Method and device for training prediction model and predicting client cooperation intention
Through semi-supervised learning technology training prediction model, the problems of low manual screening efficiency and high false positive rate in supply chain finance business are solved, accurate prediction and dynamic optimization of customer cooperation intentions are achieved, and customer expansion efficiency and model accuracy are improved.
Patent Information
- Application Number
- CN202310664259.5
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2023-06-06
- Publication Date
- 2024-12-06
AI Technical Summary
The existing supply chain financial business relies on manual experience screening in the process of customer expansion, resulting in large amount of customer screening and low efficiency. The false positive rate of traditional data mining methods is high, so it is impossible to allocate sales resources in a refined manner, affecting the efficiency of customer expansion.
The prediction model is trained using semi-supervised learning technology. By obtaining customer characteristics, calculating cooperation possibility scores, sorting and selecting target customers, updating the model based on feedback information, realizing dynamic mining and iterative optimization.
It improves customer expansion efficiency and accuracy, reduces false positive rates, supports the goal of business opportunity mining, and improves prediction accuracy through automated iterative models.
Smart Images

Figure CN119090366A_ABST
Abstract
Description
Technical Field
[0001] Embodiments of the present disclosure relate to the field of computer technology, and in particular to a method and device for training a prediction model. Background Art
[0002] Supply chain finance is currently an important financing channel for small and medium-sized enterprises. The current financial business models include movable property financing, freight factoring, and data lending. Movable property financing business refers to the credit business in which an enterprise uses movable property or property rights that it owns or that are legally owned by a third party as collateral or a bank supervises the movable property or property rights of an enterprise.
[0003] At present, when expanding business, the supply chain finance business group tends to directly screen customer groups using access conditions, that is, to set hard conditions for qualified customer groups based on expert experience to find the access customer groups. Then, sales resources are allocated to the access customer groups one by one by manually screening them according to experience. After a period of time, if the logistics customer list is updated, it needs to be repeated in this way.
[0004] In addition to the customer acquisition method that relies entirely on experience, some data mining technologies have also been applied in the field of business opportunity mining. In the field of customer acquisition, data usually only has one-sided labels, that is, "which customers have cooperated with us", but lacks the other-side labels, that is, "which customers will never cooperate with us". Such data structure is similar to the field of outlier detection. The application of some unsupervised and semi-supervised methods such as KNN (K-NearestNeighbor, nearest neighbor), isolation forest, one-class SVM (single-class support vector machine) and other technologies can play a role in similar situations, but due to the influence of methods and data, these methods have an obvious problem that they have a high recall rate, that is, problematic cases will be selected with a high probability, but at the same time have a high false positive rate, that is, a large proportion of the selected cases actually have no problems.
[0005] The customer groups screened out based on manual experience and entry criteria are usually large. For the sales team, it is impossible to assess the potential for cooperation with the merchants, and there is no way to further refine the allocation of sales resources. If you try to reach customers, it is like looking for a needle in a haystack, which has a certain impact on the efficiency of customer acquisition.
[0006] Data mining-based technologies cannot iterate the model based on the business side’s reach, so the model cannot actively iterate its own accuracy. At the same time, the data mining method applicable to this scenario is limited by the problem of too high false positive rate, and tends to provide more lists, a large proportion of which are not the target customers of precise customer development, which also has a certain impact on the efficiency of customer development. Summary of the invention
[0007] The embodiments of the present disclosure provide methods and devices for training prediction models and predicting customer cooperation intentions.
[0008] In a first aspect, an embodiment of the present disclosure provides a method for training a prediction model, comprising: obtaining features of at least one customer to be cooperated with, wherein the customer includes customers who have cooperated with and customers who have not cooperated with; inputting the features of each customer into the prediction model to obtain a cooperation possibility score for each customer; determining at least one target customer for possible cooperation based on the cooperation possibility score; sorting the target customers in order of cooperation possibility score from high to low, and selecting a portion of first target customers with the highest cooperation possibility score and a portion of second target customers with the lowest cooperation possibility score according to a predetermined ratio; sending a cooperation invitation to the first target customer and the second target customer, and receiving cooperation feedback information from the first target customer and the second target customer; updating customers who have cooperated with and customers who have not cooperated with according to the cooperation feedback information; and retraining the prediction model based on the updated features of each customer.
[0009] In some embodiments, the obtaining of the characteristics of at least one customer to be cooperated with includes: obtaining the access conditions for cooperation; obtaining the latest customer information and a list of customers who have cooperated; filtering out customer information that does not meet the access conditions; and extracting the characteristics of each customer according to predetermined fields from the retained customer information.
[0010] In some embodiments, the method further includes: if there is no predetermined field in the customer information, performing semantic analysis on the customer information to obtain features that are synonymous with the predetermined field.
[0011] In some embodiments, the prediction model includes at least one of the following: a single-class support vector machine, a nearest neighbor knot algorithm model, and an isolation forest.
[0012] In some embodiments, the prediction model is a single-class support vector machine; and the determining of at least one target customer for possible cooperation based on the cooperation possibility score includes: mapping the customer's feature points to a high-dimensional space based on a kernel function according to the atypical proportion of customers who have cooperated, and training a hyperplane; projecting the customer's feature points on the hyperplane to obtain a double teardrop-shaped area; and determining the customer whose feature points are within the lower teardrop-shaped area as a target customer for possible cooperation.
[0013] In a second aspect, an embodiment of the present disclosure provides a method for predicting customer cooperation intention, comprising: obtaining characteristics of the customer to be predicted; inputting the characteristics into a prediction model trained according to the method described in any one of the first aspects, and outputting the cooperation possibility score of the customer.
[0014] In a third aspect, an embodiment of the present disclosure provides a device for training a prediction model, comprising: an acquisition unit, configured to acquire features of at least one customer to be cooperated with, wherein the customer includes customers who have cooperated with and customers who have not cooperated with; a prediction unit, configured to input the features of each customer into the prediction model to obtain a cooperation possibility score for each customer; a determination unit, configured to determine at least one target customer for possible cooperation based on the cooperation possibility score; a selection unit, configured to sort the target customers in descending order of the cooperation possibility score, and select a part of the first target customers with the highest cooperation possibility score and a part of the second target customers with the lowest cooperation possibility score according to a predetermined ratio; an invitation unit, configured to send a cooperation invitation to the first target customer and the second target customer, and receive cooperation feedback information from the first target customer and the second target customer; an update unit, configured to update customers who have cooperated with and customers who have not cooperated with according to the cooperation feedback information; and a retraining unit, configured to retrain the prediction model based on the updated features of each customer.
[0015] In some embodiments, the acquisition unit is further configured to: acquire access conditions for cooperation; acquire the latest customer information and a list of customers who have cooperated; filter out customer information that does not meet the access conditions; and extract features of each customer from the retained customer information according to predetermined fields.
[0016] In some embodiments, the acquisition unit is further configured to: if there is no predetermined field in the customer information, perform semantic analysis on the customer information to obtain features that are synonymous with the predetermined field.
[0017] In some embodiments, the prediction model includes at least one of the following: a single-class support vector machine, a nearest neighbor knot algorithm model, and an isolation forest.
[0018] In some embodiments, the prediction model is a single-class support vector machine; and the determination unit is further configured to: map the customer's feature points to a high-dimensional space based on a kernel function according to the atypical proportion of the customers who have cooperated, and train a hyperplane; project the customer's feature points on the hyperplane to obtain a double teardrop-shaped area; and determine the customer whose feature points are within the lower teardrop-shaped area as a target customer for possible cooperation.
[0019] In a fourth aspect, an embodiment of the present disclosure provides a device for predicting customer cooperation intention, comprising: an acquisition unit, configured to acquire features of a customer to be predicted; a prediction unit, configured to input the features into a prediction model trained according to the method described in any one of the first aspects, and output a cooperation possibility score for the customer.
[0020] In a fifth aspect, an embodiment of the present disclosure provides an electronic device, comprising: one or more processors; a storage device on which one or more computer programs are stored, and when the one or more computer programs are executed by the one or more processors, the one or more processors implement a method as described in any one of the first and second aspects.
[0021] In a sixth aspect, an embodiment of the present disclosure provides a computer-readable medium having a computer program stored thereon, wherein when the computer program is executed by a processor, the method as described in any one of the first aspect and the second aspect is implemented.
[0022] The method and device for training a prediction model provided by the embodiments of the present disclosure realize dynamic mining of target customers through productization by using semi-supervised active learning technology, and automatically iterate the data mining model according to business contact feedback (cooperation or non-cooperation), so as to solve the problems of manual customer development relying on experience and low efficiency, and the inability of traditional data mining to improve model results according to feedback in a timely manner, thereby better supporting the achievement of the goal of mining business opportunities. BRIEF DESCRIPTION OF THE DRAWINGS
[0023] Other features, objects and advantages of the present disclosure will become more apparent from the detailed description of non-limiting embodiments made with reference to the following drawings:
[0024] Figure 1 is an exemplary system architecture diagram in which the present disclosure may be applied;
[0025] Figure 2 is a flowchart of an embodiment of a method for training a prediction model according to the present disclosure;
[0026] Figure 3a , 3b is a schematic diagram of an application scenario of the method for training a prediction model according to the present disclosure;
[0027] Figure 4 is a flow chart of an embodiment of a method for predicting customer cooperation intention according to the present disclosure;
[0028] Figure 5 is a structural schematic diagram of an embodiment of an apparatus for training a prediction model according to the present disclosure;
[0029] Figure 6 is a structural schematic diagram of an embodiment of a device for predicting customer cooperation intention according to the present disclosure;
[0030] Figure 7 It is a schematic diagram of the structure of a computer system suitable for implementing the electronic device of the embodiment of the present disclosure. DETAILED DESCRIPTION
[0031] The present disclosure is further described in detail below in conjunction with the accompanying drawings and embodiments. It is understood that the specific embodiments described herein are only used to explain the relevant invention, rather than to limit the invention. It is also necessary to explain that, for ease of description, only the parts related to the relevant invention are shown in the accompanying drawings.
[0032] It should be noted that, in the absence of conflict, the embodiments and features in the embodiments of the present disclosure may be combined with each other. The present disclosure will be described in detail below with reference to the accompanying drawings and in combination with the embodiments.
[0033] Figure 1 An exemplary system architecture 100 is shown to which the method for training a prediction model, the apparatus for training a prediction model, the method for predicting customer cooperation intention, or the apparatus for predicting customer cooperation intention according to the embodiments of the present disclosure can be applied.
[0034] like Figure 1 As shown, the system architecture 100 may include terminals 101, 102, a network 103, a database server 104, and a server 105. The network 103 is used to provide a medium for communication links between the terminals 101, 102, the database server 104, and the server 105. The network 103 may include various connection types, such as wired, wireless communication links, or optical fiber cables, etc.
[0035] The user 110 can use the terminal 101, 102 to interact with the server 105 through the network 103 to receive or send messages, etc. Various client applications can be installed on the terminal 101, 102, such as model training applications, customer cooperation applications, shopping applications, payment applications, web browsers, and instant messaging tools.
[0036] The terminals 101 and 102 here can be hardware or software. When the terminals 101 and 102 are hardware, they can be various electronic devices with display screens, including but not limited to smart phones, tablet computers, e-book readers, MP3 players (Moving Picture Experts Group Audio Layer III, Dynamic Image Experts Compression Standard Audio Layer 3), laptops and desktop computers, etc. When the terminals 101 and 102 are software, they can be installed in the electronic devices listed above. It can be implemented as multiple software or software modules (for example, to provide distributed services), or it can be implemented as a single software or software module. No specific limitation is made here.
[0037] The database server 104 may be a database server that provides various services. For example, a customer feature set may be stored in the database server. Each customer feature may include a customer feature and a label of whether the customer has cooperated before. In this way, the user 110 may also select a customer feature from the customer feature set stored in the database server 104 through the terminals 101 and 102.
[0038] The server 105 may also be a server that provides various services, such as a background server that supports various applications displayed on the terminals 101 and 102. The background server may train the initial model using the customer features sent by the terminals 101 and 102, and may send the training results (such as the generated prediction model) to the terminals 101 and 102. In this way, the user may use the generated prediction model to predict the cooperation intention.
[0039] The database server 104 and server 105 here can also be hardware or software. When they are hardware, they can be implemented as a distributed server cluster consisting of multiple servers, or as a single server. When they are software, they can be implemented as multiple software or software modules (for example, used to provide distributed services), or as a single software or software module. No specific limitation is made here. The database server 104 and server 105 can also be servers of a distributed system, or servers combined with blockchain. The database server 104 and server 105 can also be cloud servers, or intelligent cloud computing servers or intelligent cloud hosts with artificial intelligence technology.
[0040] It should be noted that the method for training a prediction model or the method for predicting customer cooperation intention provided in the embodiments of the present disclosure is generally executed by the server 105. Accordingly, the device for training a prediction model or the device for predicting customer cooperation intention is generally also set in the server 105.
[0041] It should be pointed out that, in the case where the server 105 can implement the related functions of the database server 104 , the database server 104 may not be provided in the system architecture 100 .
[0042] It should be understood that Figure 1 The number of terminals, networks, database servers and servers in the embodiment is only for illustration. Any number of terminals, networks, database servers and servers may be provided according to the implementation requirements.
[0043] Continue to see Figure 2 , which shows a process 200 of an embodiment of a method for training a prediction model according to the present disclosure. The method for training a prediction model may include the following steps:
[0044] Step 201: Acquire characteristics of at least one client to be cooperated with.
[0045] In this embodiment, the execution subject of the method for training the prediction model (for example Figure 1 The server 105 shown in FIG. 104 may obtain the characteristics of at least one client to be cooperated with in a variety of ways. For example, the execution subject may obtain the characteristics of at least one client to be cooperated with in a wired connection or a wireless connection from a database server (e.g. Figure 1 For example, a user can obtain the characteristics of existing customers stored in the database server 104 shown in FIG. Figure 1 The terminals 101 and 102 shown in the figure are used to collect the characteristics of the customers. In this way, the execution subject can receive the characteristics of the customers collected by the terminals and store these characteristics of the customers locally.
[0046] Among them, customers include customers who have cooperated with us and customers who have not cooperated with us. Customers who have cooperated with us can be used as positive samples of training data, that is, positive data. Customer characteristics may include information such as age, occupation, and income. The collection, collection, update, analysis, processing, use, transmission, and storage of this information are in compliance with the provisions of relevant laws and regulations, are used for legal purposes, and do not violate public order and good morals. Take necessary measures for user personal information to prevent illegal access to user personal information data and maintain the security of user personal information, network security, and national security.
[0047] Step 202, input the characteristics of each customer into the prediction model to obtain the cooperation possibility score of each customer.
[0048] In this embodiment, the prediction model can be a single classification model, such as a single-class support vector machine, a nearest neighbor knot algorithm model, an isolation forest, or a clustering model. The single classification model can output the probability of cooperation as a cooperation possibility score. The clustering model can obtain the cooperation possibility score by calculating the similarity between the characteristics of each customer and the cluster center of the customers who have cooperated.
[0049] Step 203: determine at least one target customer for possible cooperation according to the cooperation possibility score.
[0050] In this embodiment, a predetermined threshold may be set, and customers whose cooperation possibility scores are higher than the threshold may be selected as target customers.
[0051] Step 204, sort the target customers in descending order of cooperation possibility scores, and select a portion of first target customers with the highest cooperation possibility scores and a portion of second target customers with the lowest cooperation possibility scores according to a predetermined ratio.
[0052] In this embodiment, a portion of customers with the highest scores and a portion with the lowest scores are selected from potential cooperative customers and mixed together, wherein the characteristics of the portion of customers with the lowest scores can be used to learn boundary conditions for cooperation.
[0053] The number of the first target customers is greater than the number of the second target customers. The predetermined ratio may be, for example, 4:1.
[0054] Step 205: sending a cooperation invitation to the first target customer and the second target customer, and receiving cooperation feedback information from the first target customer and the second target customer.
[0055] In this embodiment, a cooperation invitation may be sent to the first target customer and the second target customer by phone, email, APP push, etc., i.e., a customer contact behavior. The first target customer and the second target customer may provide feedback, and the feedback information may include cooperation or non-cooperation. The first target customer and the second target customer may not provide feedback, and they are still marked as non-cooperation.
[0056] Step 206: Update the customers who have cooperated and the customers who have not cooperated according to the cooperation feedback information.
[0057] In this embodiment, if a cooperation message is received from a customer who has not cooperated before, it is modified to a customer who has cooperated before, that is, the number of positive samples with labels in the training data is increased.
[0058] Step 207 , retraining the prediction model based on the updated features of each customer.
[0059] In this embodiment, in active learning, it is necessary to give priority to marking those cases that may cause false positives, that is, cases that are judged as positive by the model but are actually negative. The discovery of these cases has the most obvious effect on improving the model. This is somewhat contradictory to the business goal of reaching those customers whose model infers the highest tendency to cooperate, because the possibility of false positives for these customers is not high. Therefore, when feeding back a list of potential customers to the business, the cooperation possibility scores can be arranged from large to small (the higher the score, the more likely it is to be a potential customer), and a part of the customers are selected from the top and bottom of the list of potential customers, and then mixed and fed back to the business side.
[0060] As shown in the following table, the output list can be a mixture of 80% of the customers with the highest scores among those determined as potential target customers and 20% of the customers with lower scores among those also determined as potential target customers. For example, in the following table, G Printer Flagship Store, J Sports Franchise Store, M Outdoor Franchise Store, and F Sports Franchise Store are the four stores with the highest scores, and are the merchants that the model determines are most likely to cooperate with the business. X Underwear Official Flagship Store is the merchant with the lowest score among those that the model also determines are likely to cooperate with the business. Such a mixture of customers with different possibilities may not maximize the efficiency of business customer acquisition in the short term, but in the long run, due to the active help of the model to iterate, the model will better provide support for business customer acquisition after learning the hyperplane more accurately.
[0061]
[0062] Table 1 Feedback table of potential customers mined by prediction model
[0063] After the business reaches out to the customer, it records the behavior results in the system, that is, whether the customer chooses to cooperate with us. The model can schedule tasks, such as daily, or manually trigger retraining in the form of services. New data experience will enter the model. Since data points with low cooperation possibility scores are near the boundaries of what the model has learned, the feedback from these data points can better help the model relearn.
[0064] The above-mentioned method of training the prediction model realizes dynamic mining of target customers through productization by using semi-supervised active learning technology, and automatically iterates the data mining model according to business contact feedback (cooperation or non-cooperation), so as to solve the problems of manual customer development relying on experience and low efficiency, and the inability of traditional data mining to improve model results according to feedback in a timely manner, so as to better support the achievement of business opportunity mining goals.
[0065] After new engagement behavior data enters the system, the system can use scheduled tasks or manual triggering to retrain the model. The retrained model results will reduce the incidence of false positives and provide the business with more accurate customer acquisition prediction services.
[0066] In some optional implementations of this embodiment, the obtaining of characteristics of at least one customer to be cooperated with includes: obtaining access conditions for cooperation; obtaining the latest customer information and a list of customers who have cooperated; filtering out customer information that does not meet the access conditions; and extracting characteristics of each customer according to predetermined fields from the retained customer information.
[0067] The entry conditions refer to what kind of customers meet the minimum requirements of the capital side, including but not limited to the customer's industry, the customer's main business, whether the customer is a warehousing and distribution user, the length of time the customer has cooperated with the logistics company, etc.
[0068] The list of cooperated customers refers to the list of cooperated customers provided by the business, or at least an initial list of customers that they think may cooperate with them as the earliest positive sample labels input into the system.
[0069] Features refer to the range of feature fields for data modeling. Businesses need to provide which features they want the model to use to determine whether a customer can cooperate with them, for example, age, occupation, income, etc.
[0070] In some optional implementations of this embodiment, the method further includes: if there is no predetermined field in the customer information, semantic analysis is performed on the customer information to obtain features that are synonymous with the predetermined field. For example, if there is no age information in the customer information, semantic analysis can be performed based on occupation and other information to determine the age range of the customer. For example, if the customer is an undergraduate student, the customer's age can be determined to be between 18 and 23, and if the customer is retired, the age can be determined to be over 50. This can avoid the situation where the customer's features are incomplete and cannot be used as training data, and the accuracy of the model can be improved by expanding the amount of training data.
[0071] In some optional implementations of this embodiment, the prediction model includes at least one of the following: a one-class support vector machine (one-class SVM), a nearest neighbor node algorithm model (KNN), and an isolation forest. For customer information with only one category label, a single-class prediction model is used. If uncooperative feedback information is received from customers, and the number of uncooperative customers is large enough, a two-class model can also be used as a prediction model. However, before a sufficient number of uncooperative customers are accumulated, a single-class model needs to be used as a prediction model.
[0072] In some optional implementations of the present embodiment, the prediction model is a single-class support vector machine; and the method of determining at least one target customer for possible cooperation based on the cooperation possibility score includes: mapping the customer's feature points to a high-dimensional space based on a kernel function according to the atypical proportion of customers who have cooperated, and training a hyperplane; projecting the customer's feature points on the hyperplane to obtain a double teardrop-shaped area; and determining the customer whose feature points are within the lower teardrop-shaped area as a target customer for possible cooperation.
[0073] One-class SVM is a semi-supervised learning method for data with only one label. It maps the data to a high-dimensional space through a kernel function, obtains a hyperplane through training, and places as many points of a given proportion as possible on one side of the hyperplane, while the rest of the points are on the other side.
[0074] The feature data and positive sample labels are input into one-class SVM training, and the possible proportion of outliers in the data marked with positive sample labels (which customers cooperating with the business may be atypical) are input. The model will map the data points to a high-dimensional space based on a kernel function (such as the radial basis function, RBF, in this case), and then train a hyperplane to keep the positive data points on one side as much as possible.
[0075] Figure 3a , 3b In the figure, it is shown that in actual use, two characteristic principal components are used as the horizontal axis (representing income) and the vertical axis (representing age), and data points with positive labels (hollow dots) and unlabeled data points (solid dots) are marked on the graph at the same time. The kernel function used by the model maps the 2D data space to the 3D space, and its projection on the hyperplane is the double teardrop-shaped area on the graph. The hyperplane trained by the model divides the 3D space into two parts: the inside of the teardrop-shaped area (the area of the lower teardrop-shaped area) and the outside of the teardrop-shaped area (the upper teardrop and other areas). The lower half of the teardrop-shaped area is the inferred positive data point, which includes both customers with positive labels and customers that the model determines are likely to cooperate (inferred positive). Each data point can obtain an SVM score, and the larger the score, the higher the probability that the data point is positive. In the area of the lower teardrop-shaped area, the first target customer with the highest possibility score and the second target customer with the lowest cooperation possibility score are selected.
[0076] Through one-class SVM, the potential customer base can be greatly reduced from the original full number of admitted customers. In the actual application case, we found 3190 customers from all 4862 admitted customers, eliminating about 35% of the potential customers who are unlikely to cooperate but meet the admission criteria.
[0077] After the business reaches out to the customer, it records the behavior results in the system, that is, whether or not they choose to cooperate with us. The model can schedule tasks, such as daily dimensions, or manually trigger retraining in the form of services. New data experience will enter the model. Since the data points with low SVM scores are near the boundaries learned by the model, the feedback from these data points can better help the model relearn the hyperplane, such as Figure 3b shown.
[0078] See also Figure 4 , which shows a process 400 of an embodiment of a method for predicting customer cooperation intention provided by the present disclosure. The method for predicting customer cooperation intention may include the following steps:
[0079] Step 401, obtaining the characteristics of the customer to be predicted.
[0080] In this embodiment, the execution subject of the method for predicting customer cooperation intention (for example Figure 1 The server 105 shown in FIG. 104 may obtain the characteristics of the customer to be predicted in a variety of ways. For example, the execution subject may obtain the characteristics of the customer to be predicted from a database server (e.g. Figure 1 For another example, the execution subject may also receive a terminal (such as a database server 104) to obtain the characteristics of the customer stored therein. Figure 1 The terminals 101, 102 shown in the figure) or other devices collect customer information, and then extract predetermined features from the customer information.
[0081] Step 402, input the customer's characteristics into the prediction model, and output the customer's cooperation possibility score.
[0082] In this embodiment, the execution entity can input the features obtained in step 401 into the prediction model to output the cooperation possibility score of the customer. The higher the score, the more likely the customer is to cooperate. The priority of these high-scoring customers can be set higher and processed first. Customers with scores below the threshold can be ignored and no longer contacted.
[0083] In this embodiment, the prediction model can be adopted as described above. Figure 2 The specific generation process can be found in Figure 2 The relevant description of the embodiments will not be repeated here.
[0084] It should be noted that the method for predicting customer cooperation intention in this embodiment can be used to test the prediction model generated by the above embodiments. Then, the prediction model can be continuously optimized according to the test results. This method can also be a practical application method of the prediction model generated by the above embodiments. Using the prediction model generated by the above embodiments to predict customer cooperation intention helps to improve the performance of customer cooperation intention prediction. For example, the recommended customer cooperation intention is relatively accurate.
[0085] Continue to see Figure 5 , as a response to the above Figure 2 The present disclosure provides an embodiment of a device for training a prediction model. Figure 2 Corresponding to the method embodiment shown, the device can be specifically applied to various electronic devices.
[0086] like Figure 5As shown, the device 500 for training a prediction model in this embodiment may include: an acquisition unit 501, a prediction unit 502, a determination unit 503, a selection unit 504, an invitation unit 505, an update unit 506, and a retraining unit 507. Among them, the acquisition unit 501 is configured to obtain the characteristics of at least one customer to be cooperated, wherein the customers include customers who have cooperated and customers who have not cooperated; the prediction unit 502 is configured to input the characteristics of each customer into the prediction model to obtain the cooperation possibility score of each customer; the determination unit 503 is configured to determine at least one target customer for possible cooperation according to the cooperation possibility score; the selection unit 504 is configured to sort the target customers in order from high to low according to the cooperation possibility score, and select a part of the first target customers with the highest cooperation possibility score and a part of the second target customers with the lowest cooperation possibility score according to a predetermined ratio; the invitation unit 505 is configured to send a cooperation invitation to the first target customer and the second target customer, and receive cooperation feedback information from the first target customer and the second target customer; the update unit 506 is configured to update the customers who have cooperated and the customers who have not cooperated according to the cooperation feedback information; the retraining unit 507 is configured to retrain the prediction model based on the updated characteristics of each customer.
[0087] In some optional implementations of this embodiment, the acquisition unit 501 is further configured to: obtain access conditions for cooperation; obtain the latest customer information and a list of customers who have cooperated; filter out customer information that does not meet the access conditions; and extract characteristics of each customer according to predetermined fields from the retained customer information.
[0088] In some optional implementations of this embodiment, the acquisition unit 501 is further configured to: if there is no predetermined field in the customer information, perform semantic analysis on the customer information to obtain features that are synonymous with the predetermined field.
[0089] In some optional implementations of this embodiment, the prediction model 502 includes at least one of the following: a single-class support vector machine, a nearest neighbor knot algorithm model, and an isolation forest.
[0090] In some optional implementations of this embodiment, the prediction model is a single-class support vector machine; and the determination unit 503 is further configured to: map the customer's feature points to a high-dimensional space based on a kernel function according to the atypical proportion of the customers who have cooperated, and train a hyperplane; project the customer's feature points on the hyperplane to obtain a double teardrop-shaped area; and determine the customer whose feature points are within the lower teardrop-shaped area as a target customer for possible cooperation.
[0091] Continue to see Figure 6 , as a response to the above Figure 5The present disclosure provides an embodiment of a device for predicting customer cooperation intention. Figure 4 Corresponding to the method embodiment shown, the device can be specifically applied to various electronic devices.
[0092] like Figure 6 As shown, the device 600 for predicting customer cooperation intention in this embodiment may include: an acquisition unit 601 and a prediction unit 602. The acquisition unit 601 is configured to acquire the characteristics of the customer to be predicted; the prediction unit 602 is configured to input the characteristics into the prediction model trained by the device 500, and output the cooperation possibility score of the customer.
[0093] It should be noted that the collection, collection, updating, analysis, processing, use, transmission, storage and other aspects of user personal information involved in the technical solution of this disclosure are in compliance with the provisions of relevant laws and regulations, are used for legitimate purposes, and do not violate public order and good morals. Necessary measures are taken for user personal information to prevent illegal access to user personal information data and maintain the security of user personal information, network security and national security.
[0094] According to an embodiment of the present disclosure, the present disclosure also provides an electronic device and a readable storage medium.
[0095] An electronic device comprises: one or more processors; a storage device on which one or more computer programs are stored, and when the one or more computer programs are executed by the one or more processors, the one or more processors implement the method described in process 200 or 400.
[0096] A computer-readable medium stores a computer program, wherein the computer program implements the method described in process 200 or 400 when executed by a processor.
[0097] Figure 7 A schematic block diagram of an example electronic device 700 that can be used to implement an embodiment of the present disclosure is shown. The electronic device is intended to represent various forms of digital computers, such as laptop computers, desktop computers, workstations, personal digital assistants, servers, blade servers, mainframe computers, and other suitable computers. The electronic device can also represent various forms of mobile devices, such as personal digital processing, cellular phones, smart phones, wearable devices, and other similar computing devices. The components shown herein, their connections and relationships, and their functions are merely examples and are not intended to limit the implementation of the present disclosure described and / or required herein.
[0098] like Figure 7As shown, the device 700 includes a computing unit 701, which can perform various appropriate actions and processes according to a computer program stored in a read-only memory (ROM) 702 or a computer program loaded from a storage unit 708 into a random access memory (RAM) 703. In the RAM 703, various programs and data required for the operation of the device 700 can also be stored. The computing unit 701, the ROM 702, and the RAM 703 are connected to each other via a bus 704. An input / output (I / O) interface 705 is also connected to the bus 704.
[0099] A number of components in the device 700 are connected to the I / O interface 705, including: an input unit 706, such as a keyboard, a mouse, etc.; an output unit 707, such as various types of displays, speakers, etc.; a storage unit 708, such as a disk, an optical disk, etc.; and a communication unit 709, such as a network card, a modem, a wireless communication transceiver, etc. The communication unit 709 allows the device 700 to exchange information / data with other devices through a computer network such as the Internet and / or various telecommunication networks.
[0100] The computing unit 701 may be a variety of general and / or special processing components with processing and computing capabilities. Some examples of the computing unit 701 include, but are not limited to, a central processing unit (CPU), a graphics processing unit (GPU), various dedicated artificial intelligence (AI) computing chips, various computing units running machine learning model algorithms, digital signal processors (DSPs), and any appropriate processors, controllers, microcontrollers, etc. The computing unit 701 performs the various methods and processes described above, such as a method for outputting information. For example, in some embodiments, the method for outputting information may be implemented as a computer software program, which is tangibly contained in a machine-readable medium, such as a storage unit 708. In some embodiments, part or all of the computer program may be loaded and / or installed on the device 700 via ROM 702 and / or communication unit 709. When the computer program is loaded into RAM 703 and executed by the computing unit 701, one or more steps of the method for outputting information described above may be performed. Alternatively, in other embodiments, the computing unit 701 may be configured to execute a method for outputting information in any other appropriate manner (e.g., by means of firmware).
[0101] Various implementations of the systems and techniques described above herein can be implemented in digital electronic circuit systems, integrated circuit systems, field programmable gate arrays (FPGAs), application specific integrated circuits (ASICs), application specific standard products (ASSPs), systems on chips (SOCs), load programmable logic devices (CPLDs), computer hardware, firmware, software, and / or combinations thereof. These various implementations can include: being implemented in one or more computer programs that can be executed and / or interpreted on a programmable system including at least one programmable processor, which can be a special purpose or general purpose programmable processor that can receive data and instructions from a storage system, at least one input device, and at least one output device, and transmit data and instructions to the storage system, the at least one input device, and the at least one output device.
[0102] The program code for implementing the method of the present disclosure may be written in any combination of one or more programming languages. These program codes may be provided to a processor or controller of a general-purpose computer, a special-purpose computer, or other programmable data processing device, so that the program code, when executed by the processor or controller, enables the functions / operations specified in the flow chart and / or block diagram to be implemented. The program code may be executed entirely on the machine, partially on the machine, partially on the machine and partially on a remote machine as a stand-alone software package, or entirely on a remote machine or server.
[0103] In the context of the present disclosure, a machine-readable medium may be a tangible medium that may contain or store a program for use by or in conjunction with an instruction execution system, device, or equipment. A machine-readable medium may be a machine-readable signal medium or a machine-readable storage medium. A machine-readable medium may include, but is not limited to, an electronic, magnetic, optical, electromagnetic, infrared, or semiconductor system, device, or equipment, or any suitable combination of the foregoing. A more specific example of a machine-readable storage medium may include an electrical connection based on one or more lines, a portable computer disk, a hard disk, a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), an optical fiber, a portable compact disk read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination of the foregoing.
[0104] To provide interaction with a user, the systems and techniques described herein can be implemented on a computer having: a display device (e.g., a CRT (cathode ray tube) or LCD (liquid crystal display) monitor) for displaying information to the user; and a keyboard and pointing device (e.g., a mouse or trackball) through which the user can provide input to the computer. Other types of devices can also be used to provide interaction with the user; for example, the feedback provided to the user can be any form of sensory feedback (e.g., visual feedback, auditory feedback, or tactile feedback); and input from the user can be received in any form (including acoustic input, voice input, or tactile input).
[0105] The systems and techniques described herein may be implemented in a computing system that includes back-end components (e.g., as a data server), or a computing system that includes middleware components (e.g., an application server), or a computing system that includes front-end components (e.g., a user computer with a graphical user interface or a web browser through which a user can interact with implementations of the systems and techniques described herein), or a computing system that includes any combination of such back-end components, middleware components, or front-end components. The components of the system may be interconnected by any form or medium of digital data communication (e.g., a communication network). Examples of communication networks include: a local area network (LAN), a wide area network (WAN), and the Internet.
[0106] A computer system may include a client and a server. The client and the server are generally remote from each other and usually interact through a communication network. The relationship between the client and the server is generated by computer programs running on the respective computers and having a client-server relationship with each other. The server may be a server of a distributed system, or a server combined with a blockchain. The server may also be a cloud server, or an intelligent cloud computing server or intelligent cloud host with artificial intelligence technology. The server may be a server of a distributed system, or a server combined with a blockchain. The server may also be a cloud server, or an intelligent cloud computing server or intelligent cloud host with artificial intelligence technology.
[0107] It should be understood that the various forms of processes shown above can be used to reorder, add or delete steps. For example, the steps recorded in this disclosure can be executed in parallel, sequentially or in different orders, as long as the desired results of the technical solutions disclosed in this disclosure can be achieved, and this document does not limit this.
[0108] The above specific implementations do not constitute a limitation on the protection scope of the present disclosure. It should be understood by those skilled in the art that various modifications, combinations, sub-combinations and substitutions can be made according to design requirements and other factors. Any modification, equivalent substitution and improvement made within the spirit and principle of the present disclosure shall be included in the protection scope of the present disclosure.
Claims
1. A method for training a prediction model, comprising: Acquire characteristics of at least one customer to be cooperated with, wherein the customer includes customers who have cooperated with the customer and customers who have not cooperated with the customer; Input the characteristics of each customer into the prediction model to obtain the cooperation possibility score of each customer; Determine at least one target customer for possible cooperation based on the cooperation possibility score; Sort the target customers in descending order according to their cooperation possibility scores, and select a portion of the first target customers with the highest cooperation possibility scores and a portion of the second target customers with the lowest cooperation possibility scores according to a predetermined ratio; Sending a cooperation invitation to the first target customer and the second target customer, and receiving cooperation feedback information from the first target customer and the second target customer; Update the customers who have cooperated and the customers who have not cooperated according to the cooperation feedback information; The prediction model is retrained based on the updated features of each customer.
2. The method according to claim 1, wherein: The obtaining of the characteristics of at least one client to be cooperated with includes: Obtain access conditions for cooperation; Get the latest customer information and a list of existing customers; Filter out customer information that does not meet the access conditions; The characteristics of each customer are extracted from the retained customer information according to predetermined fields.
3. The method according to claim 2, wherein: The method further comprises: If there is no predetermined field in the customer information, semantic analysis is performed on the customer information to obtain features that are synonymous with the predetermined field.
4. The method according to claim 1, wherein: The prediction model includes at least one of the following: a single-class support vector machine, a nearest neighbor knot algorithm model, and an isolation forest.
5. The method according to claim 4, wherein: The prediction model is a single-class support vector machine; as well as The step of determining at least one target customer for possible cooperation according to the cooperation possibility score includes: According to the atypical proportion of existing customers, the customer's feature points are mapped to a high-dimensional space based on the kernel function, and a hyperplane is trained; The customer's feature points are projected onto the hyperplane to obtain a double teardrop-shaped area; Customers whose feature points are within the lower water drop-shaped area are identified as target customers for possible cooperation.
6. A method for predicting customer cooperation intention, comprising: Obtain the characteristics of the customer to be predicted; The features are input into a prediction model trained according to the method of any one of claims 1 to 5, and a cooperation possibility score of the customer is output.
7. A device for training a prediction model, comprising: An acquisition unit, configured to acquire characteristics of at least one customer to be cooperated with, wherein the customer includes customers who have cooperated with the customer and customers who have not cooperated with the customer; The prediction unit is configured to input the characteristics of each customer into the prediction model to obtain a cooperation possibility score for each customer; A determination unit, configured to determine at least one target customer with whom cooperation is possible according to the cooperation possibility score; A selection unit is configured to sort the target customers in descending order of cooperation possibility scores, and select a portion of first target customers with the highest cooperation possibility scores and a portion of second target customers with the lowest cooperation possibility scores according to a predetermined ratio; an invitation unit, configured to send a cooperation invitation to the first target customer and the second target customer, and receive cooperation feedback information from the first target customer and the second target customer; An updating unit, configured to update the customers who have cooperated and the customers who have not cooperated according to the cooperation feedback information; The retraining unit is configured to retrain the prediction model based on the updated features of each customer.
8. A device for predicting customer cooperation intention, comprising: an acquisition unit configured to acquire characteristics of a customer to be predicted; A prediction unit is configured to input the features into a prediction model trained according to the method of any one of claims 1-5, and output a cooperation possibility score of the customer.
9. An electronic device, comprising: one or more processors; a storage device having one or more computer programs stored thereon, When the one or more computer programs are executed by the one or more processors, the one or more processors implement the method according to any one of claims 1 to 6.
10. A computer readable medium having a computer program stored thereon, wherein: When the computer program is executed by a processor, the method according to any one of claims 1 to 6 is implemented.
