auto insurance products and non-auto insurance products and the total number of auto insurance product sales under the name of the seat in the past three months, and calculating the non-auto co-sales rate = number of co-sales / total number of auto insurance product sales × 100%; the calculation of the proportion of non-auto premium income includes: obtaining the non-auto premium income and total premium income under the name of the seat in the past three months, and calculating the proportion of non-auto premium income = non-auto premium income / total premium income × 100%. The above data are integrated to obtain the seat information set of the seat personnel.

[0050] In this embodiment, by obtaining the customer's customer information set and the agent's seat information set, a basis can be provided for subsequent determination of the customer evaluation level and the agent evaluation level, and the target product recommendation range and wording can be matched through the level, thereby improving the accuracy of non-car product recommendations.

[0051] Step S202: Score the customer information set in multiple dimensions using a customer rating model to obtain multiple customer dimension scores, and determine the customer evaluation grade of the customer based on predefined customer dimension weights and the multiple customer dimension scores.

[0052] In this embodiment, a customer grade model is pre-constructed, and the customer grade model includes an input layer, a feature engineering layer, a classifier, a score calculation layer and an output layer; the customer information set is input into the customer grade model, and the scores of dimension one "whether the agreement is reached", dimension two "customer purchasing power", dimension three "customer attitude" and dimension four "customer personality" are calculated by the customer grade model, and the customer evaluation level of the customer is determined based on the pre-defined customer dimension weights and the scores of the above four dimensions.

[0053] In this embodiment, the customer information set is scored in multiple dimensions through a customer rating model to obtain multiple customer dimension scores, and the customer evaluation level of the customer is determined based on pre-defined customer dimension weights and the multiple customer dimension scores. This enables the customer rating model to comprehensively determine the customer evaluation level from multiple dimensions, thereby improving the accuracy of the customer evaluation level, and further improving the adaptability of subsequent non-car recommended products to customers, thereby improving the customer insurance conversion rate, that is, improving the accuracy of non-car product recommendations.

[0054] In one embodiment, the customer information set is scored in multiple dimensions using a customer rating model to obtain multiple customer dimension scores, and the customer evaluation grade of the customer is determined based on predefined customer dimension weights and the multiple customer dimension scores, including:

[0055] inputting the customer information set into the customer grade model through the input layer of the customer grade model;

[0056] Extracting features of text data in the customer information set through a feature engineering layer of the customer grade model to obtain text features, and extracting features of structured data in the customer information set to obtain structured features;

[0057] Outputting a first intent category, a second intent category, a customer personality, and a customer attitude through a classifier of the customer level model according to the text features;

[0058] Scoring and calculating the first intention type, the second intention category, the customer personality, the customer attitude, and the structured features through the scoring calculation layer of the customer grade model to obtain a plurality of customer dimension scores;

[0059] Calculate the target customer score based on the weight corresponding to each customer dimension score;

[0060] It is determined whether the target customer's score belongs to a predefined score range, and the customer evaluation grade of the customer is obtained, and the customer evaluation grade is output through the output layer of the customer grade model.

[0061] In this embodiment, a customer level model is pre-built, and a customer information set (such as text data such as "You can join the group, but the quotation you gave needs to be negotiated" and structured data such as vehicle invoice price, last year's non-vehicle insurance premium, number of companies under the customer's name, etc.) is input into the customer level model through the input layer of the customer level model. The customer information set is processed through the feature engineering layer of the customer level model. Specifically, for text data, the customer information set is first segmented and pre-processed. The segmentation and pre-processing include segmenting the text data through the stuttering segmentation. Perform word segmentation and remove stop words, punctuation marks and noise data. Use vector conversion technology to convert text data into vector representation. Use NLP technology to extract semantic, emotional and intention information of text data to obtain text features. For structured data, perform discretization on structured data, such as binning of vehicle invoice price and last year's non-vehicle insurance premium. For example, if the vehicle invoice price is less than 200,000, it is binned as "low-end", if the vehicle invoice price is between 200,000 and 500,000, it is binned as "mid-end", and if the vehicle invoice price is greater than 500,000, it is binned as "high-end". high-end"; obtain the normal value of non-auto insurance premiums in the previous year, compare the non-auto insurance premiums in the previous year with the normal value, and divide them into high price / normal price / low price; divide the number of companies into no company / a small number of companies and multiple companies, and aggregate the features of the above bins to obtain structured features; based on the above-obtained text features, use multiple classifiers of the customer grade model to output a first intent category, a second intent category, a customer personality, and a customer attitude; and based on the first intent category, second intent category, customer personality, and customer attitude output by the multiple classifiers and the structured features, calculate in the score calculation layer of the customer grade model to obtain multiple customer dimension scores; calculate according to the weight corresponding to each customer dimension score (e.g., 40% for agreement reached, 20% for purchasing power, 30% for attitude, and 10% for personality) to obtain a target customer score, determine whether the target customer score falls within a predefined score range (e.g., excellent 80-100, good 60-79, average 40-59, poor 0-39), obtain a customer evaluation level of the customer, and output the customer evaluation level through the output layer of the customer grade model.

[0062] In this embodiment, the text data and structured data in the customer information are processed separately through the customer grade model, and based on the processing results of the above data, multiple customer dimension scores are generated. Finally, according to the multiple customer dimension scores, the customer evaluation level can be accurately evaluated, which can provide a basis for subsequent target product recommendations, thereby improving the accuracy of non-automotive product recommendations; and each layer of the model is decoupled and independent. If it is necessary to add dimensions in the future, it is only necessary to add the classifier type, without rebuilding the customer grade model, thereby improving the efficiency of non-automotive product recommendations.

[0063] In another embodiment, outputting the first intent category, the second intent category, the customer personality, and the customer attitude by a classifier of the customer level model based on the text features includes:

[0064] Mapping the text features into a first binary classifier and a second binary classifier to obtain a first intent probability distribution and a second intent probability distribution, and outputting the first intent category and the second intent category according to the first intent probability distribution and the second intent probability distribution;

[0065] Mapping the text features to three classifiers to obtain a sentiment probability distribution, and outputting the customer attitude according to the sentiment probability distribution;

[0066] The text features and predefined personality prompt word features are spliced ​​to obtain a splicing vector, the splicing vector is input into a personality classifier to obtain the confidence of each personality, and the personality of the customer is determined according to the confidence.

[0067] In this embodiment, the classifiers of the above-mentioned customer level model include a first binary classifier, a second binary classifier, a third classifier and a personality classifier, wherein the personality classifier refers to a six-classifier, the above-mentioned first intention category refers to whether to agree to the group invitation agreement, and the second intention category refers to whether to agree to the first-day quotation agreement; the first binary classifier and the second binary classifier output yes or no; the above-mentioned customer personality can be pre-classified into six categories: (1) Hesitant decision-making type: the customer shows hesitation, repetition or needs more time to think when making decisions; (2) Quick decision-making type: the customer shows decisiveness, speed or directness when making decisions; (3) Dependence and trust type: the customer shows dependence on the opinions of others and tends to seek help or confirmation; (4) Independence and autonomy type: the customer shows the ability to think independently and make decisions independently; (5) Cost-sensitive type: the customer shows high sensitivity to price or cost and tends to focus on cost-effectiveness; (6) Value-oriented type: the customer pays more attention to the value and long-term benefits of the product rather than the price alone, and the user's customer personality is determined by the personality classifier; the above-mentioned customer attitudes include positive, neutral and negative.

[0068] In this embodiment, first, text features are obtained, and the text features are input into the first binary classifier and the second binary classifier respectively. At this time, the first binary classifier and the second binary classifier will output the probability distribution of the group invitation agreement and the quotation agreement, and the probability distribution is compared with the corresponding threshold value respectively, so as to output the first intention label (such as judging whether the customer has reached a group invitation agreement) through the first binary classifier, and output the second intention label (such as judging whether the customer has reached a quotation agreement) through the second binary classifier; secondly, the text is input into the three classifiers to obtain the emotion probability distribution, such as (positive: 0.7, negative: 0.15, neutral: 0.15), and the label corresponding to the highest emotion probability distribution is selected for output to obtain the customer attitude: positive; finally , obtain the personality prompt words of the above six customer personality categories, and vectorize the personality prompt words to obtain multiple personality prompt word vectors, splice the text features with each personality prompt word vector respectively, to obtain multiple splicing vectors, and gradually input the multiple splicing vectors into the personality classifier to obtain the corresponding confidence level of each personality type. For example, the confidence level corresponding to the hesitant decision-making type is: 0.5, the confidence level corresponding to the quick decision-making type is: 0.1, the confidence level corresponding to the dependent trust type is: 0.6, the confidence level corresponding to the independent type is: 0.3, the confidence level corresponding to the cost-sensitive type is: 0.8, and the confidence level corresponding to the value-oriented type is: 0.9. Select the personality type corresponding to the highest confidence level and output it as the customer personality.

[0069] In this embodiment, the intention to reach a group invitation agreement and a quotation agreement are processed separately by a double binary classifier, which can improve the accuracy of customer intention recognition; and a three-classifier is used to replace the traditional two-class classification, and the "neutral" category is introduced to reduce the misjudgment of boundary samples, thereby improving the accuracy of identifying customer attitudes; through the splicing method of text features and prompt word features, the confidence scores corresponding to each personality type can be calculated quickly and accurately, thereby improving the accuracy and efficiency of customer personality; that is, according to the above technical effects, the accuracy of customer level determination can be improved, and the accuracy of subsequent product recommendations can be improved.

[0070] In another embodiment, the scoring calculation layer of the customer grade model performs scoring calculation on the first intent type, the second intent category, the customer personality, the customer attitude, and the structured features to obtain multiple customer dimension scores, including:

[0071] Calculating a first customer dimension score based on the first intention type, the second intention category, and the customer attitude;

[0072] Calculating a second customer dimension score based on the structured features;

[0073] Calculating the third customer dimension score based on the customer attitude;

[0074] The fourth customer dimension score is calculated based on the customer personality.

[0075] In this embodiment, in the scoring calculation layer of the customer rating model, it is necessary to calculate scores for four customer dimensions, as mentioned above, namely, dimension one, "whether the agreement is reached," dimension two, "customer purchasing power," dimension three, "customer attitude," and dimension four, "customer personality." A predefined calculation rule is used to calculate the first customer dimension score (i.e., the score for dimension one, "whether the agreement is reached") based on the first intent type, the second intent category, and the customer attitude. The specific calculation method is as follows: a base score is assigned based on the dual intent achievement (e.g., dual "yes" = 100 points, single "yes" = 70 points, dual "no" = 30 points). Based on the base score, an adjustment coefficient is determined based on the customer attitude (e.g., positive: base score * 1.2, neutral: base score * 1.0, negative: base score * 0.8). Finally, the first dimension score is output. In one example, if the customer's dual intents are both "yes" and the customer attitude is positive, the first dimension score is: 100 * 1.2 = 120 points. According to the binning interval to which each data in the structured features belongs (such as the binning interval to which the vehicle invoice price belongs, the binning interval to which the non-vehicle insurance premium of the previous year belongs, and the binning interval to which the number of companies belongs), the data score corresponding to each data in the structured data is obtained, and the weighted summation is performed according to the weight corresponding to each data score to obtain the second customer dimension score; the third customer dimension score is determined according to the preset attitude score table and customer attitude, such as when the customer attitude is positive, the third customer dimension score is 50, when the customer attitude is neutral, the third customer dimension score is 0, and when the customer attitude is negative, the third customer dimension score is -50; the fourth customer dimension score is determined according to the preset personality score table, such as the score corresponding to the hesitant decision-making type is 10, the score corresponding to the quick decision-making type is 20, the score corresponding to the dependent trust type is 30, the independent type is 0, the score corresponding to the cost-sensitive type is -10, and the score corresponding to the value-oriented type is 40. The fourth customer dimension score is determined according to the customer's personality.

[0076] In this embodiment, multiple-dimensional scoring calculations are performed based on the first intention type, the second intention category, the customer personality, the customer attitude, and structured features to obtain multiple customer dimension scores, which can provide a basis for the subsequent risk level model to judge the customer evaluation level; and through comprehensive calculation of multiple-dimensional scores, the accuracy of the customer evaluation level can be improved.

[0077] Step S203: Scoring the agent information set in multiple dimensions using an agent grade model to obtain multiple agent dimension scores, and determining the agent evaluation grade of the agent based on predefined agent dimension weights and the multiple agent dimension scores;

[0078] In this embodiment, an agent grade model is constructed in advance, and the agent information set obtained in step S201 is input into the agent grade model. The agent information set is processed by the agent grade model to extract data of multiple dimensions, and the data of multiple dimensions are scored to obtain multiple agent dimension scores. The final total agent score is then calculated based on the weight corresponding to each agent dimension score, and the agent evaluation grade of the agent is determined based on the agent score. The agent evaluation grade includes.

[0079] In this embodiment, the agent information is processed through the agent grade model to obtain the agent evaluation grade of the agent. The agent evaluation grade is used to participate in the subsequent matching of the target product recommendation range, which can improve the accuracy of the target product recommendation range and thereby improve the customer insurance conversion rate, that is, improve the accuracy of non-car product recommendations.

[0080] In one embodiment, the agent rating model is used to score the agent information set in multiple dimensions to obtain multiple agent dimension scores, and the agent evaluation grade of the agent is determined based on predefined agent dimension weights and the multiple agent dimension scores, including:

[0081] Obtain the average value of each dimension in the organization to which the agent belongs;

[0082] Comparing each dimension data in the seat information set with the corresponding average value to determine the indicator status of each dimension data;

[0083] Determine the seat dimension score for each dimension data according to the indicator situation;

[0084] Based on the weight corresponding to each dimension data, weighted summation is performed on all the seat dimension scores to obtain a total seat dimension score;

[0085] The seat evaluation level of the agent is determined based on the total seat dimension score.

[0086] In this embodiment, step S201 mentions that the seat information set of the seat personnel includes: maturity renewal rate, plan achievement progress, non-car co-sales rate and non-car premium ratio; obtain the average value of maturity renewal rate, plan achievement progress, non-car co-sales rate and non-car premium ratio in the institution to which the seat personnel belongs, and compare the maturity renewal rate, plan achievement progress, non-car co-sales rate and non-car premium ratio in the seat information set with the corresponding average value to determine the indicator status of maturity renewal rate, plan achievement progress, non-car co-sales rate and non-car premium ratio. The indicator status includes: exceeding the indicator achievement, normal achievement and failure to achieve. First, assign the corresponding seat dimension score according to the indicator status of maturity renewal rate, plan achievement progress, non-car co-sales rate and non-car premium ratio, and calculate according to the weight corresponding to each seat dimension score to finally obtain the total dimension score of the seat information set, and judge the seat evaluation level of the agent according to the total dimension score.

[0087] In this embodiment, by obtaining the average value of each dimension of the organization and determining the indicator status of each dimension data based on the average value, the benchmark differences between different regions can be eliminated, and the accuracy of the indicator status determination is improved; by performing weighted summation on all the seat dimension scores, the total seat dimension score is obtained; the seat evaluation level of the seat personnel is judged based on the total seat dimension score, which can prevent accidental interference from single data and improve the accuracy of the seat evaluation level.

[0088] Step S204: Obtain a target recommended product set from a predefined non-vehicle product level table based on the customer evaluation level and the seat evaluation level;

[0089] In this embodiment, a target recommended product set is obtained from a predefined non-vehicle product level table according to the customer evaluation level and the seat evaluation level.

[0090] In one embodiment, obtaining a target product recommendation range from a predefined non-vehicle product range table based on the customer evaluation level and the seat evaluation level includes:

[0091] Matching the customer evaluation level with the first adaptation rule to obtain first gear information;

[0092] Matching the seat evaluation level and the second adaptation rule to obtain second gear information;

[0093] Determining a non-vehicle product gear from the predefined non-vehicle product gear table according to the first gear information and the second gear information;

[0094] According to the non-vehicle product level, a corresponding plurality of recommended products is obtained from a non-vehicle product knowledge base;

[0095] A plurality of the recommended products are collected to obtain the target recommended product set.

[0096] In this embodiment, the above-mentioned non-automotive product gear table includes high-end, mid-range, basic and promotional, and each gear corresponds to a set of non-automotive products; it is explained above that the customer evaluation levels include excellent, good, average, and poor; the seat evaluation levels include excellent, good, average, and poor; the above-mentioned first adaptation rule refers to the customer-gear rule, and the customer-gear rule includes: when the customer evaluation level is excellent, the corresponding first gear information is high-end, when the customer evaluation level is good, the corresponding first gear information is mid-end, when the customer evaluation level is average, the corresponding first gear information is basic, and when the customer evaluation level is poor, the corresponding first gear information is promotion; the above-mentioned second adaptation rule refers to the seat-gear rule, and the seat-gear rule includes: when the seat evaluation level is excellent, the corresponding second gear information is high-end, when the seat evaluation level is good, the corresponding second gear information is mid-end, and when the seat evaluation level is average, the corresponding Based on the second gear information, when the seat evaluation level is poor, the corresponding second gear information is promotional; after determining the first gear information and the second gear information, determine the non-vehicle product gear from the predefined non-vehicle product gear table according to the first gear information and the second gear information, specifically, when the first gear information and the second gear information are the same, the non-vehicle product gear is the first gear information or the second gear information; when the first gear information and the second gear information are adjacent, take the first gear information as the non-vehicle product gear; when the first gear information and the second gear information are one digit apart, take the median of the first gear information and the second gear information as the non-vehicle product gear; when the first gear information and the second gear information are two digits apart, take the gear close to the first gear information as the non-vehicle product gear; according to the determined non-vehicle product gear, obtain the corresponding multiple recommended products from the non-vehicle product knowledge base, and collect the multiple recommended products to obtain the target recommended product set.

[0097] In this embodiment, by determining the target recommended product set based on the customer evaluation level and the agent evaluation level, the pertinence and matching degree of the recommended products can be quickly improved, thereby significantly improving the customer underwriting rate, optimizing the operating process, and enhancing customer satisfaction and loyalty.

[0098] Step S205: Analyze the customer information set to obtain a customer information analysis result, and generate a target product recommendation script based on the customer information analysis result, the agent evaluation level, and the target recommended product set.

[0099] In this embodiment, based on the customer dimension information obtained in step S202, the corresponding labels are extracted, that is, the various labels obtained in step S202 are obtained: whether the agreement is reached label (such as quotation contract reached / not reached), customer personality label (such as independent, hesitant and dependent, impulsive decision-making type), customer attitude label (such as positive, neutral, negative), purchasing power label (high, medium, low), and various labels are summarized to obtain customer information analysis results, and the customer information analysis results and the agent evaluation level (indicator achieved, normal achieved and not achieved) and each target recommended product in the target recommended product set are input into a pre-trained speech generation model. The speech generation model will combine the customer feature adaptation dimension, the agent ability compensation dimension and the target product selling point dimension to generate the target product recommendation speech corresponding to each target recommended product; the above-mentioned speech generation model refers to a natural language model (NLP) model, including but not limited to a recurrent neural network (RNN), a Bert model, etc.

[0100] In this embodiment, the customer information analysis results, the agent evaluation level and each target recommended product in the target recommended product set are input into a speech generation model. The speech generation model extracts the various labels of the customer information analysis results, the agent evaluation level and the keywords of the target recommended products, performs feature extraction on the extracted keywords, and maps all extracted features into numerical features (numerical features of each label, numerical features of the agent evaluation level, and numerical features of product keywords); dynamic strategy matching is performed based on the numerical features, such as high-quality customers + substandard agents + high-end products, generating aggressive speech, highlighting limited-time discounts and exclusive services, etc., and ordinary customers + high-quality agents + basic products, generating conservative speech, emphasizing cost-effectiveness and basic guarantees.

[0101] Among them, in addition to the above-mentioned customer information analysis results, the agent evaluation level and each target recommended product in the target recommended product set, the customer's implementation data can also be obtained, including the customer's recent browsing history, etc. For example, when a customer frequently inquires about accident insurance, the target product recommendation script corresponding to the accident insurance will be pushed preferentially from the target recommended product set.

[0102] In this embodiment, the customer information set is analyzed to obtain customer information analysis results, and target product recommendation scripts are generated based on the customer information analysis results, the agent evaluation level, and the target recommended product set. The scripts generated based on customers, agents, and target recommended products improve the accuracy of non-auto product recommendations and increase customers' underwriting rates.

[0103] It should be emphasized that in order to further ensure the privacy and security of the above-mentioned customer information set and seat information set, the above-mentioned customer information set and seat information set can also be stored in a node of a blockchain.

[0104] The blockchain referred to in this application refers to a new application model for computer technologies such as distributed data storage, peer-to-peer transmission, consensus mechanisms, and encryption algorithms. Blockchain is essentially a decentralized database, a series of data blocks generated using cryptographic methods. Each data block contains information about a batch of network transactions, which is used to verify the validity of the information (to prevent counterfeiting) and generate the next block. Blockchain can include the blockchain underlying platform, the platform product service layer, and the application service layer.

[0105] The embodiments of the present application can acquire and process relevant data based on artificial intelligence technology. Artificial Intelligence (AI) is the theory, method, technology, and application system that uses digital computers or machines controlled by digital computers to simulate, extend, and expand human intelligence, perceive the environment, acquire knowledge, and use knowledge to achieve optimal results.

[0106] Fundamental AI technologies generally include sensors, dedicated AI chips, cloud computing, distributed storage, big data processing, operating / interaction systems, and mechatronics. AI software technologies primarily encompass computer vision, robotics, biometrics, speech processing, natural language processing, and machine learning / deep learning.

[0107] Those skilled in the art will appreciate that all or part of the processes in the above-described method embodiments can be implemented by instructing related hardware via computer-readable instructions. The computer-readable instructions can be stored in a computer-readable storage medium, and when the program is executed, it can include the processes in the above-described method embodiments. The aforementioned storage medium can be a non-volatile storage medium such as a magnetic disk, an optical disk, a read-only memory (ROM), or a random access memory (RAM).

[0108] It should be understood that although the steps in the flowcharts of the accompanying drawings are shown in sequence as indicated by the arrows, these steps are not necessarily executed in the order indicated by the arrows. Unless otherwise specified herein, there is no strict order restriction on the execution of these steps, and they can be executed in other orders. Moreover, at least some of the steps in the flowcharts of the accompanying drawings may include multiple sub-steps or multiple stages, and these sub-steps or stages are not necessarily executed at the same time, but can be executed at different times, and their execution order is not necessarily sequential, but can be executed in turn or alternately with other steps or at least a portion of the sub-steps or stages of other steps.

[0109] Further references Figure 3 , as a response to the above Figure 2 The present application provides an embodiment of a device for generating a recommendation phrase for a non-car product. Figure 2 Corresponding to the method embodiment shown, the apparatus can be specifically applied to various computer devices.

[0110] like Figure 3 As shown, the device 300 for generating recommendation words for non-car products in this embodiment includes: an information acquisition module 301, a first level judgment module 302, a second level judgment module 303, a product acquisition module 304, and a word generation module 305. Among them:

[0111] Information acquisition module 301, used to obtain the customer information set of the customer and the seat information set of the seat staff;

[0112] In one embodiment, the information acquisition module includes:

[0113] An information acquisition submodule is used to obtain the customer information tag of the customer and the call recording, platform voice and platform text in the server;

[0114] a conversion submodule, configured to convert the call recording and the platform voice into text format using speech recognition technology to obtain a call text and a platform converted text, and to aggregate the platform text, the call text, the platform converted text, and the customer information tag to obtain the customer information set;

[0115] The calculation submodule is used to obtain the business data of the agent from a preset database, calculate the business data, and obtain the agent information set.

[0116] A first rating judgment module 302 is configured to score the customer information set in multiple dimensions using a customer rating model to obtain multiple customer dimension scores, and to determine the customer evaluation level of the customer based on predefined customer dimension weights and the multiple customer dimension scores;

[0117] In one embodiment, the first level determination module includes:

[0118] A first input submodule, configured to input the customer information set into the customer grade model through an input layer of the customer grade model;

[0119] a feature extraction submodule, configured to extract features of text data in the customer information set through a feature engineering layer of the customer grade model to obtain text features, and to extract features of structured data in the customer information set to obtain structured features;

[0120] a classification submodule, configured to output a first intent category, a second intent category, a customer personality, and a customer attitude through a classifier of the customer grade model according to the text features;

[0121] A first dimension scoring submodule is configured to score and calculate the first intention type, the second intention category, the customer personality, the customer attitude, and the structured features through the scoring calculation layer of the customer grade model to obtain a plurality of customer dimension scores;

[0122] A first weight calculation submodule is used to calculate the target customer score based on the weight corresponding to each customer dimension score;

[0123] The customer grade judgment submodule is used to judge whether the target customer score belongs to a predefined score range, obtain the customer evaluation grade of the customer, and output the customer evaluation grade through the output layer of the customer grade model.

[0124] In another embodiment, the classification submodule includes:

[0125] a first output subunit, configured to map the text features into a first binary classifier and a second binary classifier to obtain a first intent probability distribution and a second intent probability distribution, and output the first intent category and the second intent category according to the first intent probability distribution and the second intent probability distribution;

[0126] A second output subunit is used to map the text features into three classifiers to obtain a sentiment probability distribution, and output the customer attitude according to the sentiment probability distribution;

[0127] The personality determination subunit is used to splice the text features and predefined personality prompt word features to obtain a splicing vector, input the splicing vector into the personality classifier to obtain the confidence of each personality, and determine the customer personality based on the confidence.

[0128] In another embodiment, the first dimension scoring submodule includes:

[0129] A first dimension scoring subunit, configured to calculate a first customer dimension score based on the first intention type, the second intention category, and the customer attitude;

[0130] A second dimension scoring subunit, configured to calculate a second customer dimension score based on the structured features;

[0131] A third dimension scoring subunit, configured to calculate the third customer dimension score based on the customer attitude;

[0132] The fourth dimension scoring subunit is used to calculate the fourth customer dimension score according to the customer personality.

[0133] The second grade judgment module 303 is configured to score the agent information set in multiple dimensions using an agent grade model to obtain multiple agent dimension scores, and determine the agent evaluation grade of the agent based on predefined agent dimension weights and the multiple agent dimension scores;

[0134] In one embodiment, the second level calculation includes:

[0135] The external mean acquisition submodule is used to obtain the average value of each dimension in the organization to which the agent belongs;

[0136] A comparison submodule, configured to compare each dimension data in the seat information set with the corresponding average value to determine the indicator status of each dimension data;

[0137] The seat dimension score calculation submodule is used to determine the seat dimension score of each dimension data according to the indicator situation;

[0138] A second weight calculation submodule is configured to perform a weighted summation of all the seat dimension scores based on the weight corresponding to each dimension data to obtain a total seat dimension score;

[0139] The seat level judgment submodule is used to judge the seat evaluation level of the seat personnel based on the total seat dimension score.

[0140] The product acquisition module 304 is configured to acquire a target recommended product set from a predefined non-vehicle product level table based on the customer evaluation level and the seat evaluation level;

[0141] In one embodiment, the product acquisition module includes:

[0142] A first matching submodule, configured to perform matching based on the customer evaluation level and a first adaptation rule to obtain first gear information;

[0143] A second matching submodule is configured to perform matching according to the seat evaluation level and the second adaptation rule to obtain second gear information;

[0144] A product gear position determination submodule, configured to determine a non-vehicle product gear position from the predefined non-vehicle product gear position table according to the first gear position information and the second gear position information;

[0145] A recommended product acquisition submodule is used to acquire a plurality of corresponding recommended products from a non-vehicle product knowledge base according to the non-vehicle product gear;

[0146] The recommended product aggregation submodule is used to aggregate multiple recommended products to obtain the target recommended product set.

[0147] The speech generation module 305 is used to analyze the customer information set to obtain the customer information analysis results, and generate target product recommendation speech based on the customer information analysis results, the agent evaluation level and the target recommended product set.

[0148] In this embodiment, by obtaining the customer's customer information set and the agent's seat information set, a basis can be provided for subsequently determining the customer evaluation level and the agent evaluation level. The target product recommendation range and sales pitch are matched based on the level, thereby improving the accuracy of non-car product recommendations.

[0149] By performing a multi-dimensional scoring calculation based on the first intention type, the second intention category, the customer personality, the customer attitude, and the structured features, multiple customer dimension scores are obtained, which can provide a basis for the subsequent risk level model to determine the customer evaluation level; and the comprehensive calculation of the multi-dimensional scores can improve the accuracy of the customer evaluation level;

[0150] By obtaining the average value of each dimension in the organization and determining the indicator status of each dimension data based on the average value, the benchmark differences between different regions can be eliminated, thereby improving the accuracy of the indicator status determination; by performing a weighted summation of all the seat dimension scores, a total seat dimension score is obtained; and the seat evaluation level of the seat personnel is determined based on the total seat dimension score, thereby preventing accidental interference from a single data and improving the accuracy of the seat evaluation level;

[0151] By determining a target product set based on customer and agent ratings, we can quickly improve the relevance and matching of recommended products, thereby significantly increasing customer underwriting rates, optimizing operational processes, and enhancing customer satisfaction and loyalty.

[0152] By analyzing the customer information set, the customer information analysis results are obtained, and according to the customer information analysis results, the agent evaluation level and the target recommended product set, the target product recommendation script is generated. The script generated based on the customer, the agent and the target recommended product improves the accuracy of non-auto product recommendations and increases the customer's underwriting rate.

[0153] In order to solve the above technical problems, the embodiment of the present application also provides a device (computer device). Figure 4 , Figure 4 This is a basic structural block diagram of the computer device in this embodiment.

[0154] The computer device 4 includes a memory 41, a processor 42, and a network interface 43 that are interconnected through a system bus. It should be noted that the figure only shows a computer device 4 with a memory 41, a processor 42, and a network interface 43, but it should be understood that it is not required to implement all the components shown, and more or fewer components can be implemented instead. Among them, those skilled in the art can understand that the computer device here is a device that can automatically perform numerical calculations and / or information processing according to pre-set or stored instructions, and its hardware includes but is not limited to a microprocessor, an application-specific integrated circuit (ASIC), a field-programmable gate array (FPGA), a digital signal processor (DSP), an embedded device, etc.

[0155] The computer device may be a desktop computer, notebook computer, PDA, cloud server, etc. The computer device may interact with the user via a keyboard, mouse, remote control, touchpad, or voice control device.

[0156] The memory 41 includes at least one type of readable storage medium, including flash memory, hard disk, multimedia card, card-type memory (e.g., SD or DX memory), random access memory (RAM), static random access memory (SRAM), read-only memory (ROM), electrically erasable programmable read-only memory (EEPROM), programmable read-only memory (PROM), magnetic memory, magnetic disk, optical disk, etc. In some embodiments, the memory 41 can be an internal storage unit of the computer device 4, such as the hard disk or memory of the computer device 4. In other embodiments, the memory 41 can also be an external storage device of the computer device 4, such as a plug-in hard disk equipped on the computer device 4, a smart memory card (SMC), a secure digital (SD) card, a flash memory card, etc. Of course, the memory 41 can also include both the internal storage unit of the computer device 4 and its external storage device. In this embodiment, the memory 41 is generally used to store the operating system and various application software installed on the computer device 4, such as computer-readable instructions for the method for generating recommendation scripts for non-automotive products. In addition, the memory 41 can also be used to temporarily store various types of data that have been output or are to be output.

[0157] In some embodiments, the processor 42 may be a central processing unit (CPU), a controller, a microcontroller, a microprocessor, or other data processing chip. The processor 42 is typically used to control the overall operation of the computer device 4. In this embodiment, the processor 42 is used to execute computer-readable instructions or process data stored in the memory 41, such as computer-readable instructions for executing the method for generating recommendation scripts for non-automotive products.

[0158] The network interface 43 may include a wireless network interface or a wired network interface. The network interface 43 is generally used to establish a communication connection between the computer device 4 and other electronic devices.

[0159] During implementation, the electronic device of the present application obtains a customer's customer information set and an agent's seat information set, thereby providing a basis for subsequently determining the customer's evaluation level and the agent's evaluation level. The levels are then used to match the target product recommendation range and sales pitch, thereby improving the accuracy of non-automotive product recommendations.

[0160] By performing a multi-dimensional scoring calculation based on the first intention type, the second intention category, the customer personality, the customer attitude, and the structured features, multiple customer dimension scores are obtained, which can provide a basis for the subsequent risk level model to determine the customer evaluation level; and the comprehensive calculation of the multi-dimensional scores can improve the accuracy of the customer evaluation level;

[0161] By obtaining the average value of each dimension in the organization and determining the indicator status of each dimension data based on the average value, the benchmark differences between different regions can be eliminated, thereby improving the accuracy of the indicator status determination; by performing a weighted summation of all the seat dimension scores, a total seat dimension score is obtained; and the seat evaluation level of the seat personnel is determined based on the total seat dimension score, thereby preventing accidental interference from a single data and improving the accuracy of the seat evaluation level;

[0162] By determining a target product set based on customer and agent ratings, we can quickly improve the relevance and matching of recommended products, thereby significantly increasing customer underwriting rates, optimizing operational processes, and enhancing customer satisfaction and loyalty.

[0163] By analyzing the customer information set, the customer information analysis results are obtained, and according to the customer information analysis results, the agent evaluation level and the target recommended product set, the target product recommendation script is generated. The script generated based on the customer, the agent and the target recommended product improves the accuracy of non-auto product recommendations and increases the customer's underwriting rate.

[0164] The present application also provides another embodiment, namely, providing a storage medium (computer-readable storage medium), wherein the computer-readable storage medium stores computer-readable instructions, and the computer-readable instructions can be executed by at least one processor to enable the at least one processor to perform the steps of the method for generating recommendation scripts for non-automotive products as described above.

[0165] During implementation, the computer-readable storage medium of the present application can obtain a customer information set and an agent information set, thereby providing a basis for subsequently determining the customer evaluation level and the agent evaluation level. The levels are then used to match the target product recommendation range and sales pitch, thereby improving the accuracy of non-automotive product recommendations.

[0166] By performing a multi-dimensional scoring calculation based on the first intention type, the second intention category, the customer personality, the customer attitude, and the structured features, multiple customer dimension scores are obtained, which can provide a basis for the subsequent risk level model to determine the customer evaluation level; and the comprehensive calculation of the multi-dimensional scores can improve the accuracy of the customer evaluation level;

[0167] By obtaining the average value of each dimension in the organization and determining the indicator status of each dimension data based on the average value, the benchmark differences between different regions can be eliminated, thereby improving the accuracy of the indicator status determination; by performing a weighted summation of all the seat dimension scores, a total seat dimension score is obtained; and the seat evaluation level of the seat personnel is determined based on the total seat dimension score, thereby preventing accidental interference from a single data and improving the accuracy of the seat evaluation level;

[0168] By determining a target product set based on customer and agent ratings, we can quickly improve the relevance and matching of recommended products, thereby significantly increasing customer underwriting rates, optimizing operational processes, and enhancing customer satisfaction and loyalty.

[0169] By analyzing the customer information set, the customer information analysis results are obtained, and according to the customer information analysis results, the agent evaluation level and the target recommended product set, the target product recommendation script is generated. The script generated based on the customer, the agent and the target recommended product improves the accuracy of non-auto product recommendations and increases the customer's underwriting rate.

[0170] The non-Company software tools or components appearing in the embodiments of this application are merely examples and do not represent actual use.

[0171] Through the description of the above implementation methods, those skilled in the art can clearly understand that the above-mentioned embodiment methods can be implemented by means of software plus the necessary general hardware platform, and of course can also be implemented by hardware, but in many cases the former is a better implementation method. Based on this understanding, the technical solution of the present application, or the part that contributes to the prior art, can be embodied in the form of a software product, which is stored in a storage medium (such as ROM / RAM, magnetic disk, optical disk), and includes a number of instructions for enabling a terminal device (which can be a mobile phone, computer, server, air conditioner, or network device, etc.) to execute the methods described in each embodiment of the present application.

[0172] Obviously, the embodiments described above are only some of the embodiments of the present application, rather than all of the embodiments. The preferred embodiments of the present application are given in the accompanying drawings, but they do not limit the patent scope of the present application. The present application can be implemented in many different forms. On the contrary, the purpose of providing these embodiments is to make the understanding of the disclosure of the present application more thorough and comprehensive. Although the present application has been described in detail with reference to the aforementioned embodiments, for those skilled in the art, it is still possible to modify the technical solutions described in the aforementioned specific embodiments, or to make equivalent replacements for some of the technical features therein. Any equivalent structure made using the contents of the present application specification and the accompanying drawings, directly or indirectly used in other related technical fields, is also within the scope of patent protection of the present application.

Claims

1. A method for generating recommendation words for non-automotive products, characterized in that: The method comprises: Get the customer's customer information set and the agent's seat information set; Scoring the customer information set in multiple dimensions using a customer rating model to obtain multiple customer dimension scores, and determining the customer evaluation level of the customer based on predefined customer dimension weights and the multiple customer dimension scores; Scoring the seat information set in multiple dimensions using a seat grading model to obtain multiple seat dimension scores, and determining the seat evaluation grade of the agent based on predefined seat dimension weights and the multiple seat dimension scores; Obtaining a target recommended product set from a predefined non-vehicle product tier table based on the customer evaluation level and the seat evaluation level; The customer information set is analyzed to obtain a customer information analysis result, and a target product recommendation script is generated based on the customer information analysis result, the agent evaluation level, and the target recommended product set.

2. The method for generating recommendation words for non-automotive products according to claim 1, characterized in that: The method of obtaining the customer information set of the customer and the agent information set of the agent includes: Obtain the customer information tag of the customer and the call recordings, platform voice and platform text in the server; Converting the call recording and the platform voice into text format using speech recognition technology to obtain a call text and a platform converted text, and collecting the platform text, the call text, the platform converted text, and the customer information tag to obtain the customer information set; The business data of the agent is obtained from a preset database, and the business data is calculated to obtain the agent information set.

3. The method for generating recommendation words for non-automotive products according to claim 1, characterized in that: Scoring the customer information set in multiple dimensions using a customer rating model to obtain multiple customer dimension scores, and determining the customer evaluation level of the customer based on predefined customer dimension weights and the multiple customer dimension scores, includes: inputting the customer information set into the customer grade model through the input layer of the customer grade model; Extracting features of text data in the customer information set through a feature engineering layer of the customer grade model to obtain text features, and extracting features of structured data in the customer information set to obtain structured features; Outputting a first intent category, a second intent category, a customer personality, and a customer attitude through a classifier of the customer level model according to the text features; Scoring and calculating the first intention type, the second intention category, the customer personality, the customer attitude, and the structured features through the scoring calculation layer of the customer grade model to obtain a plurality of customer dimension scores; Calculate the target customer score based on the weight corresponding to each customer dimension score; It is determined whether the target customer's score belongs to a predefined score range, and the customer evaluation grade of the customer is obtained, and the customer evaluation grade is output through the output layer of the customer grade model.

4. The method for generating recommendation words for non-automotive products according to claim 3, characterized in that: Outputting the first intent category, the second intent category, the customer personality, and the customer attitude through the classifier of the customer level model according to the text features includes: Mapping the text features into a first binary classifier and a second binary classifier to obtain a first intent probability distribution and a second intent probability distribution, and outputting the first intent category and the second intent category according to the first intent probability distribution and the second intent probability distribution; Mapping the text features to three classifiers to obtain a sentiment probability distribution, and outputting the customer attitude according to the sentiment probability distribution; The text features and predefined personality prompt word features are spliced ​​to obtain a splicing vector, the splicing vector is input into a personality classifier to obtain the confidence of each personality, and the personality of the customer is determined according to the confidence.

5. The method for generating recommendation words for non-automotive products according to claim 3, characterized in that: The scoring calculation layer of the customer grade model performs scoring calculation on the first intent type, the second intent category, the customer personality, the customer attitude, and the structured features to obtain multiple customer dimension scores, including: Calculating a first customer dimension score based on the first intention type, the second intention category, and the customer attitude; Calculating a second customer dimension score based on the structured features; Calculating a third customer dimension score based on the customer attitude; A fourth customer dimension score is calculated based on the customer personality.

6. The method for generating recommendation words for non-automotive products according to claim 1, characterized in that: Scoring the seat information set in multiple dimensions using the seat grade model to obtain multiple seat dimension scores, and determining the seat evaluation grade of the agent based on predefined seat dimension weights and the multiple seat dimension scores, includes: Obtain the average value of each dimension in the organization to which the agent belongs; Comparing each dimension data in the seat information set with the corresponding average value to determine the indicator status of each dimension data; Determine the seat dimension score for each dimension data according to the indicator situation; Based on the weight corresponding to each dimension data, weighted summation is performed on all the seat dimension scores to obtain a total seat dimension score; The seat evaluation level of the agent is determined based on the total seat dimension score.

7. The method for generating recommendation words for non-automotive products according to claim 1, characterized in that: The step of obtaining a target product recommendation range from a predefined non-vehicle product range table based on the customer evaluation level and the seat evaluation level includes: Matching the customer evaluation level with the first adaptation rule to obtain first gear information; Matching the seat evaluation level and the second adaptation rule to obtain second gear information; Determining a non-vehicle product gear from the predefined non-vehicle product gear table according to the first gear information and the second gear information; According to the non-vehicle product level, a corresponding plurality of recommended products is obtained from a non-vehicle product knowledge base; A plurality of the recommended products are collected to obtain the target recommended product set.

8. A device for generating recommendation words for non-automotive products, characterized in that: The device comprises: The information acquisition module is used to obtain the customer information set of the customer and the seat information set of the agent; A first grade judgment module is configured to score the customer information set in multiple dimensions using a customer grade model to obtain multiple customer dimension scores, and to determine the customer evaluation grade of the customer based on predefined customer dimension weights and the multiple customer dimension scores; A second grade judgment module is configured to score the agent information set in multiple dimensions using an agent grade model to obtain multiple agent dimension scores, and determine the agent evaluation grade of the agent based on predefined agent dimension weights and the multiple agent dimension scores; A product acquisition module, configured to acquire a target recommended product set from a predefined non-vehicle product level table based on the customer evaluation level and the seat evaluation level; The speech generation module is used to analyze the customer information set to obtain the customer information analysis results, and generate target product recommendation speech based on the customer information analysis results, the seat evaluation level and the target recommended product set.

9. A computer device, characterized in that: The computer device comprises: at least one processor; and, a memory communicatively connected to the at least one processor; wherein, The memory stores instructions that can be executed by the at least one processor, and the instructions are executed by the at least one processor so that the at least one processor can execute the method for generating recommendation words for non-vehicle products as described in any one of claims 1 to 7.

10. A computer-readable storage medium storing a computer program, characterized in that: When the computer program is executed by a processor, the method for generating recommendation words for non-automotive products as described in any one of claims 1 to 7 is implemented.

Images