The follow-up method, clue follow-up device, electronic equipment and storage medium

Abstract

The embodiment of the invention provides a clue follow-up method, a clue follow-up device, electronic equipment and a storage medium, and belongs to the technical field of artificial intelligence. The method comprises the following steps: obtaining business operation clues and a clue scoring model, scoring the business operation clues through the clue scoring model to obtain a target clue score, sorting the business operation clues according to the target clue score to obtain a clue sorting sequence, and classifying the business operation clues according to the clue sorting sequence to obtain a classification result. According to the method and the device, the qualified clues and the non-qualified clues are obtained, the qualified clues are followed according to the clue sorting sequence, the clue following intention of the non-qualified clues is obtained, and the non-qualified clues are followed according to the clue following intention, so that the clue following efficiency can be improved.

Description

Technical Field

[0001] The present application relates to the field of artificial intelligence technology, is applicable to the field of financial technology technology, and particularly relates to a lead tracking method, a lead tracking device, an electronic device and a storage medium. Background Art

[0002] Customer business leads refer to the lead information of users or potential users who intend to purchase the products for sale. In the related art, each customer business lead is manually checked and sorted, and then each customer business lead is followed up one by one, which results in a lot of time spent on lead follow-up. When faced with a large number of customer business leads, the leads cannot be followed up in time, resulting in low efficiency in lead follow-up. Summary of the invention

[0003] The main purpose of the embodiments of the present application is to propose a clue following method, a clue following device, an electronic device and a storage medium, aiming to improve the efficiency of clue following.

[0004] To achieve the above-mentioned purpose, a first aspect of an embodiment of the present application proposes a clue following method, the method comprising:

[0005] Obtain business operation leads and lead scoring models;

[0006] Scoring the business operation leads by using the lead scoring model to obtain a target lead score;

[0007] Sorting the business operation leads according to the target lead scores to obtain a lead sorting order;

[0008] Classifying the business operation leads according to the lead sorting order to obtain qualified leads and unqualified leads;

[0009] Following up on the qualified leads according to the lead ranking order;

[0010] The lead follow-up intention of the unqualified lead is obtained, and the unqualified lead is followed up according to the lead follow-up intention.

[0011] In some embodiments, the clue scoring model is trained according to the following steps:

[0012] Obtain sample business leads and sample lead scores;

[0013] The model parameters of the first preset scoring model are updated according to the sample business leads and the sample lead scores to obtain a first scoring sub-model;

[0014] The model parameters of the second preset scoring model are updated according to the sample business clues and the first scoring sub-model to obtain a second scoring sub-model;

[0015] The clue scoring model is determined according to the first scoring sub-model and the second scoring sub-model.

[0016] In some embodiments, updating the model parameters of the second preset scoring model according to the sample business clue and the first scoring sub-model to obtain the second scoring sub-model includes:

[0017] Obtaining a model gradient of the first scoring sub-model to obtain model residual data;

[0018] Performing a lead score on the sample business lead using the second preset scoring model to obtain a first predicted lead score;

[0019] Perform loss calculation according to the model residual data and the first prediction clue score to obtain first loss data;

[0020] The model parameters of the second preset scoring model are updated according to the first loss data to obtain the second scoring sub-model.

[0021] In some embodiments, determining the clue scoring model according to the first scoring sub-model and the second scoring sub-model includes:

[0022] Obtaining model weight data of the second scoring sub-model;

[0023] Performing model weighting on the second scoring sub-model according to the model weight data to obtain a reference scoring sub-model;

[0024] The first scoring sub-model and the reference scoring sub-model are subjected to model aggregation to obtain the clue scoring model.

[0025] In some embodiments, obtaining the model weight data of the second scoring sub-model includes:

[0026] Performing a clue score on the sample business clue by using the first scoring sub-model to obtain a second predicted clue score;

[0027] Calculate the loss according to the second prediction clue score and the sample clue score to obtain second loss data;

[0028] Performing a clue score on the sample business clue by using the second scoring sub-model to obtain a third predicted clue score;

[0029] Determining third loss data according to the second loss data, the third prediction clue score and a preset model weight;

[0030] The preset model weight is updated according to the third loss data to obtain the model weight data.

[0031] In some embodiments, the performing model aggregation on the first scoring sub-model and the reference scoring sub-model to obtain the clue scoring model includes:

[0032] The first scoring sub-model and the reference scoring sub-model perform model aggregation to obtain a candidate scoring model;

[0033] Obtaining the model scoring accuracy of the candidate scoring model;

[0034] The candidate scoring models are screened according to the model scoring accuracy to obtain the clue scoring model.

[0035] In some embodiments, updating the model parameters of the first preset scoring model according to the sample business leads and the sample lead scores to obtain a first scoring sub-model includes:

[0036] Performing a clue score on the sample business clue by using the first preset scoring model to obtain a fourth predicted clue score;

[0037] Calculate the loss according to the fourth prediction clue score and the sample clue score to obtain fourth loss data;

[0038] The model parameters of the first preset scoring model are updated according to the fourth loss data to obtain the first scoring sub-model.

[0039] To achieve the above-mentioned purpose, a second aspect of an embodiment of the present application provides a clue following device, the device comprising:

[0040] The acquisition module is used to obtain business operation leads and lead scoring models;

[0041] A scoring module, used to score the business operation leads by using the lead scoring model to obtain a target lead score;

[0042] A sorting module, used to sort the business operation leads according to the target lead scores to obtain a lead sorting order;

[0043] A classification module, used to classify the business operation leads according to the lead sorting order to obtain qualified leads and unqualified leads;

[0044] A first follow-up module, used for following up the qualified leads according to the lead sorting order;

[0045] The second follow-up module is used to obtain the lead follow-up intention of the unqualified lead and follow up the unqualified lead according to the lead follow-up intention.

[0046] To achieve the above objectives, a third aspect of an embodiment of the present application proposes an electronic device, which includes a memory and a processor, wherein the memory stores a computer program, and when the processor executes the computer program, the method described in the first aspect is implemented.

[0047] To achieve the above-mentioned purpose, the fourth aspect of an embodiment of the present application proposes a computer-readable storage medium, wherein the computer-readable storage medium stores a computer program, and when the computer program is executed by a processor, the method described in the first aspect is implemented.

[0048] The lead follow-up method, lead follow-up device, electronic device and storage medium proposed in the embodiment of the present application obtain business operation leads and lead scoring models, score business operation leads through the lead scoring model, and obtain target lead scores, so as to determine the probability that business operation leads can be converted into business orders by using the target lead scores. In order to obtain the lead priority of the business operation leads, the business operation leads are sorted according to the target lead scores to obtain the lead sorting order. The business operation leads are classified according to the lead sorting order to evaluate the quality of the leads and obtain qualified leads and unqualified leads. Following up qualified leads according to the lead sorting order can timely follow up business operation leads with higher priorities, reduce the time cost of combing each business operation lead, and improve the efficiency of lead follow-up. Obtain the lead follow-up intention of unqualified leads, and follow up unqualified leads according to the lead follow-up intention, without combing the leads one by one and following up on any leads, thereby improving the efficiency of lead follow-up. BRIEF DESCRIPTION OF THE DRAWINGS

[0049] Figure 1 is a flow chart of a clue following method provided in an embodiment of the present application;

[0050] Figure 2 is a flow chart of the training process of the clue scoring model provided in the embodiment of the present application;

[0051] Figure 3 yes Figure 2 Flow chart of step S220 in FIG.

[0052] Figure 4 yes Figure 2 Flow chart of step S230 in FIG.

[0053] Figure 5 yes Figure 2 Flow chart of step S240 in FIG.

[0054] Figure 6 yes Figure 5 Flow chart of step S510 in FIG.

[0055] Figure 7 yes Figure 5 Flow chart of step S530 in FIG.

[0056] Figure 8 is a schematic diagram of the structure of a clue following device provided in an embodiment of the present application;

[0057] Fig. 9 It is a schematic diagram of the hardware structure of the electronic device provided in the embodiment of the present application. DETAILED DESCRIPTION

[0058] In order to make the purpose, technical solution and advantages of the present application more clearly understood, the present application is further described in detail below in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present application and are not used to limit the present application.

[0059] It should be noted that, although the functional modules are divided in the device schematic diagram and the logical order is shown in the flowchart, in some cases, the steps shown or described may be performed in a different order than the module division in the device or the order in the flowchart. The terms "first", "second", etc. in the specification, claims and the above drawings are used to distinguish similar objects, and are not necessarily used to describe a specific order or sequence.

[0060] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as those commonly understood by those skilled in the art to which this application belongs. The terms used herein are only for the purpose of describing the embodiments of this application and are not intended to limit this application.

[0061] Customer business leads refer to the lead information of users or potential users who intend to purchase the products for sale. In the related art, each customer business lead is manually checked and sorted, and then each customer business lead is followed up one by one, which results in a lot of time spent on lead follow-up. When faced with a large number of customer business leads, the leads cannot be followed up in time, resulting in low efficiency in lead follow-up.

[0062] Based on this, the embodiments of the present application provide a clue following method, a clue following device, an electronic device and a computer-readable storage medium, aiming to improve the efficiency of clue following.

[0063] The clue following method, clue following device, electronic device and computer-readable storage medium provided in the embodiments of the present application are specifically illustrated through the following embodiments. First, the clue following method in the embodiments of the present application is described.

[0064] The clue following method provided in the embodiment of the present application relates to the field of artificial intelligence technology. The clue following method provided in the embodiment of the present application can be applied to a terminal, can also be applied to a server side, and can also be software running in a terminal or a server side. In some embodiments, the terminal can be a smart phone, a tablet computer, a laptop computer, a desktop computer, etc.; the server side can be configured as an independent physical server, or a server cluster or distributed system composed of multiple physical servers, or a cloud server that provides basic cloud computing services such as cloud services, cloud databases, cloud computing, cloud functions, cloud storage, network services, cloud communications, middleware services, domain name services, security services, CDN, and big data and artificial intelligence platforms; the software can be an application that implements the clue following method, etc., but is not limited to the above forms.

[0065] The present application can be used in many general or special computer system environments or configurations. For example: personal computers, server computers, handheld or portable devices, tablet devices, multiprocessor systems, microprocessor-based systems, set-top boxes, programmable consumer electronics, network PCs, minicomputers, mainframe computers, distributed computing environments including any of the above systems or devices, etc. The present application can be described in the general context of computer-executable instructions executed by a computer, such as program modules. Generally, program modules include routines, programs, objects, components, data structures, etc. that perform specific tasks or implement specific abstract data types. The present application can also be practiced in distributed computing environments, in which tasks are performed by remote processing devices connected through a communication network. In a distributed computing environment, program modules can be located in local and remote computer storage media including storage devices.

[0066] It should be noted that in each specific implementation of the present application, when it comes to the need to perform relevant processing based on data related to user identity or characteristics such as user information, user behavior data, user historical data, and user location information, the user's permission or consent will be obtained first, and the collection, use, and processing of these data will comply with relevant laws, regulations, and standards. In addition, when the embodiment of the present application needs to obtain the user's sensitive personal information, the user's separate permission or consent will be obtained through a pop-up window or by jumping to a confirmation page. After clearly obtaining the user's separate permission or consent, the necessary user-related data for the normal operation of the embodiment of the present application will be obtained.

[0067] Figure 1 It is an optional flow chart of the lead following method provided in the embodiment of the present application. The lead following method may include but is not limited to steps S110 to S160.

[0068] Step S110, obtaining business operation leads and lead scoring models;

[0069] Step S120, scoring the business operation leads using the lead scoring model to obtain a target lead score;

[0070] Step S130, sorting the business operation leads according to the target lead scores to obtain a lead sorting order;

[0071] Step S140, classifying the business operation leads according to the lead sorting order to obtain qualified leads and unqualified leads;

[0072] Step S150, following up on qualified leads according to the lead ranking order;

[0073] Step S160, obtaining the lead follow-up intention of the unqualified lead, and following up the unqualified lead according to the lead follow-up intention.

[0074] Steps S110 to S160 shown in the embodiment of the present application classify business operation leads according to the order of lead sorting to evaluate the quality of the leads and obtain qualified leads and unqualified leads. Following up on qualified leads according to the order of lead sorting can timely follow up on business operation leads with higher priority, reducing the time cost of sorting out each business operation lead and improving the efficiency of lead follow-up. Obtaining the lead follow-up intention of unqualified leads, and following up on unqualified leads according to the lead follow-up intention, without having to sort out the leads one by one and follow up on any leads, improves the efficiency of lead follow-up.

[0075] In step S110 of some embodiments, customer portraits are identified and screened according to business operation tasks such as trusts and funds to generate customer operation leads. Customer operation leads refer to clue information of users or potential users who intend to purchase the products for sale. Customer operation leads include basic customer information, purchase history, and customer behavior data. Customer behavior data includes application usage data and product usage data. Application usage data can be page views, page viewing time, etc., and product usage data can be product usage frequency, product usage duration, etc. In the insurance field, customer operation leads can be customer information that is interested in insurance products such as auto insurance and life insurance. The lead scoring model is used to score business operation leads.

[0076] In step S120 of some embodiments, the business operation lead is input into the lead scoring model for scoring to obtain a target lead score. The target lead score is used to indicate the lead conversion probability that the business operation lead can be converted into a business order. The higher the target lead score, the higher the lead conversion probability that the business operation lead is converted into a business order, and the higher the lead quality of the business operation lead. If the lead conversion probability is 1, it means that the business operation lead can be converted into a business order; if the lead conversion probability is 0, it means that the business operation lead cannot be converted into a business order. The target lead score output by the lead scoring model directly affects the follow-up decision results of the business operation lead, so it is necessary to ensure the prediction accuracy of the lead scoring model for the lead score.

[0077] See also Figure 2 In some embodiments, the training process of the clue scoring model may include but is not limited to steps S210 to S240:

[0078] Step S210, obtaining sample business leads and sample lead scores;

[0079] Step S220, updating the model parameters of the first preset scoring model according to the sample business leads and the sample lead scores to obtain a first scoring sub-model;

[0080] Step S230, updating the model parameters of the second preset scoring model according to the sample business clues and the first scoring sub-model to obtain a second scoring sub-model;

[0081] Step S240: determining a clue scoring model according to the first scoring sub-model and the second scoring sub-model.

[0082] In step S210 of some embodiments, sample business leads and sample lead scores may be obtained from industry websites and lead databases. The sample business leads are business leads used for model training, and the sample lead scores are real lead labels of the sample business leads, which are used to indicate the lead conversion probability of converting the sample business leads into business orders.

[0083] In step S220 of some embodiments, sample business leads and sample lead scores are used as training samples, the first preset scoring model is trained on the training samples, and the model parameters of the first preset scoring model are updated to obtain a first scoring sub-model. Due to the digital transformation of business activities and multi-channel data collection methods, the business leads obtained every day are massive. In order to improve the processing efficiency of business leads, the first preset scoring model adopts a decision tree. In addition, the structure of the decision tree is intuitive, and the decision tree can be visualized as a tree diagram, which can accurately capture the features that have a significant impact on the probability of lead conversion, thereby improving the interpretability of the lead scoring model. At the same time, the decision tree predicts the lead score through a series of decision rules, which can accurately capture the nonlinear relationship between the sample business leads and the lead score, thereby improving the accuracy of the lead score.

[0084] In step S230 of some embodiments, a second preset scoring model is created, and the second preset scoring model may be a decision tree. Specifically, a preset number of features are randomly selected from the sample business leads to obtain a sample attribute set. The information gain of each feature in the sample attribute set is calculated, and the feature with the largest information gain is selected as the splitting point of the current node. A decision tree is constructed based on the splitting point to obtain the second preset scoring model. Both the first preset scoring model and the second preset scoring model are weak learners, and may be models of the same network architecture or models of different network architectures. The network architecture is a tree structure of a decision tree, such as branches and depth of a decision tree. Model training is performed on the second preset scoring model based on the sample business leads and the first scoring sub-model, and the model parameters of the second preset scoring model are updated to capture the complex nonlinear relationship of the sample business leads and the feature interaction information between the models using the second preset scoring model, so as to correct the clue scoring error of the first scoring sub-model, thereby gradually improving the prediction accuracy of the clue scoring, until the training is completed, and the second scoring sub-model is obtained.

[0085] In step S240 of some embodiments, the first scoring sub-model and the second scoring sub-model are subjected to model aggregation to obtain a clue scoring model. Model aggregation can combine the advantages of multiple models, enhance model diversity, and achieve better scoring prediction results than a single model, so that the clue scoring model can also have good scoring prediction performance when dealing with unknown samples, thereby improving the generalization of the clue scoring model. In addition, through the joint training of multiple models, it is possible to capture complex patterns and feature interactions in clues, so as to fully learn sample business clues and improve the scoring accuracy of the clue scoring model. Specifically, the model parameters of the first scoring sub-model and the model parameters of the second scoring sub-model can be added or weighted to obtain the clue scoring model.

[0086] Through the above steps S210 to S240, a clue scoring model can be obtained to evaluate the clue quality of business operation leads based on the clue scoring model, so as to selectively follow up on business operation leads according to the clue quality, rather than following up on massive business operation leads one by one, thereby improving the efficiency of clue following up.

[0087] See also Figure 3 In some embodiments, step S220 may include but is not limited to steps S310 to S330:

[0088] Step S310, scoring the sample business leads using the first preset scoring model to obtain a fourth predicted lead score;

[0089] Step S320, performing loss calculation according to the fourth prediction clue score and the sample clue score to obtain fourth loss data;

[0090] Step S330: Update the model parameters of the first preset scoring model according to the fourth loss data to obtain a first scoring sub-model.

[0091] In step S310 of some embodiments, the sample business lead is input into the first preset scoring model for lead scoring to obtain a fourth predicted lead score. The fourth predicted lead score is the lead score for the sample business lead predicted by the first preset scoring model.

[0092] In step S320 of some embodiments, the fourth prediction clue score is used as the predicted value, and the sample clue score is used as the true value, and is input into the loss function for loss calculation to measure the difference between the true value and the predicted value, thereby obtaining fourth loss data. The loss function may be a mean square error function, a cross entropy loss function, etc.

[0093] In step S330 of some embodiments, in order to improve the accuracy of clue score prediction, the first preset score model needs to be optimized using the loss data. The fourth loss data is minimized, the model parameters of the first preset score model are updated, and the first preset score model when the fourth loss data takes the minimum value is used as the first score sub-model.

[0094] Through the above steps S310 to S330, a first scoring sub-model with better clue scoring performance can be obtained, thereby improving the scoring prediction accuracy of the first scoring sub-model.

[0095] See also Figure 4 In some embodiments, step S230 may include but is not limited to steps S410 to S440:

[0096] Step S410, obtaining the model gradient of the first scoring sub-model to obtain model residual data;

[0097] Step S420, scoring the sample business leads using a second preset scoring model to obtain a first predicted lead score;

[0098] Step S430, performing loss calculation according to the model residual data and the first prediction clue score to obtain first loss data;

[0099] Step S440: Update the model parameters of the second preset scoring model according to the first loss data to obtain a second scoring sub-model.

[0100] In step S410 of some embodiments, for sample business leads that lack key features such as purchase history, the first scoring sub-model may not make accurate score predictions for the sample business leads. In order to determine the scoring error of the first scoring sub-model, the derivative of the loss function with respect to the model parameters of the first scoring sub-model is calculated to obtain the model gradient, and the model gradient is used as the model residual data.

[0101] In step S420 of some embodiments, during the training process of the second preset scoring model, the sample business lead is input into the second preset scoring model for lead scoring to obtain a first predicted lead score. The first predicted lead score is the lead score predicted by the second preset scoring model for the sample business lead.

[0102] In step S430 of some embodiments, in order to enable the second preset scoring model to focus on samples that the first scoring sub-model fails to predict well and learn complex patterns in the samples more effectively, the model residual data and the first prediction clue score are input into the loss function for loss calculation to obtain the first loss data. The loss function may be a mean square error function, a cross entropy function, etc. The loss function is the same as the loss function used in the training process of the first scoring sub-model.

[0103] In step S440 of some embodiments, the second preset scoring model is optimized according to the first loss data to reduce the scoring error of the first scoring sub-model and gradually improve the accuracy of the scoring prediction. The first loss data is minimized, the model parameters of the second preset scoring model are updated, and the second preset scoring model when the first loss data takes the minimum value is used as the second scoring sub-model.

[0104] In the above steps S410 to S440, the second scoring sub-model is trained based on the first scoring sub-model. The features learned by the previous model can be applied to the next model, and the correlation relationship between the features extracted by different models can be established to enhance the connection between the features, so that the clue scoring model can be well adapted to clues with complex data structures and improve the accuracy of clue scoring.

[0105] See also Figure 5In some embodiments, step S240 may include but is not limited to steps S510 to S530:

[0106] Step S510, obtaining model weight data of the second scoring sub-model;

[0107] Step S520, performing model weighting on the second scoring sub-model according to the model weight data to obtain a reference scoring sub-model;

[0108] Step S530: Perform model aggregation on the first scoring sub-model and the reference scoring sub-model to obtain a clue scoring model.

[0109] In step S510 of some embodiments, the model weight data is used to indicate the contribution of the second scoring sub-model to the clue scoring model. The model weight data can be preset or obtained by real-time calculation.

[0110] In step S520 of some embodiments, the model weight data is multiplied by the model parameters of the second scoring sub-model to obtain a reference scoring sub-model.

[0111] In step S530 of some embodiments, the model parameters of the first scoring sub-model and the model parameters of the reference scoring sub-model are added to obtain a clue scoring model. A strong learner can be constructed by combining multiple weak learners, thereby improving the accuracy of clue scoring.

[0112] Through the above steps S510 to S530, the model parameters of different scoring sub-models can be integrated to enhance the generalization of the clue scoring model and improve the scoring accuracy of business operation clues.

[0113] See also Figure 6 In some embodiments, step S510 may include but is not limited to steps S610 to S650:

[0114] Step S610, scoring the sample business leads using the first scoring sub-model to obtain a second predicted lead score;

[0115] Step S620, performing loss calculation according to the second prediction clue score and the sample clue score to obtain second loss data;

[0116] Step S630, scoring the sample business leads using the second scoring sub-model to obtain a third predicted lead score;

[0117] Step S640, determining the third loss data according to the second loss data, the third prediction clue score and the preset model weight;

[0118] Step S650, updating the preset model weights according to the third loss data to obtain model weight data.

[0119] In step S610 of some embodiments, the sample business lead is input into the first scoring sub-model for lead scoring to obtain a second predicted lead score.

[0120] In step S620 of some embodiments, the second prediction clue score and the sample clue score are input into a loss function for loss calculation to obtain second loss data. The loss function may be a mean square error function, a cross entropy function, etc. It should be noted that the loss function is the same as the loss function used to train the first scoring sub-model and the loss function used to train the second scoring sub-model.

[0121] In step S630 of some embodiments, the sample business lead is input into the second scoring sub-model for lead scoring to obtain a third predicted lead score.

[0122] In step S640 of some embodiments, the preset model weight and the third prediction clue score are multiplied to obtain an intermediate clue score, and the intermediate clue score and the second loss data are added to obtain the third loss data.

[0123] In step S650 of some embodiments, the preset model weight is updated by minimizing the third loss data, and the preset model weight when the third loss data takes the minimum value is used as the model weight data.

[0124] Through the above steps S610 to S650, the optimal model weight data can be obtained to accurately evaluate the contribution of the second scoring sub-model to the clue scoring model, so as to obtain a clue scoring model with better scoring performance, thereby improving the accuracy of clue scoring.

[0125] See also Figure 7 In some embodiments, step S530 may include but is not limited to steps S710 to S730:

[0126] Step S710, the first scoring sub-model and the reference scoring sub-model are aggregated to obtain a candidate scoring model;

[0127] Step S720, obtaining the model scoring accuracy of the candidate scoring model;

[0128] Step S730, screening the candidate scoring models according to the model scoring accuracy to obtain a clue scoring model.

[0129] In step S710 of some embodiments, the model parameters of the first scoring sub-model and the model parameters of the reference scoring sub-model are added to obtain a candidate scoring model.

[0130] In step S720 of some embodiments, business leads for testing the model performance of the candidate scoring model are obtained to obtain test business leads. The actual lead scores of the test business leads are obtained to obtain test lead scores. The test business leads are input into the candidate scoring model for lead scoring to obtain candidate prediction lead scores. The difference between the test lead score and the candidate prediction lead score is calculated to obtain a difference score. The difference score is the absolute value of the difference between the test lead score and the candidate prediction lead score. If the difference score is less than a preset scoring threshold, the candidate prediction lead score is considered to be a correct score. The number of leads of the correctly scored test business leads is obtained to obtain a first number. The number of leads of the test business leads is obtained to obtain a second number. The ratio of the first number to the second number is used as the model scoring accuracy.

[0131] In step S730 of some embodiments, if the model scoring accuracy is greater than or equal to a preset accuracy threshold, the candidate scoring model is used as the clue scoring model. If the model scoring accuracy is less than the preset accuracy threshold, the candidate scoring model is retrained.

[0132] In the above steps S710 to S730, the candidate scoring models are screened according to the model scoring accuracy, so as to obtain a clue scoring model with high scoring accuracy, so as to accurately evaluate the quality of the business operation clues.

[0133] In step S130 of some embodiments, the business operation leads are sorted from large to small according to the target lead scores to obtain a lead sorting order. The lead sorting order is used to indicate the sorting order of the business operation leads in the lead sequence. The smaller the lead sorting order, the higher the lead quality of the business operation lead.

[0134] In step S140 of some embodiments, if the lead ranking order is less than or equal to the preset order threshold, it means that the lead quality of the business operation lead is high, and the business operation lead is regarded as a qualified lead. If the lead ranking order is greater than the preset order threshold, it means that the lead quality of the business operation lead is low, and the business operation lead is regarded as an unqualified lead.

[0135] In step S150 of some embodiments, qualified leads are followed up in sequence according to the lead ranking order, without having to follow up all leads, which greatly improves the efficiency of lead following up. For example, if the lead ranking order is 2, the qualified lead is taken as the second lead to be followed up.

[0136] In step S160 of some embodiments, the lead follow-up intention of the unqualified lead is obtained, and the lead follow-up intention is used to indicate whether to follow up the unqualified lead or not to follow up the unqualified lead. If the lead follow-up intention is to follow up the unqualified lead, the unqualified leads are followed up in sequence according to the lead ranking order of the unqualified leads. If the lead follow-up intention is not to follow up the unqualified lead, the unqualified lead is put into the lead pool for cultivation.

[0137] After following up on the leads, the lead follow-up voice can be obtained and converted into text to obtain the lead follow-up text. Keyword detection is performed on the lead follow-up text to obtain the keywords and the frequency of the keywords. The lead follow-up text is scored according to the keywords and frequency to obtain the follow-up score. A lead follow-up strategy is generated according to the follow-up score, keywords and frequency to improve the business capabilities of the lead follow-up object.

[0138] The lead follow-up method of the embodiment of the present application includes: obtaining sample business leads and sample lead scores. Inputting the sample business leads into a first preset scoring model for lead scoring to obtain an initial predicted lead score. Calculating losses based on the initial predicted lead score and the sample lead score to obtain initial loss data. Minimizing the initial loss data, adjusting the model parameters of the first preset scoring model, and obtaining a first scoring sub-model. Calculating the model gradient of the first scoring sub-model to obtain model residual data. Inputting the sample business leads into a second preset scoring model for lead scoring to obtain a first predicted lead score. Calculating losses based on the model residual data and the first predicted lead score to obtain first loss data. Minimizing the first loss data, adjusting the model parameters of the second preset scoring model, and obtaining a second scoring sub-model. Using the first scoring sub-model, scoring the sample business leads, and obtaining a second predicted lead score. Calculating losses based on the second predicted lead score and the sample lead score to obtain second loss data. Using the second scoring sub-model, scoring the sample business leads, and obtaining a third predicted lead score. Multiplying the third predicted lead score and the preset model weight, adding the multiplication result to the second loss data, and determining the third loss data. Minimize the third loss data to update the preset model weights to obtain the model weight data of the second scoring sub-model. Multiply the model weight data and the model parameters of the second scoring sub-model to obtain the reference scoring sub-model. Add the model parameters of the first scoring sub-model and the model parameters of the reference scoring sub-model to obtain the clue scoring model.

[0139] Obtain business operation leads and a lead scoring model, score the business operation leads through the lead scoring model to obtain a target lead score, sort the business operation leads according to the target lead score to obtain a lead sorting order, classify the business operation leads according to the lead sorting order to obtain qualified leads and unqualified leads, follow up on qualified leads according to the lead sorting order, obtain the lead follow-up intention of unqualified leads, and follow up on unqualified leads according to the lead follow-up intention.

[0140] See also Figure 8 The embodiment of the present application also provides a clue following device, which can implement the above-mentioned clue following method, and the clue following device includes:

[0141] An acquisition module 810 is used to acquire business operation leads and lead scoring models;

[0142] Scoring module 820, used to score the business operation leads through the lead scoring model to obtain the target lead score;

[0143] A sorting module 830 is used to sort the business operation leads according to the target lead scores to obtain a lead sorting order;

[0144] A classification module 840 is used to classify business operation leads according to the lead sorting order to obtain qualified leads and unqualified leads;

[0145] A first follow-up module 850, for following up on qualified leads according to the lead sorting order;

[0146] The second follow-up module 860 is used to obtain the lead follow-up intention of the unqualified lead and follow up the unqualified lead according to the lead follow-up intention.

[0147] The specific implementation of the clue following device is basically the same as the specific implementation of the clue following method described above, and will not be described in detail here.

[0148] The embodiment of the present application also provides an electronic device, the electronic device includes a memory and a processor, the memory stores a computer program, and the processor implements the above-mentioned clue tracking method when executing the computer program. The electronic device can be any intelligent terminal including a tablet computer, a car computer, etc.

[0149] See also Fig. 9 , Fig. 9 The hardware structure of an electronic device of another embodiment is illustrated, and the electronic device includes:

[0150] The processor 910 may be implemented by a general-purpose central processing unit (CPU), a microprocessor, an application-specific integrated circuit (ASIC), or one or more integrated circuits, and is used to execute relevant programs to implement the technical solutions provided in the embodiments of the present application.

[0151] The memory 920 can be implemented in the form of a read-only memory (ROM), a static storage device, a dynamic storage device, or a random access memory (RAM). The memory 920 can store an operating system and other application programs. When the technical solution provided in the embodiment of this specification is implemented by software or firmware, the relevant program code is stored in the memory 920, and the processor 910 calls and executes the clue following method of the embodiment of this application;

[0152] Input / output interface 930, used to implement information input and output;

[0153] Communication interface 940, used to realize communication interaction between the device and other devices, which can be realized through wired mode (such as USB, network cable, etc.) or wireless mode (such as mobile network, WI FI, Bluetooth, etc.);

[0154] bus 950 , which transmits information between the various components of the device (e.g., processor 910 , memory 920 , input / output interface 930 , and communication interface 940 );

[0155] The processor 910 , the memory 920 , the input / output interface 930 , and the communication interface 940 are connected to each other in communication within the device via a bus 950 .

[0156] An embodiment of the present application also provides a computer-readable storage medium, which stores a computer program. When the computer program is executed by a processor, the above-mentioned clue following method is implemented.

[0157] The memory, as a non-transient computer-readable storage medium, can be used to store non-transient software programs and non-transient computer executable programs. In addition, the memory may include a high-speed random access memory, and may also include a non-transient memory, such as at least one disk storage device, a flash memory device, or other non-transient solid-state storage device. In some embodiments, the memory may optionally include a memory remotely disposed relative to the processor, and these remote memories may be connected to the processor via a network. Examples of the above-mentioned network include, but are not limited to, the Internet, an intranet, a local area network, a mobile communication network, and combinations thereof.

[0158] The lead following method, lead following device, electronic device and computer storage medium provided in the embodiments of the present application classify business operation leads according to the lead sorting order to evaluate the quality of the leads and obtain qualified leads and unqualified leads. Following up on qualified leads according to the lead sorting order can timely follow up on business operation leads with higher priority, reducing the time cost spent on sorting out each business operation lead and improving the efficiency of lead following up. Obtaining the lead following up intention of unqualified leads, and following up on unqualified leads according to the lead following up intention, without the need to sort out the leads one by one and follow up on any leads, improves the efficiency of lead following up.

[0159] The embodiments described in the embodiments of the present application are intended to more clearly illustrate the technical solutions of the embodiments of the present application and do not constitute a limitation on the technical solutions provided in the embodiments of the present application. Those skilled in the art will appreciate that with the evolution of technology and the emergence of new application scenarios, the technical solutions provided in the embodiments of the present application are also applicable to similar technical problems.

[0160] Those skilled in the art will appreciate that the technical solutions shown in the figures do not constitute a limitation on the embodiments of the present application, and may include more or fewer steps than shown in the figures, or a combination of certain steps, or different steps.

[0161] The device embodiments described above are merely illustrative, and the units described as separate components may or may not be physically separated, that is, they may be located in one place or distributed on multiple network units. Some or all of the modules may be selected according to actual needs to achieve the purpose of the solution of this embodiment.

[0162] Those skilled in the art will appreciate that all or some of the steps in the methods disclosed above, and the functional modules / units in the systems and devices may be implemented as software, firmware, hardware, or a suitable combination thereof.

[0163] The terms "first", "second", "third", "fourth", etc. (if any) in the specification of the present application and the above-mentioned drawings are used to distinguish similar objects, and are not necessarily used to describe a specific order or sequence. It should be understood that the data used in this way can be interchangeable where appropriate, so that the embodiments of the present application described herein can be implemented in an order other than those illustrated or described herein. In addition, the terms "including" and "having" and any of their variations are intended to cover non-exclusive inclusions, for example, a process, method, system, product or device comprising a series of steps or units is not necessarily limited to those steps or units clearly listed, but may include other steps or units that are not clearly listed or inherent to these processes, methods, products or devices.

[0164] It should be understood that in the present application, "at least one (item)" means one or more, and "plurality" means two or more. "And / or" is used to describe the association relationship of associated objects, indicating that three relationships may exist. For example, "A and / or B" can mean: only A exists, only B exists, and A and B exist at the same time, where A and B can be singular or plural. The character " / " generally indicates that the objects associated before and after are in an "or" relationship. "At least one of the following" or similar expressions refers to any combination of these items, including any combination of single or plural items. For example, at least one of a, b or c can mean: a, b, c, "a and b", "a and c", "b and c", or "a and b and c", where a, b, c can be single or multiple.

[0165] In the several embodiments provided in the present application, it should be understood that the disclosed devices and methods can be implemented in other ways. For example, the device embodiments described above are only schematic. For example, the division of the above units is only a logical function division. There may be other division methods in actual implementation, such as multiple units or components can be combined or integrated into another system, or some features can be ignored or not executed. Another point is that the mutual coupling or direct coupling or communication connection shown or discussed can be through some interfaces, indirect coupling or communication connection of devices or units, which can be electrical, mechanical or other forms.

[0166] The units described above as separate components may or may not be physically separated, and the components shown as units may or may not be physical units, that is, they may be located in one place or distributed on multiple network units. Some or all of the units may be selected according to actual needs to achieve the purpose of the solution of this embodiment.

[0167] In addition, each functional unit in each embodiment of the present application may be integrated into one processing unit, or each unit may exist physically separately, or two or more units may be integrated into one unit. The above-mentioned integrated unit may be implemented in the form of hardware or in the form of software functional units.

[0168] If the integrated unit is implemented in the form of a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present application is essentially or the part that contributes to the prior art or all or part of the technical solution can be embodied in the form of a software product, and the computer software product is stored in a storage medium, including multiple instructions to enable a computer device (which can be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods of various embodiments of the present application. The aforementioned storage medium includes: U disk, mobile hard disk, read-only memory (Read-Only Memory, referred to as ROM), random access memory (Random Access Memory, referred to as RAM), disk or optical disk and other media that can store programs.

[0169] The preferred embodiments of the present application are described above with reference to the accompanying drawings, but the scope of the rights of the present application is not limited thereto. Any modification, equivalent substitution and improvement made by a person skilled in the art without departing from the scope and essence of the present application should be within the scope of the rights of the present application.

Claims

1. A clue following method, characterized in that: The method comprises: Obtain business operation leads and lead scoring models; Scoring the business operation leads by using the lead scoring model to obtain a target lead score; Sorting the business operation leads according to the target lead scores to obtain a lead sorting order; Classifying the business operation leads according to the lead sorting order to obtain qualified leads and unqualified leads; Following up on the qualified leads according to the lead ranking order; The lead follow-up intention of the unqualified lead is obtained, and the unqualified lead is followed up according to the lead follow-up intention.

2. The method according to claim 1, characterized in that: The clue scoring model is trained according to the following steps: Obtain sample business leads and sample lead scores; The model parameters of the first preset scoring model are updated according to the sample business leads and the sample lead scores to obtain a first scoring sub-model; The model parameters of the second preset scoring model are updated according to the sample business clues and the first scoring sub-model to obtain a second scoring sub-model; The clue scoring model is determined according to the first scoring sub-model and the second scoring sub-model.

3. The method according to claim 2, characterized in that The updating of the model parameters of the second preset scoring model according to the sample business clues and the first scoring sub-model to obtain the second scoring sub-model includes: Obtaining a model gradient of the first scoring sub-model to obtain model residual data; Performing a lead score on the sample business lead using the second preset scoring model to obtain a first predicted lead score; Perform loss calculation according to the model residual data and the first prediction clue score to obtain first loss data; The model parameters of the second preset scoring model are updated according to the first loss data to obtain the second scoring sub-model.

4. The method according to claim 2, characterized in that: The determining the clue scoring model according to the first scoring sub-model and the second scoring sub-model includes: Obtaining model weight data of the second scoring sub-model; Performing model weighting on the second scoring sub-model according to the model weight data to obtain a reference scoring sub-model; The first scoring sub-model and the reference scoring sub-model are subjected to model aggregation to obtain the clue scoring model.

5. The method according to claim 4, characterized in that The obtaining of the model weight data of the second scoring sub-model includes: Performing a clue score on the sample business clue by using the first scoring sub-model to obtain a second predicted clue score; Calculate the loss according to the second prediction clue score and the sample clue score to obtain second loss data; Performing a clue score on the sample business clue by using the second scoring sub-model to obtain a third predicted clue score; Determining third loss data according to the second loss data, the third prediction clue score and a preset model weight; The preset model weight is updated according to the third loss data to obtain the model weight data.

6. The method according to claim 4, characterized in that The performing model aggregation on the first scoring sub-model and the reference scoring sub-model to obtain the clue scoring model includes: The first scoring sub-model and the reference scoring sub-model perform model aggregation to obtain a candidate scoring model; Obtaining the model scoring accuracy of the candidate scoring model; The candidate scoring models are screened according to the model scoring accuracy to obtain the clue scoring model.

7. The method according to claim 2, characterized in that The updating of the model parameters of the first preset scoring model according to the sample business leads and the sample lead scores to obtain a first scoring sub-model includes: Performing a clue score on the sample business clue by using the first preset scoring model to obtain a fourth predicted clue score; Calculate the loss according to the fourth prediction clue score and the sample clue score to obtain fourth loss data; The model parameters of the first preset scoring model are updated according to the fourth loss data to obtain the first scoring sub-model.

8. A clue following device, characterized in that: The device comprises: The acquisition module is used to obtain business operation leads and lead scoring models; A scoring module, used to score the business operation leads by using the lead scoring model to obtain a target lead score; A sorting module, used to sort the business operation leads according to the target lead scores to obtain a lead sorting order; A classification module, used to classify the business operation leads according to the lead sorting order to obtain qualified leads and unqualified leads; A first follow-up module, used for following up the qualified leads according to the lead sorting order; The second follow-up module is used to obtain the lead follow-up intention of the unqualified lead and follow up the unqualified lead according to the lead follow-up intention.

9. An electronic device, characterized in that: The electronic device comprises a memory and a processor, the memory stores a computer program, and the processor implements the method according to any one of claims 1 to 7 when executing the computer program.

10. A computer-readable storage medium storing a computer program, characterized in that: When the computer program is executed by a processor, the method according to any one of claims 1 to 7 is implemented.

Images