Cross-border e-commerce platform and merchant matching recommendation method and system
By constructing a recommendation evaluation model, using the feature scoring function to calculate the matching score between merchants and platforms, solving the information asymmetry and language barriers between cross-border e-commerce platforms and merchants, realizing accurate platform-merchant matching recommendations, reducing Labor costs and improves the credibility of recommended results.
Patent Information
- Application Number
- CN202411091507.2
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2024-08-08
- Publication Date
- 2024-11-08
AI Technical Summary
There is information asymmetry and language barriers between cross-border e-commerce platforms and merchants, making it difficult to achieve accurate merchant recruitment and platform selection. Traditional investment methods consume a lot of labor costs and are not comprehensive enough.
Build a recommendation evaluation model, based on multi-source heterogeneous data analysis, use feature scoring functions to calculate the matching score between merchants and platforms, realize the two-way matching recommendation between platform-merchants, break the information wall and language barriers, and use data mining technology to unify Score and optimize the model.
It has achieved accurate matching and recommendations between cross-border e-commerce platforms and merchants, reduced labor costs, improved the credibility and matching of recommendation results, and helped merchants choose platforms that fit their own business development.
Smart Images

Figure CN118917875A_ABST
Abstract
Description
Technical Field
[0001] One or more embodiments of this specification relate to the fields of recommendation models and customer mining, and in particular, to a matching recommendation method and system for an e-commerce platform and merchants. Background Art
[0002] With the rapid development of computer technology and the rapid popularization of e-commerce, cross-border e-commerce is welcoming new development opportunities, and various cross-border e-commerce platforms are constantly emerging. At the same time, the types of goods that can be traded across borders through the Internet are also increasing, attracting more and more merchants (sellers) to engage in cross-border e-commerce and open stores on cross-border e-commerce platforms. These merchants quickly expand their business scale by replicating and applying their business experience of advantageous businesses on multiple professional cross-border e-commerce platforms, and cross-border e-commerce platforms continuously improve their overall competitiveness and market share by introducing cross-border merchants that match the platform's positioning, achieving a win-win situation.
[0003] Among them, the natural language barriers and information walls between cross-border merchants and cross-border e-commerce platforms have led to serious information asymmetry between platforms and merchants, becoming the biggest problem hindering the two-way selection between platforms and merchants. For example, for local merchants, due to language problems, they have very little access to information about overseas e-commerce platforms in Latin America, Europe, Southeast Asia, etc., and it is difficult to evaluate whether they are compatible with these platforms.
[0004] On the one hand, in order to meet the growing demand of the cross-border e-commerce market, cross-border e-commerce platforms need to continuously recruit merchants with professional capabilities and successful operating experience. Among them, merchants who have achieved good results on other platforms and sell similar categories of goods have become the preferred recruitment targets of cross-border e-commerce platforms. In order to attract high-quality merchants to join, cross-border e-commerce platforms usually need to unite various service platforms (for example, payment platforms, logistics platforms, etc.), invite merchants under their names, conduct joint investment promotion activities, and screen out high-quality cross-border merchants that meet the conditions for joining the e-commerce platform. However, this method of investment promotion requires a lot of labor costs to conduct KYC (Know Your Customer) and KYB (Know Your Business) audits on merchants. Cross-border merchants in different overseas regions have their own characteristics in language and business characteristics, and corresponding matching strategies need to be customized for different cross-border merchants. Moreover, the operation of e-commerce is often affected by many business factors. The success of cross-border merchants on a certain platform does not mean that they can be successfully replicated on other cross-border e-commerce platforms. Only relying on product categories as the standard for investment promotion evaluation lacks a comprehensive and scientific consideration of cross-border merchants, and cannot achieve accurate investment promotion.
[0005] On the other hand, for merchants focusing on the cross-border e-commerce field, they have an urgent need to understand the compatibility between their own business and the cross-border e-commerce platform. Merchants hope to fully understand the operation of the cross-border e-commerce platform, evaluate whether their qualifications meet the merchant entry standards of the cross-border e-commerce platform, and whether the product categories they operate match the target customer base and logistics and distribution capabilities of the cross-border e-commerce platform, so as to ensure that they can choose and settle in the e-commerce platform suitable for their business development among the numerous cross-border e-commerce platforms, and succeed in the fiercely competitive cross-border e-commerce market. However, due to the very limited cross-border e-commerce platform operation data that merchants can master, merchants face great uncertainty when evaluating and selecting a cross-border e-commerce platform that is compatible with their own business development.
[0006] Therefore, we hope to have a solution that can break down the information wall between cross-border merchants and cross-border e-commerce platforms through technical means, fill the language barrier between cross-border merchants and cross-border e-commerce platforms, comprehensively evaluate the compatibility between cross-border merchants and cross-border e-commerce platforms, and be able to provide platform-merchant matching recommendations, conduct two-way mining, and solve the matching problem between merchants and e-commerce platforms in cross-border scenarios. Summary of the invention
[0007] One or more embodiments of this specification describe a matching recommendation method and system for e-commerce platforms and merchants, and construct a recommendation evaluation model. The model can be based on the analysis of multi-source, heterogeneous data, using corresponding scoring functions formulated for different types of business data, comprehensively calculating the recommendation value between the e-commerce platform and the merchant, and comprehensively evaluating the matching degree, and providing platform-merchant matching recommendations, so as to achieve the purpose of two-way mining of e-commerce platforms and merchants. Data mining technology is used to break the information wall between cross-border merchants and cross-border e-commerce platforms, and a unified scoring standard is used to score platforms-merchants, avoiding information gaps caused by language barriers and achieving accurate recommendations. Furthermore, the embodiments of this specification also evaluate the recommendation evaluation model from the perspectives of recommendation value scoring quality and business implementation effect, and based on the evaluation results of the model, the model is continuously optimized to improve the model quality and enhance the credibility of the recommendation results.
[0008] According to a first aspect, a matching recommendation method for a cross-border e-commerce platform and a merchant is provided, comprising:
[0009] Acquire merchant characteristics of a target merchant and platform characteristics of a target e-commerce platform; the merchant characteristics and platform characteristics have several common characteristic items, wherein the target e-commerce platform is a cross-border e-commerce platform;
[0010] The target merchant and the target e-commerce platform are subjected to target processing by using a recommendation evaluation model, wherein the recommendation evaluation model includes a feature scoring function corresponding to each of a plurality of feature items, and the target processing includes: for any common feature item, using the feature scoring function corresponding to the feature item, calculating the matching score of the merchant feature and the platform feature in the feature item; determining the recommendation value between the target merchant and the target e-commerce platform according to the matching scores of the respective common feature items;
[0011] According to the recommendation value, a platform-merchant matching recommendation is made to the target merchant or the target e-commerce platform.
[0012] In one implementation, the plurality of common feature items include a first feature item, a feature value of the first feature item is of a numerical type, and a first feature scoring function corresponding to the first feature item is a piecewise function.
[0013] The calculating the matching score of the merchant feature and the platform feature in the feature item includes:
[0014] Comparing the characteristic values of the merchant characteristic and the platform characteristic in the first characteristic item to obtain a first result;
[0015] The matching score is determined according to the domain segment within which the first result falls into the first feature scoring function.
[0016] According to an implementation manner, the plurality of common feature items include a second feature item, a feature value of the second feature item is of a set type, and a second feature scoring function corresponding to the second feature item is a piecewise function.
[0017] The calculating the matching score of the merchant feature and the platform feature in the feature item includes:
[0018] Performing a set operation on the feature values of the merchant feature and the platform feature in the second feature item to obtain a second set;
[0019] The matching score is determined according to the segment of the domain of the second feature scoring function into which the number of elements of the second set falls.
[0020] In one implementation, the set operation includes: an intersection operation, a union operation, or a difference operation.
[0021] In one implementation, determining the recommendation value between the target merchant and the target e-commerce platform includes:
[0022] The matching scores of the several common feature items are directly summed to obtain a sum result, and the recommendation value is obtained according to the sum result, wherein the several common feature items include a third feature item and a fourth feature item, the importance of the third feature item is higher than that of the fourth feature item, and the range of the third feature scoring function corresponding to the third feature item is larger than the range of the fourth feature scoring function corresponding to the fourth feature item.
[0023] According to one embodiment, determining the recommendation value between the target merchant and the target e-commerce platform includes:
[0024] The matching scores of the plurality of common feature items are weighted and summed according to the feature item weight to obtain a sum result, and the recommendation value is obtained according to the sum result, and the feature item weight is positively correlated with the importance of the feature item.
[0025] In one scenario of the above implementation, obtaining the recommended value according to the summation result includes:
[0026] The basic score corresponding to the target e-commerce platform is added to the summed result to obtain the recommended value.
[0027] In one implementation, the target e-commerce platform is one of a plurality of pre-stored e-commerce platforms; and performing a platform-merchant matching recommendation to the target merchant or the target e-commerce platform according to the recommendation value includes:
[0028] Sorting the multiple recommendation values between the target merchant and the multiple e-commerce platforms;
[0029] According to the ranking results, several e-commerce platforms are recommended to the target merchant.
[0030] In one implementation, the recommending several e-commerce platforms to the target merchant includes:
[0031] The ranking results are displayed to the target merchants, and recommended value demarcations are given.
[0032] In one implementation, the target merchant is one of a plurality of pre-stored merchants; and performing a platform-merchant matching recommendation to the target merchant or the target e-commerce platform according to the recommendation value includes:
[0033] sorting the multiple recommendation values between the target e-commerce platform and the multiple merchants;
[0034] According to the ranking results, several merchants are recommended to the target e-commerce platform.
[0035] In one implementation, the recommending several merchants to the target e-commerce platform includes:
[0036] The ranking result is displayed to the target e-commerce platform, including the recommendation value between the target e-commerce platform and the multiple merchants, the matching score corresponding to each common feature item, and text annotations.
[0037] In one embodiment, the several common feature items include at least one of the following: a region supporting cross-border payments, a region with cross-border logistics deployment.
[0038] In one embodiment, after determining the recommendation value between the target merchant and the target e-commerce platform, the method further includes:
[0039] The recommendation evaluation model is evaluated according to the model evaluation index; the model evaluation index includes several items of the following: recommendation value variance, recommendation value median and recommendation value average.
[0040] In one implementation, the recommendation evaluation model is a regression model; after evaluating the recommendation evaluation model, the method further includes: optimizing the regression model according to the evaluation result.
[0041] In one embodiment, the method further includes: performing business effectiveness evaluation on the recommendation evaluation model according to business evaluation indicators, wherein the business evaluation indicators include several of the following: the number of merchants successfully reached, the number of merchants successfully converted, and the turnover of merchants successfully converted.
[0042] According to a second aspect, a matching recommendation system for a cross-border e-commerce platform and merchants is provided, comprising:
[0043] Acquisition module: configured to acquire merchant characteristics of a target merchant and platform characteristics of a target e-commerce platform; the merchant characteristics and platform characteristics have several common characteristic items, wherein the target e-commerce platform is a cross-border e-commerce platform;
[0044] Target processing module: configured to perform target processing on the target merchant and target e-commerce platform using a recommendation evaluation model, wherein the recommendation evaluation model includes a feature scoring function corresponding to each of a plurality of feature items, and the target processing includes: for any common feature item, using the feature scoring function corresponding to the feature item, calculating the matching score of the merchant feature and the platform feature in the feature item; determining the recommendation value between the target merchant and the target e-commerce platform according to the matching scores of the respective common feature items;
[0045] Matching recommendation module: configured to make platform-merchant matching recommendations to the target merchant or the target e-commerce platform based on the recommendation value.
[0046] According to a third aspect, a computer program product is provided, comprising a computer program / instruction, which, when executed by a processor, implements the steps of the method described in the first aspect.
[0047] According to a fourth aspect, a computing device is provided, comprising a memory and a processor, wherein an executable code is stored in the memory, and when the processor executes the executable code, the method described in the first aspect is implemented.
[0048] In the embodiments of this specification, a matching recommendation method and system for a cross-border e-commerce platform and a merchant are proposed, involving three aspects: data definition, data collection and processing, and modeling and evaluation of a platform-merchant recommendation evaluation model. First, based on the classification and evaluation criteria of the business, the common feature items of the merchant and the e-commerce platform are defined. Next, the feature values of the e-commerce platform and the merchant under the common feature items are obtained from multiple data sources, and the feature values are standardized. Then, based on expert experience, the recommendation evaluation model is modeled, and the feature scoring function corresponding to each feature item is defined. The matching score of the e-commerce platform and the merchant under each feature item is calculated, and the recommendation value between the e-commerce platform and the merchant is determined based on the matching score. Finally, according to the recommendation value, a platform-merchant matching recommendation is made to the merchant or the e-commerce platform. Further, in the embodiments of this specification, the recommendation evaluation model is evaluated from the perspectives of the quality of the recommendation value scoring and the effect of business implementation, and the model is continuously optimized based on the evaluation results of the model. Furthermore, a regression model is introduced in the embodiments of this specification. According to the model evaluation results, a machine learning algorithm is used to iteratively optimize the recommendation evaluation model to obtain a more reliable model than that set by manual experience. BRIEF DESCRIPTION OF THE DRAWINGS
[0049] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the following briefly introduces the drawings required for use in the description of the embodiments. Obviously, the drawings described below are only some embodiments of the present invention, and for ordinary technicians in this field, other drawings can be obtained based on these drawings without creative work.
[0050] Figure 1 This is a schematic diagram of an exemplary business data collection scenario disclosed in this specification;
[0051] Figure 2 A flowchart of a matching recommendation method between an e-commerce platform and a merchant according to an embodiment of this specification;
[0052] Figure 3 A schematic diagram of a piecewise function definition domain provided according to an embodiment of this specification;
[0053] Figure 4A diagram schematically showing a diagram for recommending an e-commerce platform to a merchant according to an embodiment of this specification;
[0054] Figure 5 A schematic diagram of a list of merchants recommended to an e-commerce platform according to an embodiment of this specification;
[0055] Figure 6 A schematic diagram of a matching recommendation system between an e-commerce platform and merchants according to an embodiment of the present specification. DETAILED DESCRIPTION
[0056] The solution provided in this specification is described below in conjunction with the accompanying drawings.
[0057] As mentioned above, with the development of computer technology and the popularization of e-commerce, various e-commerce platforms have sprung up, and have also attracted a large number of merchants to enter the e-commerce field. In order to meet the needs of business development, e-commerce platforms need to recruit merchants who meet the platform's investment standards and have successful operating experience. However, traditional investment methods have problems such as high labor costs and information asymmetry, making it difficult to achieve accurate matching of merchants. At the same time, when merchants choose the e-commerce platform they are going to enter, they are also faced with the problem of how to reasonably choose an e-commerce platform that is consistent with their own business development because they cannot conduct a comprehensive and objective evaluation of the e-commerce platform.
[0058] These problems are particularly prominent in cross-border e-commerce scenarios. On the one hand, cross-border merchants are distributed in all corners of the world, and the information communication channels between them are very limited, resulting in the inability to share non-commercial secret business data. On the other hand, when collecting information from cross-border merchants, the most common problem encountered by cross-border e-commerce platforms is language barriers, resulting in the collected cross-border merchant information not being able to accurately reflect the merchant's business conditions. In addition, due to the lack of a standard evaluation method, the cross-border e-commerce platform's evaluation method for cross-border merchants is less generalized.
[0059] Therefore, for cross-border e-commerce platforms or merchants, how to efficiently and comprehensively evaluate the matching degree of both parties, formulate a complete set of business data collection methods, standard evaluation methods, and give matching recommendations are technical problems that need to be solved at present. In response to this problem, the inventors studied the business data owned by cross-border e-commerce platforms, cross-border merchants, and cross-border e-commerce service providers and found that data mining can be performed on e-commerce business data stored in different cross-border e-commerce business systems, and several typical business feature items can be selected from them. Then, different scoring rules are formulated for different feature items, and a set of recommendation evaluation models between cross-border e-commerce platforms and cross-border merchants is established. The model can give corresponding matching scores to cross-border e-commerce platforms and cross-border merchants under different feature items according to the scoring rules, and further calculate the recommendation value between cross-border e-commerce platforms and cross-border merchants according to the matching points. After evaluating the recommendation values one by one between multiple cross-border e-commerce platforms and multiple cross-border merchants, potential high-quality merchants can be recommended to cross-border e-commerce platforms according to the recommended value sorting, and cross-border e-commerce platforms suitable for their business development and worth settling in can also be recommended to cross-border merchants. Since the multi-dimensional business feature items from multiple cross-border business systems are used for comprehensive evaluation in the process of calculating the platform-merchant recommendation value, the embodiment of the method of the present invention proposed by the inventor can comprehensively and accurately evaluate the matching degree between the cross-border e-commerce platform and the cross-border merchant, thereby realizing accurate recommendation of the cross-border e-commerce platform or the cross-border merchant. The following is an overview of the implementation scenario examples of the technical concept.
[0060] Before explaining the exemplary scenarios, it should be pointed out that the embodiments of the present invention will mainly describe the methods and systems based on cross-border e-commerce platforms and cross-border e-commerce merchants in overseas e-commerce businesses. In the following text, for the sake of brevity, the e-commerce platform will be used to represent the cross-border e-commerce platform, and the merchant will be used to represent the cross-border merchant. However, this does not represent a limitation on the embodiments of this specification. Any scenario related to the technical concept provided by the embodiments of the present invention can apply the method or device provided by the embodiments of the present invention to calculate the matching score between the platform and the merchant.
[0061] Figure 1 An exemplary business data collection implementation scenario is shown. Figure 1In a typical e-commerce scenario, multiple business stages are involved. In each business stage, the generated phased business data will be deposited in different business systems. For example, in the merchant registration stage, the merchant's principal information, credit information, business scope and other industrial and commercial data will be stored in the business system of the enterprise information service provider (for example, the National Enterprise Credit Information Publicity System, Qichacha, etc.). In the online sales stage, the merchant's listed product information, sales transaction information and other business data will be stored in the e-commerce system of each e-commerce platform where the merchant is stationed. In the logistics stage, the merchant's shipping place, the transaction's receiving place and other logistics information will be stored in the business system of each logistics service provider. In the process of e-commerce transactions, every fund transaction data that occurs will be stored in the business system of the fund service provider. By legally collecting business data in these business systems with authorization, the data assets of the e-commerce platform and the merchant can be formed, and the portrait of the e-commerce platform or the merchant can be constructed based on this.
[0062] In addition, with the continuous operation of e-commerce business, a lot of business data will be generated on the merchant side, such as financial data, business scale data, credit assessment data, intellectual property data, legal litigation data, etc. These data are scattered and stored in different business systems, and can also be used to improve the merchant portrait from multiple angles.
[0063] In addition, by using methods such as online forms, product questionnaires, sales visits, etc., merchants can actively fill in and provide their own business information, so that their business information can be accurately obtained and the merchant portrait can be improved.
[0064] In addition to the several business data collection channels listed above, you can also search and collect business data related to merchants in a reasonable and legal manner on the public Internet based on known merchant information (for example, merchant name, brand name, contact information, etc.).
[0065] The business data collected through various channels described in the above examples are Figure 1 In summary, it is represented as e-commerce platform and merchant data assets. What needs to be understood is that Figure 1 This is only an exemplary representation of the source of business data collection for an e-commerce platform or merchant in one embodiment. In actual scenarios, there may be more data collection channels based on the massive e-commerce transactions that actually occur, or relying on different technical means, for example, collecting commodity turnover information through the business system of a warehousing service provider, data acquisition applications developed based on the API open to the e-commerce platform, etc. The channels and means for collecting business data are not listed one by one in the embodiments of this specification.
[0066] The above embodiments exemplarily illustrate the channels for obtaining various business data of e-commerce platforms and merchants. These business data come from different heterogeneous data sources. For the same category of data items, the collected business data will have different data expressions. For example, in the data category of average monthly turnover, some business systems will store it as two decimal places, while some business systems will store it in scientific notation. These different data expressions for the same type of business data will have a negative impact on the convenience of operating the data items in subsequent steps.
[0067] Therefore, in order to unify data expression and facilitate data comparison and calculation, in one implementation of this embodiment, the collected business data will be standardized, which can not only unify the data format of the business data, but also ensure that the content of the data is unambiguous. For example, for country codes, the three-letter code of ISO 3166 is uniformly adopted; for digital categories, two decimal places are uniformly used for storage; for currency codes, the three-letter code of ISO 4217 is uniformly adopted, and so on.
[0068] In the above embodiment, a description of business data mining for e-commerce platforms and merchants is given. Business data related to e-commerce platforms and merchants are collected from multiple business systems using a variety of technical means. After data standardization, these business data are stored in the database to form data assets of e-commerce platforms and merchants. At the same time, according to the portraits of merchants and e-commerce platforms, several feature items are summarized from the business data. These feature items will be used as basic evaluation elements in subsequent steps to evaluate the matching degree between e-commerce platforms and merchants.
[0069] In one embodiment, with the e-commerce platform as the center, based on the basic information of the e-commerce platform itself and the demands of the merchants on the business data, several standardized feature items are summarized to construct the platform features. Table 1 below gives an example of the feature items of the platform features.
[0070] Table 1
[0071]
[0072] Similar to the above-mentioned summary of standardized feature items for e-commerce platforms, for merchants, several standardized feature items can also be summarized based on their own business data to construct merchant characteristics, such as the exemplary merchant feature items shown in Table 2 below.
[0073] Table 2
[0074]
[0075]
[0076] The platform feature items and merchant feature items shown in the above table are examples of embodiments of this specification and do not represent a limitation on the feature items. In actual business scenarios, based on the platform's specific requirements for merchants and the different operating characteristics of potential merchants, the platform and merchant feature items will be expanded in combination with specific scenarios, which will not be specifically elaborated in the embodiments of this specification.
[0077] Combining Table 1 and Table 2 above, corresponding feature items can be found in merchant features for certain feature items of platform features. For example, for the feature item "store opening fee" of the e-commerce platform, the "store opening fee" of the merchant is the corresponding feature item; for the feature item "commission" of the e-commerce platform, the "payment commission" of the merchant is its corresponding feature item, and so on. These feature items that can establish a corresponding relationship between platform features and merchant features are called common feature items of e-commerce platforms and merchants.
[0078] It should be pointed out that shared feature items not only include one-to-one feature item mapping relationships, but also one-to-many, many-to-one, or many-to-many feature item mapping relationships in some set feature items. For example, for the "dominant category" feature item of an e-commerce platform, its feature value is a collection of several commodity category codes. Correspondingly, in the merchant feature, the union of the merchant's "main sales category" and "other sales categories" constitutes a shared feature item with this feature item.
[0079] In addition, the understanding of e-commerce platforms and merchants should not be limited to a narrow understanding of their business scope. In many embodiments of this specification, the business scope of the e-commerce platform is not specifically limited. These e-commerce platforms include not only domestic e-commerce platforms that provide local e-commerce services, but also cross-border e-commerce platforms involved in cross-border e-commerce business. Similarly, for merchants, it includes both domestic e-commerce merchants operating local e-commerce businesses and cross-border e-commerce merchants engaged in cross-border e-commerce businesses. For cross-border e-commerce platforms and cross-border e-commerce merchants whose business scope covers cross-border e-commerce, some feature items with cross-border business attributes will be their important feature items, such as "regions supporting cross-border payments", "regions with cross-border logistics deployment", and so on.
[0080] The above is an overview of the implementation scenarios and examples of data mining and feature item definition in the technical concept of this specification. Based on the massive business data and feature item definitions of e-commerce platforms and merchants, the matching degree between e-commerce platforms and merchants can be calculated, and platform-merchant matching recommendations can be given. The following will be described in detail with reference to the accompanying drawings and embodiments.
[0081] Figure 2The following is a flow chart of a matching recommendation method between an e-commerce platform and a merchant according to an embodiment of the present specification. It can be understood that the method can be executed by any device, equipment, platform, or device cluster with computing and processing capabilities. Figure 2 In this embodiment, the e-commerce platform is a cross-border e-commerce platform, and the merchant is a cross-border e-commerce merchant. The method includes at least the following steps: Step S201: Obtain the merchant characteristics of the target merchant and the platform characteristics of the target e-commerce platform; the merchant characteristics and the platform characteristics have several common feature items. Step S203: Use the recommendation evaluation model to perform target processing on the target merchant and the target e-commerce platform. The recommendation evaluation model contains feature scoring functions corresponding to multiple feature items. The target processing includes: for any common feature item, using the feature scoring function corresponding to the feature item, calculating the matching score of the merchant characteristics and the platform characteristics in the feature item; according to the matching scores of the several common feature items, determining the recommendation value between the target merchant and the target e-commerce platform. Step S205: According to the recommendation value, make a platform-merchant matching recommendation to the target merchant or the target e-commerce platform.
[0082] The specific implementation methods of the above steps will be described in detail below with reference to the accompanying drawings.
[0083] In step S201, merchant characteristics of a target merchant and platform characteristics of a target e-commerce platform are obtained; the merchant characteristics and platform characteristics have several common characteristic items.
[0084] Based on the above definition of the platform characteristics and merchant characteristics, relevant source data is extracted from various business systems, and data standardization is performed to form e-commerce platform and merchant data assets. The data acquisition and feature item definition of this data asset have been introduced above and will not be repeated here.
[0085] In one embodiment, the merchant feature X includes several feature items x, that is, X = {x1, x2, ..., x j The platform feature Y contains several feature items y, that is, Y = {y1, y2, ..., y k As mentioned above, between merchant feature X and platform feature Y, several groups of common feature items with corresponding relationships can be found. <x i ,y i > = {x i ,y i |x i ∈X∧y i ∈Y}.
[0086] Based on the several common feature items determined in the above steps, in step S203, the target merchant and the target e-commerce platform are subjected to target processing using a recommendation evaluation model, wherein the recommendation evaluation model includes feature scoring functions corresponding to each of the multiple feature items. The target processing includes: for any common feature item, using the feature scoring function corresponding to the feature item, calculating the matching score of the merchant feature and the platform feature in the feature item.
[0087] According to one implementation, various forms of feature scoring functions can be defined for feature items according to the feature value type of the feature item. For example, a piecewise function or a continuous function can be used as the feature scoring function of the feature item; a linear function or a nonlinear function can also be used as the feature scoring function of the feature item. This is not limited in the embodiments of this specification.
[0088] For a common feature item, the corresponding feature scoring function can be expressed as f i (x i ,y i ), where x i represents the merchant feature item in the common feature items, y i represents the platform feature item among the common feature items, f i (·) represents the feature scoring function corresponding to the common feature item. Based on the feature scoring function, the matching score between the merchant feature and the platform feature in the common feature item can be determined.
[0089] In one scenario, the feature value of the feature item is of numerical type, and the feature scoring function corresponding to the feature item is a piecewise function. Correspondingly, in this scenario, the feature value of the common feature item is of numerical type, and the calculation of the matching score of the merchant feature and the platform feature in the feature item includes: performing a comparison operation on the feature values of the merchant feature and the platform feature in the feature item, and determining the matching score according to the domain segmentation of the feature scoring function into which the obtained result falls.
[0090] According to one implementation of this scenario, the common feature item is the "average monthly turnover" feature item of the merchant feature (represented by x) and the "average monthly turnover" feature item required for merchants in the platform feature (represented by y). The characteristic value of this feature item is of numerical type (two decimal places) and is comparable. Therefore, in the feature scoring function, the definition domain is divided into three sections according to the proportional relationship between x and y: the merchant's average monthly turnover exceeds 2 times the platform requirement, the merchant's average monthly turnover meets the platform requirement but does not exceed 2 times, and the merchant's average monthly turnover does not meet the platform requirement. Exemplarily, the feature scoring function is defined as follows:
[0091]
[0092] The domain of this piecewise function includes three domain segments. The domain segment it falls into can be determined based on the comparison results of the "average monthly turnover" feature value of the merchant characteristics and the "average monthly turnover" feature value of the platform characteristics.
[0093] Figure 3 The domain diagram of the piecewise function is provided. According to the diagram, when the merchant's average monthly turnover exceeds twice the platform requirement, the result of the eigenvalue comparison operation falls into the first domain segment of the feature scoring function ( Figure 3 In the figure, it is marked with ①. According to the definition of the feature scoring function, it can be determined that within this domain segment, the matching score between the merchant feature and the platform feature in this common feature item is 2. Similarly, when the merchant's monthly average turnover reaches the platform's requirement but does not exceed 2 times, it falls into the second domain segment ( Figure 3 The matching score is 1; when the merchant's monthly average turnover does not meet the platform's requirements, it falls into the third definition domain segment ( Figure 3 The match score is -1.
[0094] In another embodiment, the characteristic value of the characteristic item is of numerical type, and the characteristic scoring function corresponding to the characteristic item is a continuous function. Since the characteristic value is of numerical type, it can be directly calculated, so the characteristic scoring function can be designed as a continuous numerical calculation function that is compatible with the characteristic item. For example, assuming that the common characteristic item is still "average monthly turnover", the characteristic scoring function can also be designed as: f(x,y)=2(xy), where x is the average monthly turnover value of the merchant, and y is the average monthly turnover value required by the platform.
[0095] In another scenario, the feature value of the feature item is of set type, and the feature scoring function corresponding to the feature item is a piecewise function. Correspondingly, in this scenario, the feature value of the common feature item is of set type, and the calculation of the matching score of the merchant feature and the platform feature in the feature item includes: performing a set operation on the feature values of the merchant feature and the platform feature in the feature item, and determining the matching score according to the number of elements contained in the result set that fall into the domain segment of the feature scoring function.
[0096] According to one implementation of this scenario, the common feature items are the "currency used" feature item of the merchant feature (represented by x) and the "settlement currency" feature item in the platform feature (represented by y) representing the currency that the platform can support. The feature value of this feature item is a set type (an enumeration of ISO 4217 three-letter codes), and set operations can be performed. Therefore, in the feature scoring function, the domain is divided into three sections according to the number of elements in the result set obtained by the intersection operation between x and y: the platform can support more than 2 currencies used by merchants, the platform only supports 1 currency used by merchants, and the platform does not support currencies used by merchants. Exemplarily, the feature scoring function is defined as follows:
[0097]
[0098] Among them, |·| represents the number of elements contained in the set. The domain of the piecewise function includes 3 domain segments. After the intersection operation of the merchant feature "currency used" feature value and the platform feature "settlement currency" feature value is performed, the domain segment it falls into can be determined according to the number of elements in the result set. For example, the merchant feature "currency used" includes 3 currencies: x = {CNY, HKD, USD}, and the platform feature "settlement currency" includes 5 currencies: y = {CNY, GBP, EUR, HKD, MOP}. The intersection of the two feature values gives the result set: x∩y = {CNY, HKD}. The result set contains 2 elements, namely |x∩y| = 2, which falls into the domain segment of |x∩y|≥2. According to the definition of the feature scoring function, it can be determined that within this domain segment, the matching score of the merchant feature and the platform feature in this common feature item is 4.
[0099] In this implementation, the intersection operation is used as an exemplary set operation to illustrate the matching score calculation scenario. In other implementations, the set operation not only includes the intersection operation, but also can be a union operation or a difference operation, which is not exemplified one by one in the embodiments of this specification. In addition, when the common feature item is a set type, the feature scoring function can also be designed as a continuous function, for example, directly performing arithmetic operations based on the number of elements contained in the above result set to obtain the matching score under the feature item.
[0100] In another scenario, the feature value of the feature item is of Boolean type, and the feature scoring function corresponding to the feature item is a piecewise function. Correspondingly, in this scenario, the calculation of the matching score of the merchant feature and the platform feature in the feature item includes: performing a comparison operation on the feature values of the merchant feature and the platform feature in the feature item, and determining the matching score according to the domain segmentation of the feature scoring function into which the obtained result falls.
[0101] According to one implementation of this scenario, the common feature item is the "whether it is operated globally" feature item of the merchant feature (represented by x) and the "whether it is operated globally" feature item required for merchants in the platform feature (represented by y). The feature value of this feature item is of Boolean type and can be logically compared. Therefore, in the feature scoring function, according to the logical relationship between x and y, the definition domain is divided into two sections: the feature values of the merchant and the platform on this feature item are the same, or different. Exemplarily, the feature scoring function is defined as follows:
[0102]
[0103] The logic for determining the matching score is the same as that described in the above scenario. Based on the comparison result of the "whether it is a global operation" feature value of the merchant feature and the "whether it is a global operation" feature value of the platform feature, it is possible to determine the domain segmentation in which it falls into the feature scoring function, and then obtain the matching score of the merchant feature and the platform feature in the common feature item. This will not be elaborated in detail here.
[0104] In one scenario of this embodiment, some common feature items reflect the strong appeal of the platform to the merchants, or the strong appeal of the merchants to the platform, and their importance is higher than other common feature items. In order to reflect the impact of the importance of the feature items representing strong appeals on the matching score, in the recommendation evaluation model, the corresponding feature scoring function is specifically defined according to the importance of the feature items, and the feature scoring function is given an extremely high score value on a certain domain segment, or an extremely low score value is given to expand the value range, so as to reflect the importance of the feature items.
[0105] In one example, the common feature item "average monthly turnover" is a strong demand indicator for the platform to assess merchants. For merchants who cannot meet the platform's average monthly turnover requirements, the platform will not invite them to settle in. Therefore, in the feature scoring function of this feature item, the domain corresponding to the merchant not meeting the platform's requirements can be segmented and assigned a very low score value, thereby reflecting the importance of this item. An exemplary function definition is as follows.
[0106]
[0107] In another example, the common feature item "average monthly turnover" is a strong indicator for the platform to assess merchants. The platform will give priority to merchants whose average monthly turnover exceeds twice the platform's requirement. Therefore, in the feature scoring function of this feature item, the domain segment corresponding to the merchant's average monthly turnover exceeding twice the platform's requirement can be assigned a very high score value, thereby reflecting the importance of this item. An exemplary function definition is as follows.
[0108]
[0109] It should be noted that in the above-mentioned scenarios, in order to describe the piecewise function concisely, a piecewise constant function (i.e., step function) is used as an example. The feature scoring function of the feature item is defined by the step function, which can simplify the complex function into a series of constant linear segments. In this way, a simple linear model can be used for analysis on each domain segment, thereby simplifying the complexity of the matching score evaluation. However, it does not represent a limitation on the feature scoring function. In other scenarios, the feature scoring function can be a piecewise function with different function expressions for different value ranges of the independent variable.
[0110] In this way, for each common feature item, using its corresponding feature scoring function, the matching score of the merchant feature and the platform feature in each of the common feature items can be calculated. Therefore, in step S203, the recommendation value between the target merchant and the target e-commerce platform is determined according to the matching scores of the common feature items.
[0111] In one embodiment, the recommendation value is the sum of the matching scores of the plurality of common feature items. m represents the target e-commerce platform, n represents the number of common feature items between the target e-commerce platform and the target merchant, S m Represents the recommendation value between the target e-commerce platform and the target merchant. The recommendation value formula is defined as follows.
[0112]
[0113] In this embodiment, if different feature items have different degrees of importance, where an important feature item means that the platform has a strong appeal related to the feature item, then its importance can be reflected by setting a larger value range for the feature scoring function of the important feature item (a larger value range will make the matching score of the feature item occupy a larger proportion in the total recommendation value), as shown in the above formulas (4) and (5).
[0114] In another embodiment, the recommendation value is the sum of the matching scores of the several common feature items according to the weight of the feature items. Specifically, among the several common feature items, each common feature item has a different importance for the recommendation value between the target e-commerce platform and the target merchant. When the matching scores of the common feature items are summed to calculate the recommendation value, a higher weight value is assigned to the matching score of the common feature item with a higher importance, and vice versa. Therefore, in this implementation, W i Representing the weight value of the common feature items, the formula for the recommended value can be defined as follows.
[0115]
[0116] For example, when an e-commerce platform recruits merchants, the key consideration is whether the merchant's "unit price per customer" can meet the platform's requirements, while the merchant's "average monthly turnover" is not a key factor considered by the platform. Then, when calculating the recommendation value between the e-commerce platform and the merchant, the matching score of the common feature item "unit price per customer" can be given a larger weight value, such as 10, to increase the proportion of this matching score in the recommendation value, and the common feature item "average monthly turnover" can be given a smaller weight value, such as 0.5, to reduce the impact of this matching score on the recommendation value.
[0117] According to one implementation, the recommendation value is obtained by superimposing the basic score corresponding to the target e-commerce platform on the sum result. Specifically, since each e-commerce platform is in a different stage of development at a certain moment, under the condition that other business conditions are the same, the matching degree of different e-commerce platforms for merchants should also be different. For example, some e-commerce platforms are already the leading platforms in the relevant e-commerce field, while some e-commerce platforms may be in the initial stage. The business maturity of these e-commerce platforms at different stages of business development will have an impact on their attractiveness and fit with merchants. Therefore, when calculating the recommendation value between the target e-commerce platform and the target merchant, the platform basic score is introduced as a variable to reflect the impact of influencing factors other than the common feature items on the recommendation value. Therefore, with b m Representing the basic score of platform m, the formula for the recommended value can be defined as follows.
[0118]
[0119] The above embodiments describe in detail a method for recommending matching between e-commerce platforms and merchants. In one embodiment, according to this method, a recommendation evaluation model can be used to calculate the platform-merchant recommendation values between several merchants and several e-commerce platforms, and according to the ranking of the recommendation values, a priority-arranged list of e-commerce platform or merchant recommendations can be obtained.
[0120] According to one implementation, the target e-commerce platform is one of a plurality of pre-stored e-commerce platforms. After the recommendation values between merchants and e-commerce platforms are calculated, the plurality of recommendation values between the target merchant and the plurality of e-commerce platforms are sorted, and according to the sorting results, a plurality of e-commerce platforms are recommended to the target merchant.
[0121] In one scenario of this implementation, the ranking results are displayed to the target merchant in a graphical form, and recommendation value demarcations are given.
[0122] Figure 4An exemplary chart for recommending e-commerce platforms to merchants is provided. In the chart, the horizontal axis corresponds to various e-commerce platforms, including cross-border e-commerce platforms in various geographical regions, and the vertical axis is the recommendation value of the e-commerce platform for the target merchant. It can be seen that according to the matching recommendation method given in the above embodiment, the recommendation value of each e-commerce platform for the target merchant is calculated, and after the recommendation values are sorted, they are displayed in a bar chart from high to low. The e-commerce platform with the highest recommendation value for the target merchant is on the far left of the bar chart ( Figure 4 ), and the one on the far right is the e-commerce platform with the lowest recommendation value to the target merchant (in Figure 4 In this example, it is 1 point). In addition, in this scenario, the target merchant is also provided with a recommended value delimitation, see Figure 4 The recommended value boundary is shown by a horizontal dotted line, and a dividing line is drawn in the bar chart with the recommended value of 7 points as the boundary. E-commerce platforms with recommended values exceeding the recommended value boundary are the platforms that are recommended to the target merchants, while e-commerce platforms with values below the recommended value boundary are not given priority when recommended to the target merchants.
[0123] According to another implementation, the target merchant is one of a plurality of pre-stored merchants. After the recommendation values between the e-commerce platform and the merchants are calculated, the plurality of recommendation values between the target e-commerce platform and the plurality of merchants are sorted, and a plurality of merchants are recommended to the target e-commerce platform according to the sorting result.
[0124] In one scenario of this implementation, the ranking results are displayed to the target e-commerce platform in a list format, including the recommendation values between the target e-commerce platform and the multiple merchants, the matching scores corresponding to each common feature item, and text annotations.
[0125] Figure 5 An exemplary list of merchants recommended to an e-commerce platform is provided. In the list, the list items correspond to merchants, and the columns in the list display data including: merchant name, recommendation value of the target e-commerce platform corresponding to each merchant (shown in the column "recommended total score" in the figure), and the matching score of the target e-commerce platform corresponding to each merchant on each common feature item (for example, dominant category matching score, monthly average flow matching score, etc.). In addition, in this scenario, for each matching score, a text annotation corresponding to the matching score is also given to the target e-commerce platform to help the target e-commerce platform better understand the reason for the matching score.
[0126] The recommendation value calculation process presented in each of the above embodiments is obtained by calculating and summing the matching scores of the common feature items using the feature scoring function in the recommendation evaluation model. Therefore, evaluating the output results of the recommendation evaluation model can help understand the quality of the recommendation value output of the model and grasp the business effect of the platform-merchant matching recommendation. In multiple embodiments of this specification, not only the recommendation value output by the recommendation evaluation model is evaluated from a statistical perspective, but also the business effect of the platform-merchant matching recommendation is evaluated by monitoring the business data.
[0127] According to one embodiment, after determining the recommendation value between the target merchant and the target e-commerce platform, the recommendation evaluation model is evaluated according to the model evaluation index. The model evaluation index includes several of the following items: recommendation value variance, recommendation value median, and recommendation value average. S is set to represent the platform-merchant recommendation value output by the recommendation evaluation model. In this embodiment, the recommendation evaluation model calculates and outputs m platform-merchant recommendation values, where:
[0128] Average value of recommended values: It can be used to measure the tendency of the recommended value output by the recommendation evaluation model. If the average value is too large, it means that the model tends to output higher recommended values, and vice versa. Therefore, whether the average value of the recommended value is too large or too small, it means that the output of the recommendation evaluation model has a certain tendency, and it is necessary to optimize and adjust the feature scoring function and the recommended value calculation rules in the model. The calculation formula of the average value of the recommended value can be defined as follows.
[0129]
[0130] Median of recommended values: can be used to show the central tendency of multiple recommended values output by the recommendation evaluation model. The distribution characteristics of multiple recommended values can be reflected by comparing the median of recommended values with the average of recommended values. If the median of recommended values is less than the average of recommended values, it indicates that most of the multiple recommended values output by the recommendation evaluation model are low-scoring recommended values, while high-scoring recommended values are relatively few, and vice versa. Therefore, whether the median of recommended values is too large or too small, it indicates that the output of the recommendation evaluation model has a certain inclination, and it is necessary to optimize and adjust the feature scoring function and the recommended value calculation rules in the model. When calculating the median of recommended values, first sort the m platform-merchant recommended values output by the recommendation evaluation model according to the size of the values (for example, from small to large). For the ordered series of recommended values after sorting, find the value in the middle, which is the median of recommended values median(S), which can be defined as follows.
[0131]
[0132] Recommendation value variance: By calculating the average of the squares of the deviations between each recommendation value output by the recommendation evaluation model and the average value of the recommendation value, it reflects the measure of the degree of dispersion of multiple platform-merchant recommendation values output by the model. If the fluctuation range of the dispersion of each recommendation value is small, that is, the variance value is small, it means that the distinction between the multiple recommendation values output by the model is not high, and the feature scoring function and recommendation value calculation rules in the model need to be optimized and adjusted, and vice versa. The evaluation function of the recommendation value variance can be defined as follows.
[0133]
[0134] In addition to the above-mentioned embodiment, in which the recommendation value output by the recommendation evaluation model is evaluated from a statistical perspective, in another embodiment, the business effect of the recommendation evaluation model is evaluated based on business evaluation indicators; the business evaluation indicators include several of the following: the number of successfully reached merchants, the number of successfully converted merchants, and the turnover of successfully converted merchants.
[0135] In one scenario of this embodiment, the values of various business evaluation indicators are compared with the market values of the same period, or AB testing is performed on the business in practice, and the business effect fluctuations on various business evaluation indicators can be obtained after the recommendation value given by the recommendation evaluation model is applied to the platform-merchant matching recommendation, compared with when the model is not applied. Therefore, the recommendation evaluation model can be optimized and adjusted according to the comparison results of the business effects.
[0136] The recommendation evaluation model described in the above multiple embodiments can be implemented and optimized in a variety of ways. In one possible implementation, each feature scoring function and recommendation value calculation formula in the recommendation evaluation model is initialized based on the experience of industry experts. The recommendation evaluation model built based on expert experience has high interpretability and credibility, and is fast to build, which can achieve rapid implementation of the model. In the later stage, based on the model evaluation method given in the above embodiments, the model is iteratively optimized and adjusted using expert experience according to the evaluation results.
[0137] In another possible implementation, the recommendation evaluation model is a regression model; after evaluating the recommendation evaluation model, the regression model is optimized according to the evaluation result. The regression model can be a logistic regression model or a ridge regression model, which is not specifically limited.
[0138] In the initialization stage of the regression model, the recommendation evaluation model can be modeled using expert experience, or the historical data of the aforementioned e-commerce platform and merchant data assets can be used for manual labeling to initialize the recommendation evaluation model. In the subsequent steps, after evaluating the recommendation value output by the model, the difference between the expected value and the value is calculated using the set loss function (e.g., MSE, RMSE, MAE, etc.) according to the evaluation result, and the recommendation evaluation model is iteratively optimized using mathematical methods (e.g., gradient descent method), and function fitting is performed on each feature scoring function so that the matching score output by the model can gradually approach the feedback of the actual business situation.
[0139] In this specification, the word "first" in terms such as the first feature item, the first feature scoring function, and the corresponding "second", "third", "fourth" (if any), etc. in the text are merely for the convenience of distinction and description and do not have any limiting meaning.
[0140] The foregoing describes certain embodiments of the present specification, and other embodiments are within the scope of the appended claims. In some cases, the actions or steps described in the claims may be performed in an order different from that in the embodiments, and the desired results may still be achieved. In addition, the processes depicted in the accompanying drawings do not necessarily have to be performed in the specific order or sequential order shown to achieve the desired results. In some embodiments, multitasking and parallel processing are also possible or may be advantageous.
[0141] Figure 6 Schematic diagram of a matching recommendation system for e-commerce platforms and merchants according to an embodiment of this specification. The system 600 is deployed in a computing device, which can be implemented by any device, equipment, platform, device cluster, etc. with computing and processing capabilities. Figure 2 In the system, the e-commerce platform is a cross-border e-commerce platform, the merchant is a cross-border e-commerce merchant, and the system 600 includes:
[0142] Acquisition module 601: configured to acquire merchant characteristics of a target merchant and platform characteristics of a target e-commerce platform; the merchant characteristics and platform characteristics have several common characteristic items.
[0143] Target processing module 602: configured to perform target processing on the target merchant and target e-commerce platform using a recommendation evaluation model, wherein the recommendation evaluation model includes feature scoring functions corresponding to multiple feature items, and the target processing includes: for any common feature item, using the feature scoring function corresponding to the feature item, calculating the matching score of the merchant feature and the platform feature in the feature item; determining the recommendation value between the target merchant and the target e-commerce platform based on the matching scores of the several common feature items.
[0144] Matching recommendation module 603: configured to make platform-merchant matching recommendations to the target merchant or the target e-commerce platform according to the recommendation value.
[0145] According to another embodiment of the present specification, there is also provided a computer program product, including a computer program / instruction, which implements the aforementioned combination when executed by a processor. Figure 2 The steps of the method.
[0146] According to another embodiment, the present specification further provides a computing device, including a memory and a processor, wherein the memory stores executable code, and when the processor executes the executable code, the aforementioned combination is implemented. Figure 2 The steps of the method.
[0147] Those skilled in the art should be aware that in one or more of the above examples, the functions described in the embodiments of the present invention may be implemented using hardware, software, firmware, or any combination thereof. When implemented using software, these functions may be stored in a computer-readable medium or transmitted as one or more instructions or codes on a computer-readable medium.
[0148] The specific implementation methods described above further describe the purpose, technical solutions and beneficial effects of the embodiments of the present invention in detail. It should be understood that the above description is only a specific implementation method of the embodiments of the present invention and is not intended to limit the scope of protection of the present invention. Any modification, equivalent replacement, improvement, etc. made on the basis of the technical solution of the present invention shall be included in the scope of protection of the present invention.
Claims
1. A matching recommendation method for a cross-border e-commerce platform and a merchant, comprising: Acquire merchant characteristics of a target merchant and platform characteristics of a target e-commerce platform; the merchant characteristics and platform characteristics have several common characteristic items; The target e-commerce platform is a cross-border e-commerce platform; The target merchant and the target e-commerce platform are subjected to target processing by using a recommendation evaluation model, wherein the recommendation evaluation model includes a feature scoring function corresponding to each of a plurality of feature items, and the target processing includes: for any common feature item, using the feature scoring function corresponding to the feature item, calculating the matching score of the merchant feature and the platform feature in the feature item; determining the recommendation value between the target merchant and the target e-commerce platform according to the matching scores of the respective common feature items; According to the recommendation value, a platform-merchant matching recommendation is made to the target merchant or the target e-commerce platform.
2. The method according to claim 1, wherein: The plurality of common feature items include a first feature item, a feature value of the first feature item is a numerical type, and a first feature scoring function corresponding to the first feature item is a piecewise function; The calculating the matching score of the merchant feature and the platform feature in the feature item includes: Comparing the characteristic values of the merchant characteristic and the platform characteristic in the first characteristic item to obtain a first result; The matching score is determined according to the domain segment within which the first result falls into the first feature scoring function.
3. The method according to claim 1, wherein: The plurality of common feature items include a second feature item, a feature value of the second feature item is a set type, and a second feature scoring function corresponding to the second feature item is a piecewise function; The calculating the matching score of the merchant feature and the platform feature in the feature item includes: Performing a set operation on the feature values of the merchant feature and the platform feature in the second feature item to obtain a second set; The matching score is determined according to the segment of the domain of the second feature scoring function into which the number of elements of the second set falls.
4. The method according to claim 3, wherein: The set operation includes: intersection operation, union operation or difference operation.
5. The method according to claim 1, wherein: The determining of the recommendation value between the target merchant and the target e-commerce platform includes: The matching scores of the several common feature items are directly summed to obtain a sum result, and the recommendation value is obtained according to the sum result, wherein the several common feature items include a third feature item and a fourth feature item, the importance of the third feature item is higher than that of the fourth feature item, and the range of the third feature scoring function corresponding to the third feature item is larger than the range of the fourth feature scoring function corresponding to the fourth feature item.
6. The method according to claim 1, wherein: The determining of the recommendation value between the target merchant and the target e-commerce platform includes: The matching scores of the plurality of common feature items are weighted and summed according to the feature item weight to obtain a sum result, and the recommendation value is obtained according to the sum result, and the feature item weight is positively correlated with the importance of the feature item.
7. The method according to claim 5 or 6, wherein: The obtaining the recommended value according to the summation result includes: The basic score corresponding to the target e-commerce platform is added to the summed result to obtain the recommended value.
8. The method according to claim 1, wherein: The target e-commerce platform is one of a plurality of pre-stored e-commerce platforms; and performing a platform-merchant matching recommendation to the target merchant or the target e-commerce platform according to the recommendation value, comprises: Sorting the multiple recommendation values between the target merchant and the multiple e-commerce platforms; According to the ranking results, several e-commerce platforms are recommended to the target merchant.
9. The method according to claim 8, wherein: The recommending several e-commerce platforms to the target merchant includes: The ranking results are displayed to the target merchants, and recommended value demarcations are given.
10. The method according to claim 1, wherein: The target merchant is one of a plurality of pre-stored merchants; and performing a platform-merchant matching recommendation to the target merchant or the target e-commerce platform according to the recommendation value, including: sorting the multiple recommendation values between the target e-commerce platform and the multiple merchants; According to the ranking results, several merchants are recommended to the target e-commerce platform.
11. The method according to claim 10, wherein: The recommending several merchants to the target e-commerce platform includes: The ranking result is displayed to the target e-commerce platform, including the recommendation value between the target e-commerce platform and the multiple merchants, the matching score corresponding to each common feature item, and text annotations.
12. The method according to claim 1, wherein: The several common feature items include at least one of the following: regions supporting cross-border payments, regions with cross-border logistics deployment.
13. The method according to claim 1, wherein: After determining the recommendation value between the target merchant and the target e-commerce platform, the method further includes: The recommendation evaluation model is evaluated according to the model evaluation index; the model evaluation index includes several items of the following: recommendation value variance, recommendation value median and recommendation value average.
14. The method according to claim 13, wherein: The recommendation evaluation model is a regression model; after evaluating the recommendation evaluation model, the method further includes: According to the evaluation results, the regression model is optimized.
15. The method according to claim 1, further comprising: The business effect of the recommendation evaluation model is evaluated according to business evaluation indicators, and the business evaluation indicators include several items among the following: the number of merchants successfully reached, the number of merchants successfully converted, and the turnover of merchants successfully converted.
16. A matching recommendation system between a cross-border e-commerce platform and merchants, comprising: Acquisition module: configured to acquire merchant characteristics of a target merchant and platform characteristics of a target e-commerce platform; the merchant characteristics and platform characteristics have several common characteristic items; wherein the target e-commerce platform is a cross-border e-commerce platform; Target processing module: configured to perform target processing on the target merchant and target e-commerce platform using a recommendation evaluation model, wherein the recommendation evaluation model includes a feature scoring function corresponding to each of a plurality of feature items, and the target processing includes: for any common feature item, using the feature scoring function corresponding to the feature item, calculating the matching score of the merchant feature and the platform feature in the feature item; determining the recommendation value between the target merchant and the target e-commerce platform according to the matching scores of the respective common feature items; Matching recommendation module: configured to make platform-merchant matching recommendations to the target merchant or the target e-commerce platform based on the recommendation value.
17. A computer program product, comprising a computer program / instruction, which implements the steps of the method according to any one of claims 1 to 15 when executed by a processor.
18. A computing device comprising a memory and a processor, characterized in that: The memory stores executable codes, and when the processor executes the executable codes, the method according to any one of claims 1 to 15 is implemented.
Citation Information
Patent Citations
Vertical e-commerce platform commodity intelligent recommendation method based on big data analysis and cloud computing and cloud service platform
CN113283960A
Cross-border e-commerce platform recommendation method and related products
CN117010996A
KR20230138988A
