Strategy generation method and device, computer device and storage medium

Abstract

The application relates to a strategy generation method and device, computer equipment and a storage medium, and relates to the field of big data. The method comprises the following steps: determining customer labels of a target customer under each label category based on a decision tree corresponding to each label category, and constructing a customer label set of the target customer by using each customer label; determining a first label similarity between the customer label set of the target customer and customer label sets of each reference customer according to the customer label set of the target customer; determining a target reference customer for the target customer from each reference customer according to the first label similarity; and generating an interaction strategy for the target customer based on an interaction strategy for the target reference customer. The method can improve the accuracy and practicality of the interaction strategy.

Description

Technical Field

[0001] This application relates to the field of big data, and in particular to a strategy generation method, apparatus, computer device, and storage medium. Background Technology

[0002] In the course of its operations, banks have many scenarios that require interaction with customers, such as recommending products to customers and communicating with customers on a certain matter.

[0003] Currently, when conducting interactions, staff often need to proactively organize customer information and rely on experience to determine interaction strategies for each customer. However, this approach is highly dependent on the staff's subjective judgment, resulting in low precision in the interaction strategies and a low success rate for the interactions. Summary of the Invention

[0004] Therefore, it is necessary to provide a strategy generation method, apparatus, computer device, and storage medium to address the aforementioned technical problems.

[0005] Firstly, this application provides a strategy generation method. The method includes:

[0006] Based on the decision tree corresponding to each tag category, the customer tags of the target customer under each tag category are determined, and the customer tag set of the target customer is constructed using each customer tag.

[0007] Based on the target customer's customer tag set, determine the first tag similarity between the target customer's customer tag set and the customer tag sets of each reference customer;

[0008] Based on the first tag similarity, target reference customers for the target customer are determined from each of the reference customers.

[0009] Based on the interaction strategy for the target reference customer, an interaction strategy for the target customer is generated.

[0010] In one embodiment, determining the first tag similarity between the target customer's customer tag set and the customer tag sets of each reference customer, based on the target customer's customer tag set, includes:

[0011] Based on the target customer's customer tag set, determine the second tag similarity between the target customer's customer tag set and the benchmark customer tag set;

[0012] If the similarity of the second tag is greater than the similarity threshold, the target customer category to which the target customer belongs is determined based on the similarity of the second tag;

[0013] For any reference customer corresponding to the target customer category, determine the first tag similarity between the target customer's customer tag set and the reference customer's customer tag set.

[0014] In one embodiment, the target reference customer has a customer identifier, which is used to characterize the interaction results of the target reference customer's historical interactions. The interaction strategy includes interaction strategy elements. Generating an interaction strategy for the target customer based on the interaction strategy for the target reference customer includes:

[0015] If the customer identifier corresponding to the target reference customer indicates that the target reference customer has had successful historical interactions, then the interaction strategy elements of the target reference customer are obtained.

[0016] Based on the interaction strategy elements of the target reference customer, an interaction strategy is generated for the target customer.

[0017] In one embodiment, determining the customer tags of the target customer under each tag category based on the decision tree corresponding to each tag category includes:

[0018] For any of the aforementioned label categories, determine the target customer characteristics of the target customer in relation to the decision tree;

[0019] The customer feature value corresponding to the target customer feature is input into the decision tree to obtain the customer tag corresponding to the target customer under the tag category.

[0020] In one embodiment, determining the first tag similarity between the target customer's customer tag set and the customer tag sets of each reference customer, based on the target customer's customer tag set, includes:

[0021] For any of the reference customers, the customer tags corresponding to each of the tag categories in the customer tag set of the target customer are sequentially concatenated according to the preset order corresponding to each of the tag categories to obtain a first tag string, and the customer tags corresponding to each of the tag categories in the customer tag set of the reference customer are sequentially concatenated to obtain a second tag string;

[0022] Determine the similarity between the first tag string and the second tag string;

[0023] The similarity between the first tag string and the second tag string is used as the first tag similarity between the target customer's customer tag set and the customer tag sets of each reference customer.

[0024] In one embodiment, determining the similarity between the first tag string and the second tag string includes:

[0025] Determine the edit distance between the first tag string and the second tag string;

[0026] The edit distance is used as the similarity between the first tag string and the second tag string.

[0027] Secondly, this application also provides a strategy generation apparatus. The apparatus includes:

[0028] The first determining module is used to determine the customer tags of the target customer under each of the tag categories based on the decision tree corresponding to each tag category, and to construct the customer tag set of the target customer using each of the customer tags;

[0029] The second determining module is used to determine the first tag similarity between the target customer's customer tag set and the customer tag sets of each reference customer, based on the target customer's customer tag set.

[0030] The third determining module is used to determine the target reference customer for the target customer from among the reference customers based on the first tag similarity.

[0031] The generation module is used to generate an interaction strategy for the target customer based on the interaction strategy for the target reference customer.

[0032] In one embodiment, the second determining module is further configured to:

[0033] Based on the target customer's customer tag set, determine the second tag similarity between the target customer's customer tag set and the benchmark customer tag set;

[0034] If the similarity of the second tag is greater than the similarity threshold, the target customer category to which the target customer belongs is determined based on the similarity of the second tag;

[0035] For any reference customer corresponding to the target customer category, determine the first tag similarity between the target customer's customer tag set and the reference customer's customer tag set.

[0036] In one embodiment, the target reference customer has a customer identifier, which is used to characterize the interaction results of the target reference customer's historical interactions. The interaction strategy includes interaction strategy elements. The generation module is further configured to:

[0037] If the customer identifier corresponding to the target reference customer indicates that the target reference customer has had successful historical interactions, then the interaction strategy elements of the target reference customer are obtained.

[0038] Based on the interaction strategy elements of the target reference customer, an interaction strategy is generated for the target customer.

[0039] In one embodiment, the first determining module is further configured to:

[0040] For any of the aforementioned label categories, determine the target customer characteristics of the target customer in relation to the decision tree;

[0041] The customer feature value corresponding to the target customer feature is input into the decision tree to obtain the customer tag corresponding to the target customer under the tag category.

[0042] In one embodiment, the second determining module is further configured to:

[0043] For any of the reference customers, the customer tags corresponding to each of the tag categories in the customer tag set of the target customer are sequentially concatenated according to the preset order corresponding to each of the tag categories to obtain a first tag string, and the customer tags corresponding to each of the tag categories in the customer tag set of the reference customer are sequentially concatenated to obtain a second tag string;

[0044] Determine the similarity between the first tag string and the second tag string;

[0045] The similarity between the first tag string and the second tag string is used as the first tag similarity between the target customer's customer tag set and the customer tag sets of each reference customer.

[0046] In one embodiment, the second determining module is further configured to:

[0047] Determine the edit distance between the first tag string and the second tag string;

[0048] The edit distance is used as the similarity between the first tag string and the second tag string.

[0049] Thirdly, this application also provides a computer device. The computer device includes a memory and a processor, the memory storing a computer program, and the processor executing the computer program to implement any of the methods described above.

[0050] Fourthly, this application also provides a computer-readable storage medium. The computer-readable storage medium stores a computer program thereon, which, when executed by a processor, implements any of the above methods.

[0051] Fifthly, this application also provides a computer program product. The computer program product includes a computer program, which, when executed by a processor, implements any of the methods described above.

[0052] The aforementioned strategy generation method, apparatus, computer equipment, and storage medium can generate a customer tag set for the target customer through decision trees corresponding to each tag category. Based on the first tag similarity between the customer tag sets of the target customer and the customer tag sets of each reference customer, a target reference customer is determined. Then, an interaction strategy for the target customer is generated based on the interaction strategy for the target reference customer. Since customer tags are equivalent to the sum of the target customer's various customer feature values, determining the first tag similarity based on the target customer's and reference customer's customer feature values ​​reduces the dimensionality of comparison, eliminates redundant information, and improves the accuracy of determining the similarity between the target customer and the reference customer compared to directly determining similarity based on the target customer's and reference customer's customer feature values. Therefore, it improves the accuracy of determining the target reference customer. Furthermore, the interaction strategy for the target reference customer is also applicable to the target customer to some extent. Therefore, generating an interaction strategy for the target customer based on the interaction strategy for the target reference customer improves the accuracy and practicality of the interaction strategy. Further, interacting with the target customer based on this generated interaction strategy can significantly increase the probability of successful interaction. Attached Figure Description

[0053] Figure 1 This is a flowchart illustrating a strategy generation method in one embodiment;

[0054] Figure 2 This is a flowchart illustrating step 104 in one embodiment;

[0055] Figure 3 This is a flowchart illustrating step 108 in one embodiment;

[0056] Figure 4 This is a flowchart illustrating step 102 in one embodiment;

[0057] Figure 5 This is a flowchart illustrating step 104 in one embodiment;

[0058] Figure 6 This is a flowchart illustrating step 504 in one embodiment;

[0059] Figure 7 This is a schematic diagram of the editing distance algorithm in one embodiment;

[0060] Figure 8 This is a schematic diagram of a strategy generation method in one embodiment;

[0061] Figure 9 This is a schematic diagram illustrating the generation of a decision tree in one embodiment;

[0062] Figure 10 This is a schematic diagram of a decision tree in one embodiment;

[0063] Figure 11 This is a schematic diagram of the process of generating a decision tree in one embodiment;

[0064] Figure 12 This is a flowchart illustrating the process of generating an interaction strategy in one embodiment;

[0065] Figure 13 This is a structural block diagram of a strategy generation device in one embodiment;

[0066] Figure 14 This is an internal structural diagram of a computer device in one embodiment. Detailed Implementation

[0067] To make the objectives, technical solutions, and advantages of this application clearer, the following detailed description is provided in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative and not intended to limit the scope of this application.

[0068] In one embodiment, such as Figure 1 As shown, a strategy generation method is provided. This embodiment illustrates the application of this method to a terminal. It is understood that this method can also be applied to a server, and further to a system including both a terminal and a server, and is implemented through interaction between the terminal and the server. In this embodiment, the method includes the following steps:

[0069] Step 102: Based on the decision tree corresponding to each tag category, determine the customer tags of the target customer under each tag category, and use each customer tag to construct the customer tag set of the target customer.

[0070] In this embodiment, the target customer is the customer to be interacted with, and the customer tag is used to represent the customer's value category on a certain evaluation dimension. In this embodiment, the interaction can include interactive operations in various scenarios, and this embodiment does not limit this. For example, it can include interactive operations such as making relevant recommendations or push notifications to the target customer, or negotiating with the target customer on a certain matter. The evaluation dimensions are different in different scenarios, and the corresponding tag categories are also different.

[0071] In this embodiment of the application, the following example is taken as an application to a banking system, with the target customer being a customer to whom repayment negotiation is pending. For example, if the evaluation dimension is the overdue amount, then the customer label under the overdue amount can be large overdue amount, relatively large overdue amount, small overdue amount, small overdue amount, etc. One evaluation dimension can correspond to one label category.

[0072] For any given tag category, a decision tree corresponding to that tag category can be constructed in advance based on historical customer data. This application does not specifically limit the method for constructing the decision tree; any method that can construct decision trees for each tag category based on historical customer data is applicable to this application, such as the ID3 algorithm, C4.5 algorithm, and CART (Classification and Regression Tree) algorithm.

[0073] After determining the target customer's tags under each tag category using the decision tree corresponding to each tag category, these tags can be assembled into a customer tag set. For example, when the tag categories are overdue amount, repayment willingness, and customer's current economic situation, the target customer's tag under the decision tree corresponding to overdue amount can be "large overdue amount," the tag under the decision tree corresponding to repayment willingness can be "low repayment willingness," and the tag under the decision tree corresponding to the customer's current economic situation can be "good economic situation." After assembling these tags into a customer tag set, the target customer's customer tag set could be {large overdue amount, low repayment willingness, good economic situation}.

[0074] Step 104: Based on the target customer's customer tag set, determine the first tag similarity between the target customer's customer tag set and the customer tag sets of each reference customer.

[0075] In this embodiment, the reference customer is a pre-selected customer who has already completed the interaction. The interaction result of the reference customer can be either successful or unsuccessful. Taking the target customer as the customer to be negotiated repayment, successful interaction can mean that the customer agrees to the proposed solution in the interaction strategy and repays according to the solution. Unsuccessful interaction can mean that the customer does not agree to the proposed solution in the interaction strategy, or that the customer does not repay according to the solution. Based on the first tag similarity between the target customer's customer tag set and the reference customer's customer tag set, the possible interaction results of using the interaction strategy for the reference customer with the target customer can be predicted, and thus an interaction strategy for the target customer can be generated accordingly.

[0076] The first label similarity is used to characterize the degree of similarity between the target customer's customer label set and the reference customer's customer label set. For example, when the target customer's customer label set is {large overdue amount, low repayment willingness, good economic situation}, and the reference customer's customer label set is {large overdue amount, low repayment willingness, relatively good economic situation}, the first label similarity between the target customer's customer label set and the reference customer's customer label set is high; conversely, when the target customer's customer label set is {large overdue amount, low repayment willingness, good economic situation}, and the reference customer's customer label set is {small overdue amount, high repayment willingness, relatively good economic situation}, the first label similarity between the target customer's customer label set and the reference customer's customer label set is low. This application does not specifically limit the method for determining the first label similarity between the target customer's customer label set and the reference customer's customer label set; any method that can determine the first label similarity between the target customer's customer label set and the reference customer's customer label set is applicable to this application.

[0077] Step 106: Based on the first tag similarity, determine the target reference customers for the target customer from among the reference customers.

[0078] In this embodiment, one or more target reference customers can be determined from each reference customer based on the first tag similarity. For example, when there is only one target reference customer, the reference customer with the highest first tag similarity among all reference customers can be selected as the target reference customer; when there are multiple target reference customers, the reference customers can be sorted according to the first tag similarity of each reference customer, and the reference customer with the preset ranking before sorting can be selected as the target reference customer, or a first tag similarity threshold can be set, and the reference customers with the first tag similarity greater than the first tag similarity threshold can be selected as the target reference customers. This embodiment does not specifically limit this.

[0079] Step 108: Generate an interaction strategy for the target customer based on the interaction strategy for the target reference customer.

[0080] In this embodiment, the interaction strategy can be a strategy for interacting with a target reference customer. Taking repayment negotiation as an example, the interaction strategy can be a communication strategy for negotiating repayment with the target customer. When there is only one target reference customer, an interaction strategy for the target customer can be generated directly based on the interaction strategy for the target reference customer. For example, if the interaction result with the target reference customer is successful, an interaction strategy similar to the interaction strategy for the target reference customer can be generated; if the interaction result with the target reference customer is unsuccessful, an interaction strategy opposite to the interaction strategy for the target reference customer can be generated. This embodiment does not specifically limit this.

[0081] When there are multiple target reference customers, an interaction strategy for each target reference customer can be generated based on the interaction results of each target reference customer and the interaction strategy for each target reference customer. For example, when the interaction results of each target reference customer are all successes or all failures, the interaction strategy that appears most frequently among the target reference customers can be selected as a reference interaction strategy, and an interaction strategy for the target customer can be generated based on the reference interaction strategy; when the interaction results of each target reference customer are both successes and failures, an interaction strategy similar to the interaction strategy of the target reference customer with a successful interaction result, but opposite to the interaction strategy of the target reference customer with a failed interaction result, can be generated, etc. The embodiments of this application do not specifically limit this.

[0082] The strategy generation method provided in this application can generate a customer tag set for the target customer through a decision tree corresponding to each tag category. Based on the first tag similarity between the customer tag sets of the target customer and the customer tag sets of each reference customer, a target reference customer is determined. Then, an interaction strategy for the target customer is generated based on the interaction strategy for the target reference customer. Since customer tags are equivalent to the sum of the target customer's various customer feature values, determining the first tag similarity based on the target customer's customer tag set and the reference customer's customer tag set reduces the dimensions of comparison, eliminates redundant information, and improves the accuracy of determining the similarity between the target customer and the reference customer compared to directly determining the similarity based on the target customer's customer feature values ​​and the reference customer's customer feature values. Therefore, it can improve the accuracy of determining the target reference customer. Furthermore, the interaction strategy for the target reference customer is also applicable to the target customer to a certain extent. Therefore, generating an interaction strategy for the target customer based on the interaction strategy for the target reference customer can improve the accuracy and practicality of the interaction strategy. Further, interacting with the target customer based on the generated interaction strategy can significantly increase the probability of successful interaction.

[0083] In one embodiment, such as Figure 2 As shown, in step 104, based on the target customer's customer tag set, the first tag similarity between the target customer's customer tag set and the customer tag sets of each reference customer is determined, including:

[0084] Step 202: Based on the target customer's customer tag set, determine the second tag similarity between the target customer's customer tag set and the benchmark customer tag set.

[0085] In this embodiment, the baseline customer tag set is a preset customer tag set, which includes preset customer tags corresponding to each tag category. The baseline customer tag set is used to represent a certain ideal customer, and the difficulty of interacting with this ideal customer can be quantified (for example, the ideal customer can be a customer whose interaction result is successful regardless of the interaction strategy used, that is, the difficulty of interacting with this ideal customer is extremely low). Therefore, the difficulty of interacting with the target customer can be determined based on the second tag similarity between the target customer's customer tag set and the baseline customer tag set. This embodiment does not specifically limit the method for determining the second tag similarity between the target customer's customer tag set and the baseline customer tag set; any method that can determine the second tag similarity between the target customer's customer tag set and the baseline customer tag set is applicable to this embodiment.

[0086] Step 204: If the similarity of the second label is greater than the similarity threshold, determine the target customer category to which the target customer belongs based on the similarity of the second label.

[0087] In this embodiment, customer categories are used to characterize the difficulty of interacting with customers. For example, customer categories can be "low interaction difficulty," "relatively low interaction difficulty," "relatively high interaction difficulty," and "high interaction difficulty." The similarity threshold is a preset value. When the similarity of the second tag is less than or equal to the similarity threshold, it can be considered that the interaction difficulty with the target customer is too high, and the expected success rate of interacting with the target customer is too low. Therefore, the strategy generation process can be terminated, and no further interaction strategies will be generated for the target customer. The value of the similarity threshold can be selected by those skilled in the art according to actual needs. For example, when it is necessary to ensure that the interaction difficulty of the target customer for whom an interaction strategy is generated is low, the similarity threshold can be set higher; when it is necessary to generate interaction strategies for more target customers, the similarity threshold can be set lower.

[0088] For example, a second label similarity interval can be set for each customer category. When the second label similarity between the target customer's customer label set and the baseline customer label set falls within the second label similarity interval corresponding to a certain customer category, it can be determined that the customer category belongs to the target customer category. Taking the baseline customer label set representing extremely low customer interaction difficulty, and the larger the second label similarity, the more similar the target customer's customer label set is to the baseline customer label set, as an example, the second label similarity interval corresponding to "low interaction difficulty" can be A ≤ second label similarity, the second label similarity interval corresponding to "relatively low interaction difficulty" can be B ≤ second label similarity < A, the second label similarity interval corresponding to "relatively high interaction difficulty" can be C ≤ second label similarity < B, and the second label similarity interval corresponding to "high interaction difficulty" can be second label similarity < C, where A > B > C. Therefore, when the target customer's customer label set is greater than or equal to B and less than A, it can be determined that the target customer category to which the target customer belongs is "relatively low interaction difficulty".

[0089] Step 206: For any reference customer corresponding to the target customer category, determine the first tag similarity between the target customer's customer tag set and the reference customer's customer tag set.

[0090] In this embodiment of the application, each customer category may correspond to a different reference customer. For example, the customer category to which each reference customer belongs can be determined by determining the third tag similarity between the customer tag set of each reference customer and the benchmark customer tag set. Alternatively, reference customers corresponding to each customer category can be preset. This embodiment of the application does not specifically limit this.

[0091] After determining the target customer category to which the target customer belongs, the first label similarity between the target customer and each reference customer corresponding to the target customer category can be determined.

[0092] The strategy generation method provided in this application can set a baseline customer tag set, and determine the target customer category to which the target customer belongs by determining the second tag similarity between the target customer's customer tag set and the baseline customer tag set, and then determine the first tag similarity between the reference customer's customer tag set corresponding to the target customer category and the target customer's customer tag set. This application embodiment can first determine the target customer category to which the target customer belongs based on the second tag similarity, and then determine the first tag similarity between the reference customer's customer tag set and the target customer's customer tag set based on the reference customer corresponding to the target customer category. Therefore, it can improve the similarity between the reference customer and the target customer, and improve the accuracy of determining the target reference customer. Since it is only necessary to determine the first tag similarity between the target customer's customer tag set and the reference customer's customer tag set corresponding to the target customer category, without needing to determine the first tag similarity for the customer tag sets of all reference customers, it can also reduce the computational load and speed up the determination of the target reference customer.

[0093] In one embodiment, such as Figure 3 As shown, the target reference customer has a customer identifier, which is used to characterize the interaction results of the target reference customer's historical interactions. The interaction strategy includes interaction strategy elements. In step 108, based on the interaction strategy for the target reference customer, an interaction strategy for the target customer is generated, including:

[0094] Step 302: If the customer identifier corresponding to the target reference customer indicates that the target reference customer's historical interactions were successful, obtain the interaction strategy elements of the target reference customer.

[0095] Step 304: Generate an interaction strategy for the target customer based on the interaction strategy elements of the target reference customer.

[0096] In this embodiment, each reference customer has a corresponding customer identifier, which represents the historical interaction results with the reference customer. These results can include success or failure. The interaction strategy can consist of interaction strategy elements and preset templates. The interaction strategy elements can be used to indicate the proposed solution to the user. For example, interaction strategy elements can be "negotiated installment period of 12 months" or "interest and fees can be reduced by 5%". The preset templates can be "For [target customer name], we can provide [interaction strategy element 1], [interaction strategy element 2], and [interaction strategy element 3], with a high expected success rate," or "For [target customer name], the success rate of using [interaction strategy element 1], [interaction strategy element 2], and [interaction strategy element 3] is low; it is recommended to use other methods for interaction," etc. For example, a complete interaction strategy could be "For XXX, we can provide a negotiated installment period of 12 months and a 5% interest and fees reduction, with a high expected success rate."

[0097] If the customer identifier corresponding to the target reference customer indicates successful historical interactions with that customer, it can be assumed that after generating an interaction strategy for the target customer using the target reference customer's interaction strategy elements, the success rate of interacting with the target customer through this strategy will also be high. Therefore, an interaction strategy for the target customer can be generated based on the target reference customer's interaction strategy elements. For example, when the target reference customer's interaction strategy elements are "negotiable installment period of 12 months" and "interest and fee reduction ratio of 5%", an interaction strategy for the target customer can be generated based on these elements: "For XXX, a negotiated installment period of 12 months and an interest and fee reduction ratio of 5% are available, with a high expected success rate."

[0098] It should be noted that, even if the customer identifier corresponding to the target reference customer indicates that the target reference customer's historical interactions have failed, an interaction strategy can still be generated for the target customer based on the target reference customer's interaction strategy elements. This strategy would instruct the target customer to use other methods to interact with the target customer. For example, if the target reference customer's interaction strategy elements are "negotiated installment period of 12 months" and "interest and fee reduction of 5%", the generated interaction strategy for the target customer could be: "The success rate of interactions with XXX using a negotiated installment period of 12 months and an interest and fee reduction of 5% is low; it is recommended to use other methods to interact."

[0099] The strategy generation method provided in this application can generate an interaction strategy for a target customer based on the interaction strategy elements of the target reference customer, provided that the customer identifier corresponding to the target reference customer indicates successful historical interactions with the target reference customer. Since the target reference customer is quite similar to the target customer, it can be assumed that if an interaction with the target reference customer is successful using the interaction strategy for the target reference customer, the success rate of interacting with the target customer through the generated interaction strategy using the interaction strategy elements of the target reference customer is also high. Therefore, the accuracy of determining the interaction strategy for the target customer can be improved.

[0100] In one embodiment, such as Figure 4 As shown, in step 102, based on the decision tree corresponding to each tag category, the customer tags of the target customer under each tag category are determined, including:

[0101] Step 402: For any decision tree corresponding to a label category, determine the target customer characteristics of the target customer in relation to the decision tree.

[0102] Step 404: Input the customer feature values ​​corresponding to the target customer features into the decision tree to obtain the customer tags corresponding to the target customer under the tag category.

[0103] In this embodiment, since the customer features used when constructing different decision trees may differ, the customer feature values ​​input into each decision tree may also differ. For example, if the customer features used when constructing decision tree A corresponding to label category A are customer gender, customer age group, outstanding amount, and total consumption amount, and the customer features used when constructing decision tree B corresponding to label category B are outstanding amount, outstanding interest, account balance, and number of past overdue payments, then when determining the customer label of a target customer under label category A, the target customer features for decision tree A can be determined first: customer gender, customer age group, outstanding amount, and total consumption amount. The corresponding customer feature values ​​for the target customer under these target customer features are then input into decision tree A to obtain the customer label of the target customer under label category A. Similarly, when determining the customer label of a target customer under label category B, the target customer features for decision tree B can be determined first: outstanding amount, outstanding interest, account balance, and number of past overdue payments. The corresponding customer feature values ​​for the target customer under these target customer features are then input into decision tree B to obtain the customer label of the target customer under label category B.

[0104] The strategy generation method provided in this application can determine the target customer features of a target customer against a decision tree, and input the customer feature values ​​corresponding to the target customer features into the decision tree to obtain the customer tags corresponding to the target customer under the tag category. Since the decision tree can integrate the various customer feature values ​​of the target customer and output the corresponding customer tags, compared to directly determining the similarity based on the customer feature values ​​of the target customer and the reference customer, obtaining the customer tag set of the target customer through the decision tree and determining the first tag similarity based on the customer tag set of the target customer and the customer tag set of the reference customer can reduce the dimensions of comparison, eliminate redundant information, and improve the accuracy of determining the similarity between the target customer and the reference customer. Therefore, it can improve the accuracy of determining the target reference customer.

[0105] In one embodiment, such as Figure 5 As shown, in step 104, based on the target customer's customer tag set, the first tag similarity between the target customer's customer tag set and the customer tag sets of each reference customer is determined, including:

[0106] Step 502: For any reference customer, according to the preset order corresponding to each tag category, the customer tags corresponding to each tag category in the target customer's customer tag set are concatenated in sequence to obtain the first tag string, and the customer tags corresponding to each tag category in the reference customer's customer tag set are concatenated in sequence to obtain the second tag string.

[0107] Step 504: Determine the similarity between the first tag string and the second tag string.

[0108] Step 506: The similarity between the first tag string and the second tag string is used as the first tag similarity between the target customer's customer tag set and the customer tag sets of each reference customer.

[0109] In this embodiment of the application, each customer tag can be concatenated to obtain a tag string, and by determining the text similarity between the first tag string of the target customer and the second tag string of the reference customer, the first tag similarity between the customer tag set of the target customer and the customer tag set of each reference customer can be obtained.

[0110] For example, if the preset order of the tag categories is overdue amount, repayment willingness, and economic status, and the target customer's customer tag set is {large overdue amount, low repayment willingness, good economic status}, the reference customer A's customer tag set is {small overdue amount, high repayment willingness, good economic status}, and the reference customer B's customer tag set is {relatively small overdue amount, low repayment willingness, relatively good economic status}, then the customer tags corresponding to each tag category in the target customer's customer tag set can be sequentially concatenated to obtain the first tag string {large overdue amount, low repayment willingness, good economic status}. Then, the customer tags corresponding to each tag category in the reference customer A's customer tag set can be sequentially concatenated to obtain the second tag string A {small overdue amount, high repayment willingness, good economic status}. Finally, the customer tags corresponding to each tag category in the reference customer B's customer tag set can be sequentially concatenated to obtain the second tag string B {relatively small overdue amount, low repayment willingness, relatively good economic status}.

[0111] Furthermore, the similarity between the first tag string and the second tag string A, and the similarity between the first tag string and the second tag string B, can be determined separately. The similarity between the first tag string and the second tag string A is used as the first tag similarity between the target customer's customer tag set and the reference customer A's customer tag set, and the similarity between the first tag string and the second tag string B is used as the first tag similarity between the target customer's customer tag set and the reference customer B's customer tag set. This application embodiment does not specifically limit the method for determining the similarity between the first tag string and the second tag string; any method that can determine the similarity between two strings is applicable to this application embodiment, such as: Euclidean distance algorithm, cosine distance algorithm, edit distance algorithm, etc.

[0112] The strategy generation method provided in this application can concatenate customer tags to obtain tag strings, and obtain the first tag similarity between the target customer's customer tag set and the customer tag sets of each reference customer by determining the text similarity between the first tag string of the target customer and the second tag string of the reference customer. Since customer tags are equivalent to the sum of the customer feature values ​​of the target customer, compared with directly determining the similarity based on the customer feature values ​​of the target customer and the reference customer, determining the first tag similarity based on the customer tag sets of the target customer and the reference customers can reduce the dimensions of comparison, eliminate redundant information, and improve the accuracy of determining the similarity between the target customer and the reference customers. Therefore, it can improve the accuracy of determining the target reference customer.

[0113] In one embodiment, such as Figure 6 As shown, in step 504, determining the similarity between the first tag string and the second tag string includes:

[0114] Step 602: Determine the edit distance between the first tag string and the second tag string.

[0115] Step 604: Use the edit distance as the similarity between the first tag string and the second tag string.

[0116] In this embodiment, the similarity between the first tag string and the second tag string can be determined using an edit distance algorithm. The edit distance algorithm is a method for calculating the literal similarity between two strings. By determining how many character insertion, deletion, or replacement operations are required to obtain the second tag string from the first tag string, the similarity between the first and second tag strings can be obtained. See also... Figure 7 As shown, we can first construct an initial matrix containing the first tag string and the second tag string, and then fill the matrix according to the following logic: For the empty cell (D[i,j]) located in the i-th row and j-th column of the matrix, the value of D[i,j] is the minimum value among D[i-1,j]+1 (D[i-1,j] is the matrix cell located in the (i-1)-th row and j-th column), D[i,j-1]+1 (D[i,j-1] is the matrix cell located in the i-th row and (j-1)-th column), D[i-1,j-1]+1 (if the i-th character of the first tag string is not equal to the j-th character of the second tag string), and D[i-1,j-1] (if the i-th character of the first tag string is equal to the j-th character of the second tag string). After filling all the matrix cells in this way, the value of the matrix cell in the lower right corner of the matrix is ​​the edit distance between the first tag string and the second tag string.

[0117] After determining the edit distance between the first tag string and the second tag string, the edit distance between the first tag string and the second tag string can be used as the similarity between the first tag string and the second tag string.

[0118] The strategy generation method provided in this application can determine the similarity between a first tag string and a second tag string using an edit distance algorithm. Since the similarity between customer tags within the same tag category can usually be represented by the literal similarity between customer tags, using the edit distance algorithm to determine the similarity between the first tag string and the second tag string is more accurate, improving the precision of determining the similarity between the first tag string and the second tag string, and further enhancing the precision of determining the similarity between the target customer and the reference customer.

[0119] To enable those skilled in the art to better understand the embodiments of this application, the embodiments of this application are described below through specific examples.

[0120] Reference Figure 8 The diagram shows a flowchart of a strategy generation method.

[0121] In this embodiment, a decision tree corresponding to each tag category can first be constructed based on historical customer data. For example, the decision tree for each tag category can be constructed based on the C4.5 algorithm. (Refer to...) Figure 9 , Figure 10 As shown, taking repayment willingness as the tag category, and customer tags under this tag category as high repayment willingness, relatively high repayment willingness, low repayment willingness, and low repayment willingness, and using the debt amount, historical overdue number, average number of collections, and account balance as examples to construct the decision tree corresponding to this tag category, the C4.5 algorithm can calculate the information gain ratio of each customer feature and use the customer feature with the highest information gain ratio as the split point of the decision tree. The information gain ratio is the ratio of the information gain of the customer feature (used to characterize the contribution of the customer feature to determining the customer tag; after splitting the decision tree according to the value of the customer feature, the greater the proportion of customers with the same customer tag on each branch, the higher the information gain of the customer feature) to the split information of the customer feature (used to characterize the number and size of branches after splitting the decision tree based on a certain customer feature) (see formula (I)).

[0122]

[0123] Where GainRatio(A) is the information gain ratio of customer feature A, Gain(A) is the information gain of customer feature A, and SplitInfo(A) is the split information of customer feature A.

[0124] Information gain can be obtained by calculating the information entropy of the unsplit node before the decision tree splits (used to characterize the purity of each customer belonging to the unsplit node; the greater the proportion of customers with the same customer label among the customers belonging to the unsplit node, the higher the purity) and the conditional entropy of the decision tree after splitting based on customer features (used to characterize the purity of customers on each branch after the decision tree splits based on customer features) (see formula (II)):

[0125] Gain(A) = Info(D) - Info A (D) Formula (II)

[0126] Where Gain(A) is the information gain of customer feature A, Info(D) is the information entropy of the unsplit node, and Info A (D) represents the conditional entropy of the decision tree after splitting based on customer features.

[0127] The definition of information entropy can be found in Formula (III):

[0128]

[0129] Where m is the total number of customer tag types for customers under unsplit nodes, and p i It is the ratio of the number of customers with the i-th customer label to the total number of customers under the unsplit node.

[0130] The definition of conditional entropy can be found in formula (IV):

[0131]

[0132] Where v is the total number of branches in the decision tree after splitting based on customer features, |D j | represents the number of customers belonging to the j-th branch, and |D| represents the total number of customers under the unsplit node.

[0133] The definition of splitting information can be found in Formula (V):

[0134]

[0135] Where v is the total number of branches in the decision tree after splitting based on customer features, |D j | represents the number of customers belonging to the j-th branch, and |D| represents the total number of customers under the unsplit node.

[0136] By selecting historical customer information as a training set, decision trees for each label category can be trained. For example, historical customer data from a specific quarter in a region across the country can be selected, and based on the business architecture, historical customer information can be obtained from systems across different business areas of the bank. For instance, basic customer information, including name, gender, age, place of origin, ID type, ID number, and contact information, can be obtained from the customer management system; card information, including card number, card status, issuing region, issuing branch, and account balance, can be obtained from the card product system; recent historical transaction details can be obtained from the transaction clearing system; and information such as the amount owed, interest, overdue days, historical overdue frequency, and historical collection records can be obtained from the post-loan management system. This historical customer information can then be stored in a database.

[0137] For example, refer to Figure 11 As shown, decision trees T for each label category can be trained using the following method: Select a training set D. For any label category, multiple customer features can be selected to form a customer feature set A, and an information gain ratio threshold ε can be set. 1) If all customers in the training set D have the same customer label, then decision tree T is a single-node decision tree, and the customer label is used as the class label for that node. 2) If customer feature set A is empty, then decision tree T is a single-node decision tree, and the customer label with the largest number of corresponding customers in the training set D is used as the class label for that node. 3) If a customer in the training set D corresponds to at least two customer labels, and customer feature set A is not empty, then the information gain ratio of all customer features in customer feature set A is calculated. 4) If the information gain ratio of all customer features is less than the information gain ratio threshold ε, then decision tree T is a single-node decision tree, and the customer label with the largest number of corresponding customers in the training set D is used as the class label for that node. 5) If there exists a customer feature with an information gain ratio greater than or equal to the information gain ratio threshold ε, then select the customer feature with the largest information gain ratio as the splitting node of the decision tree T, delete this customer feature from the customer feature set A, and obtain subsets D1 to D2 corresponding to the training set D. i decision tree child nodes T1~T i 6) For any decision tree child node T1 to T2 i Repeat steps 1) to 5) above until a complete decision tree is obtained.

[0138] It should be noted that for customer features with continuous numerical values, these features can first be divided into multiple numerical intervals. For example, customer gender, customer age group, range of outstanding amount, range of outstanding interest, range of account balance, consumption amount, number of overdue days, number of historical overdue payments, and average number of collection attempts during historical overdue periods can be selected as customer features. For customer features with continuous numerical values ​​(customer age group, range of outstanding amount, range of outstanding interest, range of account balance, consumption amount, number of overdue days, number of historical overdue payments, and average number of collection attempts during historical overdue periods), these can be divided into multiple numerical intervals. When splitting the decision tree based on the above customer features, each numerical interval corresponds to a decision tree child node.

[0139] After obtaining the trained decision trees corresponding to each label category, the corresponding customer feature values ​​of the target customer can be input into the decision trees corresponding to each label category to obtain the target customer labels of the target customer under each label category, and the customer label set of the target customer can be constructed accordingly.

[0140] Furthermore, the similarity of the second label between the target customer's customer tag set and the benchmark customer tag set can be determined to identify the target customer category. For example, the benchmark customer tag set could be {large overdue amount, poor current economic situation, low recent spending, good historical credit, high repayment willingness}. Since these customers typically have good credit and a high willingness to repay, and are only unable to repay due to current financial difficulties, they can be considered customers with low interaction difficulty. Conversely, when the target customer's customer tag set is {small overdue amount, poor current economic situation, low recent spending, poor historical credit, high repayment willingness}, the second label similarity between the target customer's tag set and the benchmark customer tag set is high, therefore the target customer category can be considered low interaction difficulty; when the target customer's customer tag set is {large overdue amount, good current economic situation, high recent spending, poor historical credit, low repayment willingness}, the second label similarity between the target customer's tag set and the benchmark customer tag set is low, therefore the target customer category can be considered high interaction difficulty.

[0141] See Figure 12 As shown, the system allows registration of interaction strategy elements for each reference customer. When identifying the target reference customer for the target customer and generating an interaction strategy for the target customer based on the interaction strategy elements of the target reference customer, the system can also include the target customer's basic information, account information, transaction information, etc., as reference data for negotiating repayment, making it convenient for staff to review during the negotiation process with the target customer.

[0142] After staff have finished interacting with the target customer, they can also register the interaction strategy for the target customer in the system so that the target customer can be used as a new reference customer.

[0143] The strategy generation method provided in this application allows banks to proactively predict which customers are currently facing repayment pressure and to actively negotiate repayment with them, thereby improving the bank's foresight in handling non-performing assets. Furthermore, it addresses the issues of cumbersome customer information collection and organization by bank staff during repayment negotiations, and the reliance on staff's historical experience, thus improving the convenience of repayment negotiations.

[0144] It should be understood that although the steps in the flowcharts of the embodiments described above are shown sequentially according to the arrows, these steps are not necessarily executed in the order indicated by the arrows. Unless explicitly stated herein, there is no strict order restriction on the execution of these steps, and they can be executed in other orders. Moreover, at least some steps in the flowcharts of the embodiments described above may include multiple steps or multiple stages. These steps or stages are not necessarily completed at the same time, but can be executed at different times. The execution order of these steps or stages is not necessarily sequential, but can be performed alternately or in turn with other steps or at least some of the steps or stages of other steps.

[0145] Based on the same inventive concept, this application also provides a strategy generation apparatus for implementing the strategy generation method described above. The solution provided by this apparatus is similar to the implementation described in the above method; therefore, the specific limitations in one or more strategy generation apparatus embodiments provided below can be found in the limitations of the strategy generation method described above, and will not be repeated here.

[0146] In one embodiment, such as Figure 13 As shown, a strategy generation device 1300 is provided, including: a first determining module 1302, a second determining module 1304, a third determining module 1306, and a generation module 1308, wherein:

[0147] The first determining module 1302 is used to determine the customer tags of the target customer under each of the tag categories based on the decision tree corresponding to each tag category, and to construct the customer tag set of the target customer using each of the customer tags;

[0148] The second determining module 1304 is used to determine the first tag similarity between the target customer's customer tag set and the customer tag sets of each reference customer based on the target customer's customer tag set;

[0149] The third determining module 1306 is used to determine the target reference customer for the target customer from among the reference customers based on the first tag similarity.

[0150] The generation module 1308 is used to generate an interaction strategy for the target customer based on the interaction strategy for the target reference customer.

[0151] The strategy generation apparatus provided in this application can generate a customer tag set for a target customer through a decision tree corresponding to each tag category. Based on the first tag similarity between the customer tag sets of each reference customer and the target customer's customer tag set, it determines a target reference customer for the target customer. Then, based on the interaction strategy for the target reference customer, it generates an interaction strategy for the target customer. Since customer tags are equivalent to the sum of the target customer's various customer feature values, determining the first tag similarity based on the target customer's customer tag set and the reference customer's customer tag set reduces the dimensions of comparison, eliminates redundant information, and improves the accuracy of determining the similarity between the target customer and the reference customer compared to directly determining the similarity based on the target customer's customer feature values ​​and the reference customer's customer feature values. Therefore, it improves the accuracy of determining the target reference customer. Furthermore, the interaction strategy for the target reference customer is also applicable to the target customer to some extent. Therefore, generating an interaction strategy for the target customer based on the interaction strategy for the target reference customer improves the accuracy and practicality of the interaction strategy. Finally, interacting with the target customer based on the generated interaction strategy significantly increases the probability of successful interaction.

[0152] In one embodiment, the second determining module 1304 is further configured to:

[0153] Based on the target customer's customer tag set, determine the second tag similarity between the target customer's customer tag set and the benchmark customer tag set;

[0154] If the similarity of the second tag is greater than the similarity threshold, the target customer category to which the target customer belongs is determined based on the similarity of the second tag;

[0155] For any reference customer corresponding to the target customer category, determine the first tag similarity between the target customer's customer tag set and the reference customer's customer tag set.

[0156] In one embodiment, the target reference customer has a customer identifier, which is used to characterize the interaction results of the target reference customer's historical interactions. The interaction strategy includes interaction strategy elements. The generation module 1308 is further used to:

[0157] If the customer identifier corresponding to the target reference customer indicates that the target reference customer has had successful historical interactions, then the interaction strategy elements of the target reference customer are obtained.

[0158] Based on the interaction strategy elements of the target reference customer, an interaction strategy is generated for the target customer.

[0159] In one embodiment, the first determining module 1302 is further configured to:

[0160] For any of the aforementioned label categories, determine the target customer characteristics of the target customer in relation to the decision tree;

[0161] The customer feature value corresponding to the target customer feature is input into the decision tree to obtain the customer tag corresponding to the target customer under the tag category.

[0162] In one embodiment, the second determining module 1304 is further configured to:

[0163] For any of the reference customers, the customer tags corresponding to each of the tag categories in the customer tag set of the target customer are sequentially concatenated according to the preset order corresponding to each of the tag categories to obtain a first tag string, and the customer tags corresponding to each of the tag categories in the customer tag set of the reference customer are sequentially concatenated to obtain a second tag string;

[0164] Determine the similarity between the first tag string and the second tag string;

[0165] The similarity between the first tag string and the second tag string is used as the first tag similarity between the target customer's customer tag set and the customer tag sets of each reference customer.

[0166] In one embodiment, the second determining module 1304 is further configured to:

[0167] Determine the edit distance between the first tag string and the second tag string;

[0168] The edit distance is used as the similarity between the first tag string and the second tag string.

[0169] Each module in the aforementioned strategy generation device can be implemented entirely or partially through software, hardware, or a combination thereof. These modules can be embedded in or independent of the processor in a computer device, or stored in the memory of a computer device as software, so that the processor can invoke and execute the operations corresponding to each module.

[0170] In one embodiment, a computer device is provided, which may be a server, and its internal structure diagram may be as follows: Figure 14 As shown, the computer device includes a processor, memory, and a network interface connected via a system bus. The processor provides computing and control capabilities. The memory includes non-volatile storage media and internal memory. The non-volatile storage media stores the operating system and computer programs. The internal memory provides an environment for the operation of the operating system and computer programs in the non-volatile storage media. The network interface is used to communicate with external terminals via a network connection. When the computer program is executed by the processor, it implements a policy generation method.

[0171] Those skilled in the art will understand that Figure 14 The structure shown is merely a block diagram of a portion of the structure related to the present application and does not constitute a limitation on the computer device to which the present application is applied. Specific computer devices may include more or fewer components than those shown in the figure, or combine certain components, or have different component arrangements.

[0172] In one embodiment, a computer device is provided, including a memory and a processor, wherein the memory stores a computer program, and the processor executes the computer program to implement the steps in the above-described method embodiments.

[0173] In one embodiment, a computer-readable storage medium is provided having a computer program stored thereon that, when executed by a processor, implements the steps in the above method embodiments.

[0174] In one embodiment, a computer program product is provided, including a computer program that, when executed by a processor, implements the steps in the above method embodiments.

[0175] It should be noted that the user information (including but not limited to user device information, user personal information, etc.) and data (including but not limited to data used for analysis, data stored, data displayed, etc.) involved in this application are all information and data authorized by the user or fully authorized by all parties.

[0176] Those skilled in the art will understand that all or part of the processes in the methods of the above embodiments can be implemented by a computer program instructing related hardware. The computer program can be stored in a non-volatile computer-readable storage medium, and when executed, it can include the processes of the embodiments of the above methods. Any references to memory, databases, or other media used in the embodiments provided in this application can include at least one of non-volatile and volatile memory. Non-volatile memory can include read-only memory (ROM), magnetic tape, floppy disk, flash memory, optical memory, high-density embedded non-volatile memory, resistive random access memory (ReRAM), magnetic random access memory (MRAM), ferroelectric random access memory (FRAM), phase change memory (PCM), graphene memory, etc. Volatile memory can include random access memory (RAM) or external cache memory, etc. By way of illustration and not limitation, RAM can take many forms, such as Static Random Access Memory (SRAM) or Dynamic Random Access Memory (DRAM). The databases involved in the embodiments provided in this application may include at least one type of relational database and non-relational database. Non-relational databases may include, but are not limited to, blockchain-based distributed databases. The processors involved in the embodiments provided in this application may be general-purpose processors, central processing units, graphics processing units, digital signal processors, programmable logic devices, quantum computing-based data processing logic devices, etc., and are not limited to these.

[0177] The technical features of the above embodiments can be combined in any way. For the sake of brevity, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.

[0178] The embodiments described above are merely illustrative of several implementation methods of this application, and while the descriptions are specific and detailed, they should not be construed as limiting the scope of this patent application. It should be noted that those skilled in the art can make various modifications and improvements without departing from the concept of this application, and these all fall within the protection scope of this application. Therefore, the protection scope of this application should be determined by the appended claims.

Claims

1. A strategy generation method, characterized in that, The method includes: Based on the decision tree corresponding to each tag category, the customer tags of the target customer under each tag category are determined, and the customer tag set of the target customer is constructed using each customer tag. Based on the target customer's customer tag set, determine the first tag similarity between the target customer's customer tag set and the customer tag sets of each reference customer; Based on the first tag similarity, target reference customers for the target customer are determined from each of the reference customers. The target reference customer has a customer identifier, which is used to characterize the interaction results of the target reference customer's historical interactions. The interaction strategy includes interaction strategy elements. If the customer identifier corresponding to the target reference customer indicates that the target reference customer has had successful historical interactions, then the interaction strategy elements of the target reference customer are obtained. Based on the interaction strategy elements of the target reference customer, an interaction strategy is generated for the target customer.

2. The method according to claim 1, characterized in that, The step of determining the first tag similarity between the target customer's customer tag set and the customer tag sets of each reference customer, based on the target customer's customer tag set, includes: Based on the target customer's customer tag set, determine the second tag similarity between the target customer's customer tag set and the benchmark customer tag set; If the similarity of the second tag is greater than the similarity threshold, the target customer category to which the target customer belongs is determined based on the similarity of the second tag; For any reference customer corresponding to the target customer category, determine the first tag similarity between the target customer's customer tag set and the reference customer's customer tag set.

3. The method according to claim 1, characterized in that, The process of determining the customer tags for each target customer under each tag category based on the decision tree corresponding to each tag category includes: For any of the aforementioned label categories, determine the target customer characteristics of the target customer in relation to the decision tree; The customer feature value corresponding to the target customer feature is input into the decision tree to obtain the customer tag corresponding to the target customer under the tag category.

4. The method according to claim 1, characterized in that, The step of determining the first tag similarity between the target customer's customer tag set and the customer tag sets of each reference customer, based on the target customer's customer tag set, includes: For any of the reference customers, the customer tags corresponding to each of the tag categories in the customer tag set of the target customer are sequentially concatenated according to the preset order corresponding to each of the tag categories to obtain a first tag string, and the customer tags corresponding to each of the tag categories in the customer tag set of the reference customer are sequentially concatenated to obtain a second tag string; Determine the similarity between the first tag string and the second tag string; The similarity between the first tag string and the second tag string is used as the first tag similarity between the target customer's customer tag set and the customer tag sets of each reference customer.

5. The method according to claim 4, characterized in that, Determining the similarity between the first tag string and the second tag string includes: Determine the edit distance between the first tag string and the second tag string; The edit distance is used as the similarity between the first tag string and the second tag string.

6. A strategy generation apparatus, characterized in that, The device includes: The first determining module is used to determine the customer tags of the target customer under each of the tag categories based on the decision tree corresponding to each tag category, and to construct the customer tag set of the target customer using each of the customer tags; The second determining module is used to determine the first tag similarity between the target customer's customer tag set and the customer tag sets of each reference customer based on the target customer's customer tag set; The third determining module is used to determine a target reference customer for the target customer from among the reference customers based on the first tag similarity; the target reference customer has a customer identifier, which is used to characterize the interaction results of the target reference customer's historical interactions, and the interaction strategy includes interaction strategy elements. The generation module is used to obtain the interaction strategy elements of the target reference customer based on the customer identifier corresponding to the target reference customer, indicating that the target reference customer's historical interactions were successful; and to generate an interaction strategy for the target reference customer based on the interaction strategy elements of the target reference customer.

7. A computer device comprising a memory and a processor, wherein the memory stores a computer program, characterized in that, When the processor executes the computer program, it implements the steps of the method according to any one of claims 1 to 5.

8. A computer-readable storage medium having a computer program stored thereon, characterized in that, When the computer program is executed by a processor, it implements the steps of the method according to any one of claims 1 to 5.

9. A computer program product, comprising a computer program, characterized in that, When the computer program is executed by a processor, it implements the steps of the method according to any one of claims 1 to 5.

Images