Electronic commerce integrated management system and method based on big data analysis

Abstract

The invention relates to the technical field of e-commerce management, in particular to an e-commerce comprehensive management system and method based on big data analysis, and the system comprises a collaborative filtering risk analysis module which corresponds to the historical after-sales information record of a user. And generating a collaborative filtering risk value of the corresponding user based on the biased intervention feature set corresponding to the after-sales e-commerce shop. According to the method, retrieval keywords and commodity browsing information of a user are considered, the influence of after-sales information of the user on commodity recommendation information is also considered, and a portrait preference association set of the user and a deviation intervention feature set of each e-commerce shop in the corresponding set are generated; meanwhile, according to the historical after-sales information record of the user, dynamically generating a collaborative filtering risk value of the corresponding user based on a biased intervention feature set corresponding to the after-sales e-commerce shop; therefore, intelligent screening of the user preference recommendation content set is realized, and effective management of e-commerce information is realized.

Description

Technical Field

[0001] The present invention relates to the technical field of e-commerce management, and in particular to an e-commerce integrated management system and method based on big data analysis. Background Art

[0002] With the continuous development of social economy and Internet technology, the popularity and application of e-commerce is becoming increasingly higher. In order to increase user stickiness, existing e-commerce software often makes intelligent product recommendations to users based on their historical search records and registration information, predicts their product needs, and pushes corresponding product information.

[0003] However, when existing e-commerce integrated management systems based on big data analysis generate user product recommendation information, they often only rely on users' search keywords and product browsing information. The product recommendation information generated in this way is relatively one-sided and does not take into account the impact of users' after-sales information on product recommendation information. As a result, the conversion rate of recommended products for users in the existing technology is low. Therefore, the existing technology has major defects. Summary of the Invention

[0004] The purpose of the present invention is to provide an e-commerce integrated management system and method based on big data analysis to solve the problems raised in the above background technology.

[0005] In order to solve the above technical problems, the present invention provides the following technical solution: an e-commerce comprehensive management method based on big data analysis, the method comprising the following steps:

[0006] S1. Obtain the user's behavior records within a recent preset time period to generate the user's preferred content information; combine the characteristic keywords of the e-commerce store's corresponding products to obtain the profile matching value of each e-commerce store based on the user's behavior records; record the set of e-commerce stores whose profile matching values ​​based on the user's behavior records are greater than a threshold as the corresponding user's profile preference association set;

[0007] S2. Based on the product operation feature information set of each e-commerce store in the user's profile preference association set, obtain the biased intervention feature set corresponding to each e-commerce store; and combine the historical after-sales information records of the corresponding user to generate the biased intervention feature analysis value corresponding to the biased intervention feature set of each e-commerce store based on the corresponding user's after-sales behavior profile;

[0008] S3. Combine the historical after-sales information records of the corresponding user to generate a collaborative filtering risk value for the corresponding user based on the biased intervention feature set corresponding to the after-sales e-commerce store;

[0009] S4. Extract the biased intervention feature analysis value corresponding to each element in the corresponding portrait preference association set based on the corresponding user's after-sales behavior portrait, as well as the collaborative filtering risk value of the corresponding user based on the biased intervention feature set corresponding to the after-sales e-commerce store; and filter and optimize the portrait preference association set corresponding to the corresponding user based on the obtained collaborative filtering risk value and the obtained biased intervention feature analysis value, generate the corresponding user's preference recommendation content set, and feed it back to the corresponding user's e-commerce account display terminal.

[0010] Furthermore, the user behavior record includes a keyword set consisting of keywords corresponding to titles of products browsed, clicked, and purchased by the user and user search keywords, as well as an array of product records consisting of products browsed, clicked, and purchased by the user;

[0011] The user's preference content information includes one or more preference content data groups, each preference content data group corresponds to a product category, and the preference content data group is recorded as [PC, PF], where PC represents the product category corresponding to the corresponding preference content data group; PF represents the frequency of product records belonging to PC in the corresponding preference content data group in the product record array in the user behavior record;

[0012] The characteristic keywords of the products of the e-commerce store are the keywords set by the administrator of the corresponding e-commerce store according to the product characteristics;

[0013] When the portrait matching value of each e-commerce store based on the user behavior record is obtained, the portrait matching value of the i-th e-commerce store based on the user behavior record is recorded as Pm i ,

[0014]

[0015] Among them, GKB represents the keyword set in the user behavior record; GKF represents the set of characteristic keywords corresponding to the products of each element in the product record array in the user behavior record; both GKB and GKF are not empty sets; Gi represents the set of characteristic keywords corresponding to the products of the i-th e-commerce store; PF i represents the frequency of product records corresponding to the product category of the i-th e-commerce store in the user's preference content information; ZPF represents the sum of the frequency of corresponding product records in each preference content data group in the user's preference content information; count{} represents a function that counts the number of elements in a set; μ represents a preset normalization coefficient.

[0016] The present invention constructs user portraits through user behavior records (including a keyword set consisting of keywords in the titles of products browsed, clicked and purchased by users and user search keywords, and an array of product records consisting of products browsed, clicked and purchased by users), generates user preference content information, and analyzes the relationship (matching status) between e-commerce shops and user preference content information based on the characteristic keywords of the products corresponding to each e-commerce shop, thereby quantifying the degree of association of e-commerce shops based on users.

[0017] Furthermore, the e-commerce store corresponding to the nth element in the user's profile preference association set is recorded as SN n , SN n The corresponding commodity operation feature information set is recorded as RCC n ;

[0018] The RCC n It includes fraud information assessment value, transshipment information assessment curve and out-of-stock risk assessment value;

[0019] RCC n The fraud information evaluation value in is recorded as QZn, QZn=CBn+r·STn / SAn; r represents the preset weight coefficient; CBn represents SN n The percentage of negative reviews in the product reviews; SAn stands for SN n The number of times the product was involved in the complaint platform; STn represents SN n The number of successful user complaints in the ruling results of the corresponding complaints of the internal products;

[0020] Get RCC n When evaluating the transit information curve in the , obtain the corresponding e-commerce store SN in the historical data n The corresponding transshipment risk data pairs of the corresponding products at different transportation distances, wherein the first data in the transshipment risk data pair is the transportation distance of the corresponding product, and the second data is the proportion of orders with abnormal product damage in all orders whose transportation distance between the user order location and the corresponding product delivery warehouse is less than the corresponding first data; the SN-based n The transshipment risk data of the products are fitted with a curve, and the function corresponding to the fitted curve is used as the RCC n The transport information assessment curve in ;

[0021] Get RCC n When evaluating the out-of-stock risk value in n The out-of-stock risk assessment value in is recorded as DHn.

[0022] DHn=Vn / Cn·TPn,

[0023] Where Cn represents SN n The average inventory of the product at different times; Vn represents SN n The average sales rate of the product at different times; TPn represents SN n The average value of the delivery delay deviation corresponding to each product order in the delivery delay deviation represents the interval between the courier pickup time of the logistics company to which the corresponding product order belongs and the time when the corresponding user submits the product order;

[0024] Get SN n The corresponding bias intervention feature set includes the fraud information assessment value, the transshipment information bias risk value and the out-of-stock risk assessment value; when obtaining the transshipment information bias risk value, obtain the distance between the delivery order location submitted by the corresponding user and the corresponding product shipping warehouse, recorded as L; substitute L into RCC n The corresponding function value in the transport information evaluation curve is used as SN n The corresponding biased risk value of the transport information in the biased intervention feature set.

[0025] When constructing a set of commodity operation feature information, the present invention takes into consideration three aspects: fraud information evaluation value, transshipment information evaluation curve, and out-of-stock risk evaluation value, and displays a comprehensive evaluation of e-commerce stores, providing data support for generating corresponding user after-sales behavior portraits in subsequent steps based on the biased intervention feature analysis values ​​corresponding to the biased intervention feature set of each e-commerce store.

[0026] Furthermore, when generating the biased intervention feature analysis value corresponding to the biased intervention feature set of each e-commerce store based on the historical after-sales information record of the corresponding user, the biased intervention feature analysis value corresponding to the biased intervention feature set of the nth element in the user's portrait preference association set of the corresponding user's after-sales behavior portrait is recorded as BFAV n ,

[0027] BFAV n =Pm in / exp(r1·QZn+r2·ZYn+r3·DHn),

[0028] Among them, QZn represents SN n The corresponding bias intervention feature set fraud information evaluation value; ZYn represents SN n The corresponding bias risk value of the transfer information in the bias intervention feature set; DHn represents SN n The corresponding out-of-stock risk assessment value in the biased intervention feature set; Pm in Indicates SN n Profile matching value based on user behavior records;

[0029] r1 represents the proportion of orders with fraud information assessment values ​​greater than or equal to QZn in the corresponding biased intervention feature set in the historical after-sales information records; r2 represents the proportion of orders with transit information bias risk values ​​greater than or equal to ZYn in the corresponding biased intervention feature set in the historical after-sales information records; r3 represents the proportion of orders with out-of-stock risk assessment values ​​greater than or equal to DHn in the corresponding biased intervention feature set in the historical after-sales information records.

[0030] Furthermore, when generating the collaborative filtering risk value of the corresponding user based on the biased intervention feature set of the after-sales e-commerce store,

[0031] Obtain the biased intervention feature set corresponding to each after-sales order in the historical after-sales information record of the corresponding user and the portrait matching value of the e-commerce store to which the corresponding order belongs based on the user behavior record at the corresponding after-sales time point, and combine the calculation formula of the biased intervention feature analysis value corresponding to the biased intervention feature set of the nth element in the user's portrait preference association set based on the after-sales behavior portrait of the corresponding user to generate the biased intervention feature analysis value corresponding to each after-sales order in the historical after-sales information record of the corresponding user, and record the average value of the biased intervention feature analysis value corresponding to each after-sales order in the historical after-sales information record of the corresponding user as the collaborative filtering risk value of the corresponding user based on the biased intervention feature set corresponding to the after-sales e-commerce store.

[0032] The present invention combines the biased intervention feature set corresponding to each after-sales order in the user's historical after-sales information record and the portrait matching value of the e-commerce store to which the corresponding order belongs based on the user behavior record at the corresponding after-sales time point, and generates a collaborative filtering risk value of the corresponding user based on the biased intervention feature set corresponding to the after-sales e-commerce store. The obtained collaborative filtering risk value is dynamically changing and will continue to change according to changes in the user's historical after-sales information record. The obtained collaborative filtering risk value can realize the filtering optimization of the elements in the portrait preference association set corresponding to the user. The obtained collaborative filtering risk value is equivalent to the threshold standard for performing filtering optimization operations on the portrait preference association set corresponding to the user.

[0033] Furthermore, in the process of filtering and optimizing the corresponding user's profile preference association set based on the obtained collaborative filtering risk value and the obtained biased intervention feature analysis value to generate the corresponding user's preference recommendation content set,

[0034] The remaining set after removing the elements whose corresponding preference intervention feature analysis value is less than or equal to the collaborative filtering risk value in the corresponding portrait preference association set is used as the preference recommendation content set of the corresponding user.

[0035] An integrated e-commerce management system based on big data analysis, comprising the following modules:

[0036] A preference feature association analysis module, which obtains the user's user behavior records within a recent preset time period and generates the user's preferred content information; combines the characteristic keywords of the e-commerce store's corresponding products to obtain the profile matching value of each e-commerce store based on the user's behavior records; and records the set of e-commerce stores whose profile matching values ​​based on the user's behavior records are greater than a threshold as the corresponding user's profile preference association set;

[0037] An operation feature bias analysis module, which obtains a bias intervention feature set corresponding to each e-commerce store based on the product operation feature information set of each e-commerce store in the user's profile preference association set; and generates a bias intervention feature analysis value corresponding to the bias intervention feature set of each e-commerce store based on the corresponding user's historical after-sales information record;

[0038] A collaborative filtering risk analysis module, wherein the collaborative filtering risk analysis module generates a collaborative filtering risk value for the corresponding user based on a biased intervention feature set corresponding to the after-sales e-commerce store corresponding to the user's historical after-sales information record;

[0039] A preference filtering management module extracts the biased intervention feature analysis value corresponding to each element in the corresponding portrait preference association set based on the corresponding user's after-sales behavior portrait, as well as the collaborative filtering risk value of the corresponding user based on the biased intervention feature set corresponding to the after-sales e-commerce store; and based on the obtained collaborative filtering risk value and the obtained biased intervention feature analysis value, the portrait preference association set corresponding to the corresponding user is filtered and optimized to generate the corresponding user's preference recommendation content set, and feed it back to the corresponding user's e-commerce account display terminal.

[0040] Furthermore, the operation feature bias analysis module includes a bias intervention feature set acquisition unit and a bias intervention feature analysis value generation unit;

[0041] The biased intervention feature set acquisition unit acquires the biased intervention feature set corresponding to each e-commerce store based on the product operation feature information set of each e-commerce store in the user's portrait preference association set;

[0042] The biased intervention feature analysis value generation unit combines the historical after-sales information records of the corresponding user to generate a biased intervention feature analysis value corresponding to the biased intervention feature set of each e-commerce store based on the after-sales behavior portrait of the corresponding user.

[0043] Furthermore, the preference filtering management module includes a preference filtering optimization unit and a preference feedback management unit.

[0044] The preference filtering optimization unit extracts the biased intervention feature analysis value corresponding to each element in the corresponding profile preference association set based on the corresponding user's after-sales behavior profile, as well as the collaborative filtering risk value of the corresponding user based on the biased intervention feature set corresponding to the after-sales e-commerce store; and based on the obtained collaborative filtering risk value and the obtained biased intervention feature analysis value, filters and optimizes the profile preference association set corresponding to the corresponding user to generate a preference recommendation content set for the corresponding user;

[0045] The preference feedback management unit obtains the user preference recommendation content set generated by the preference filtering optimization unit and feeds it back to the e-commerce account display terminal of the corresponding user.

[0046] Compared with the prior art, the beneficial effects achieved by the present invention are: in the process of obtaining recommended products for users, the present invention not only takes into account the user's search keywords and product browsing information, but also takes into account the influence of the user's after-sales information on the product recommendation information, generates the user's portrait preference association set and the biased intervention feature set of each e-commerce store in the corresponding set, and realizes the dynamic quantification of the biased intervention feature analysis value corresponding to the biased intervention feature set of each e-commerce store for the user's after-sales behavior portrait; at the same time, according to the user's historical after-sales information records, dynamically generates the collaborative filtering risk value of the corresponding user based on the biased intervention feature set corresponding to the after-sales e-commerce store; thereby realizing the intelligent screening of the user's preference recommendation content set, and realizing the effective management of e-commerce information. BRIEF DESCRIPTION OF THE DRAWINGS

[0047] The accompanying drawings are used to provide a further understanding of the present invention and constitute a part of the specification. Together with the embodiments of the present invention, they are used to explain the present invention and do not constitute a limitation of the present invention. In the accompanying drawings:

[0048] Figure 1 It is a structural diagram of the e-commerce integrated management system based on big data analysis of the present invention;

[0049] Figure 2 It is a flow chart of the comprehensive e-commerce management method based on big data analysis of the present invention. DETAILED DESCRIPTION

[0050] The following will clearly and completely describe the technical solutions in the embodiments of the present invention in conjunction with the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative efforts are within the scope of protection of the present invention.

[0051] See also Figure 1 The present invention provides a technical solution: an e-commerce integrated management system based on big data analysis, the system includes the following modules:

[0052] A preference feature association analysis module, which obtains the user's user behavior records within a recent preset time period and generates the user's preferred content information; combines the characteristic keywords of the e-commerce store's corresponding products to obtain the profile matching value of each e-commerce store based on the user's behavior records; and records the set of e-commerce stores whose profile matching values ​​based on the user's behavior records are greater than a threshold as the corresponding user's profile preference association set;

[0053] An operation characteristic bias analysis module, comprising a bias intervention characteristic set acquisition unit and a bias intervention characteristic analysis value generation unit;

[0054] The biased intervention feature set acquisition unit acquires the biased intervention feature set corresponding to each e-commerce store based on the product operation feature information set of each e-commerce store in the user's portrait preference association set;

[0055] The biased intervention feature analysis value generation unit combines the historical after-sales information records of the corresponding user to generate a biased intervention feature analysis value corresponding to the biased intervention feature set of each e-commerce store based on the after-sales behavior portrait of the corresponding user.

[0056] A collaborative filtering risk analysis module, wherein the collaborative filtering risk analysis module generates a collaborative filtering risk value for the corresponding user based on a biased intervention feature set corresponding to the after-sales e-commerce store corresponding to the user's historical after-sales information record;

[0057] A preference filtering management module, which includes a preference filtering optimization unit and a preference feedback management unit.

[0058] The preference filtering optimization unit extracts the biased intervention feature analysis value corresponding to each element in the corresponding profile preference association set based on the corresponding user's after-sales behavior profile, as well as the collaborative filtering risk value of the corresponding user based on the biased intervention feature set corresponding to the after-sales e-commerce store; and based on the obtained collaborative filtering risk value and the obtained biased intervention feature analysis value, filters and optimizes the profile preference association set corresponding to the corresponding user to generate a preference recommendation content set for the corresponding user;

[0059] The preference feedback management unit obtains the user preference recommendation content set generated by the preference filtering optimization unit and feeds it back to the e-commerce account display terminal of the corresponding user.

[0060] like Figure 2 As shown, the e-commerce comprehensive management method based on big data analysis includes the following steps:

[0061] S1. Obtain the user's behavior records within a recent preset time period to generate the user's preferred content information; combine the characteristic keywords of the e-commerce store's corresponding products to obtain the profile matching value of each e-commerce store based on the user's behavior records; record the set of e-commerce stores whose profile matching values ​​based on the user's behavior records are greater than a threshold as the corresponding user's profile preference association set;

[0062] The user behavior record includes a keyword set consisting of keywords corresponding to the titles of the products browsed, clicked, and purchased by the user and the user's search keywords, as well as an array of product records consisting of the products browsed, clicked, and purchased by the user;

[0063] The user's preference content information includes one or more preference content data groups, each preference content data group corresponds to a product category, and the preference content data group is recorded as [PC, PF], where PC represents the product category corresponding to the corresponding preference content data group; PF represents the frequency of product records belonging to PC in the corresponding preference content data group in the product record array in the user behavior record;

[0064] The characteristic keywords of the products of the e-commerce store are the keywords set by the administrator of the corresponding e-commerce store according to the product characteristics;

[0065] When the portrait matching value of each e-commerce store based on the user behavior record is obtained, the portrait matching value of the i-th e-commerce store based on the user behavior record is recorded as Pm i ,

[0066]

[0067] Among them, GKB represents the keyword set in the user behavior record; GKF represents the set of characteristic keywords corresponding to the products of each element in the product record array in the user behavior record; both GKB and GKF are not empty sets; Gi represents the set of characteristic keywords corresponding to the products of the i-th e-commerce store; PF i represents the frequency of product records corresponding to the product category of the i-th e-commerce store in the user's preference content information; ZPF represents the sum of the frequency of corresponding product records in each preference content data group in the user's preference content information; count{} represents a function that counts the number of elements in a set; μ represents a preset normalization coefficient.

[0068] S2. Based on the product operation feature information set of each e-commerce store in the user's profile preference association set, obtain the biased intervention feature set corresponding to each e-commerce store; and combine the historical after-sales information records of the corresponding user to generate the biased intervention feature analysis value corresponding to the biased intervention feature set of each e-commerce store based on the corresponding user's after-sales behavior profile;

[0069] The e-commerce store corresponding to the nth element in the user's profile preference association set is denoted as SN n , SN n The corresponding commodity operation feature information set is recorded as RCC n ;

[0070] The RCC n It includes fraud information assessment value, transshipment information assessment curve and out-of-stock risk assessment value;

[0071] RCC n The fraud information evaluation value in is recorded as QZn, QZn=CBn+r·STn / SAn; r represents the preset weight coefficient; CBn represents SN n The percentage of negative reviews in the product reviews; SAn stands for SN n The number of times the product was involved in the complaint platform; STn represents SN n The number of successful user complaints in the ruling results of the corresponding complaints of the internal products;

[0072] Get RCC n When evaluating the transit information curve in the , obtain the corresponding e-commerce store SN in the historical data n The corresponding transshipment risk data pairs of the corresponding products at different transportation distances, wherein the first data in the transshipment risk data pair is the transportation distance of the corresponding product, and the second data is the proportion of orders with abnormal product damage in all orders whose transportation distance between the user order location and the corresponding product delivery warehouse is less than the corresponding first data; the SN-based n The transshipment risk data of the products are fitted with a curve, and the function corresponding to the fitted curve is used as the RCC n The transport information assessment curve in ;

[0073] In this embodiment, curve fitting is performed by using curve fitting tools, such as Python (NumPy, SciPy), origin, MATLAB, Igor Pro, and SciDAVis. The sigmoid function model is an "S"-shaped curve model.

[0074] Get RCC n When evaluating the out-of-stock risk value in n The out-of-stock risk assessment value in is recorded as DHn.

[0075] DHn=Vn / Cn·TPn,

[0076] Where Cn represents SN n The average inventory of the product at different times; Vn represents SN nThe average sales rate of the product at different times; TPn represents SN n The average value of the delivery delay deviation corresponding to each product order in the delivery delay deviation represents the interval between the courier pickup time of the logistics company to which the corresponding product order belongs and the time when the corresponding user submits the product order;

[0077] Get SN n The corresponding bias intervention feature set includes the fraud information assessment value, the transshipment information bias risk value and the out-of-stock risk assessment value; when obtaining the transshipment information bias risk value, obtain the distance between the delivery order location submitted by the corresponding user and the corresponding product shipping warehouse, recorded as L; substitute L into RCC n The corresponding function value in the transport information evaluation curve is used as SN n The corresponding biased risk value of the transport information in the biased intervention feature set.

[0078] Combined with the historical after-sales information records of the corresponding user, when generating the biased intervention feature analysis value corresponding to the biased intervention feature set of each e-commerce store, the biased intervention feature analysis value corresponding to the biased intervention feature set of the nth element in the user's portrait preference association set of the corresponding user's after-sales behavior portrait is recorded as BFAV n ,

[0079] BFAV n =Pm in / exp(r1·QZn+r2·ZYn+r3·DHn),

[0080] Among them, QZn represents SN n The corresponding bias intervention feature set fraud information evaluation value; ZYn represents SN n The corresponding bias risk value of the transfer information in the bias intervention feature set; DHn represents SN n The corresponding out-of-stock risk assessment value in the biased intervention feature set; Pm in Indicates SN n Profile matching value based on user behavior records;

[0081] r1 represents the proportion of orders with fraud information assessment values ​​greater than or equal to QZn in the corresponding biased intervention feature set in the historical after-sales information records; r2 represents the proportion of orders with transit information bias risk values ​​greater than or equal to ZYn in the corresponding biased intervention feature set in the historical after-sales information records; r3 represents the proportion of orders with out-of-stock risk assessment values ​​greater than or equal to DHn in the corresponding biased intervention feature set in the historical after-sales information records.

[0082] S3. Combine the historical after-sales information records of the corresponding user to generate a collaborative filtering risk value for the corresponding user based on the biased intervention feature set corresponding to the after-sales e-commerce store;

[0083] When generating the collaborative filtering risk value of the corresponding user based on the biased intervention feature set of the after-sales e-commerce store,

[0084] Obtain the biased intervention feature set corresponding to each after-sales order in the historical after-sales information record of the corresponding user and the portrait matching value of the e-commerce store to which the corresponding order belongs based on the user behavior record at the corresponding after-sales time point, and combine the calculation formula of the biased intervention feature analysis value corresponding to the biased intervention feature set of the nth element in the user's portrait preference association set based on the after-sales behavior portrait of the corresponding user to generate the biased intervention feature analysis value corresponding to each after-sales order in the historical after-sales information record of the corresponding user, and record the average value of the biased intervention feature analysis value corresponding to each after-sales order in the historical after-sales information record of the corresponding user as the collaborative filtering risk value of the corresponding user based on the biased intervention feature set corresponding to the after-sales e-commerce store.

[0085] In this embodiment, the collaborative filtering risk value of the corresponding user based on the biased intervention feature set corresponding to the after-sales e-commerce store is obtained according to the historical after-sales information records of the corresponding user. The obtained collaborative filtering risk value will change dynamically with the changes in the historical after-sales information records of the user. The obtained collaborative filtering risk value can be used as a reference standard for filtering and optimizing the user's portrait preference association set in subsequent steps, and the elements in the corresponding portrait preference association set whose corresponding preference intervention feature analysis value is greater than the collaborative filtering risk value are screened. The set composed of the obtained screening results is the preference recommendation content set of the corresponding user, thereby realizing the management of recommended products for the corresponding user at the current time.

[0086] S4. Extract the biased intervention feature analysis value corresponding to each element in the corresponding profile preference association set based on the corresponding user's after-sales behavior profile, as well as the collaborative filtering risk value of the corresponding user based on the biased intervention feature set corresponding to the after-sales e-commerce store; and filter and optimize the profile preference association set corresponding to the corresponding user based on the obtained collaborative filtering risk value and the obtained biased intervention feature analysis value, generate the corresponding user's preferred content set, and feed it back to the corresponding user's e-commerce account display terminal;

[0087] In the process of filtering and optimizing the corresponding user's profile preference association set based on the obtained collaborative filtering risk value and the obtained bias intervention feature analysis value to generate the corresponding user's preference recommendation content set,

[0088] The remaining set after removing the elements whose corresponding preference intervention feature analysis value is less than or equal to the collaborative filtering risk value in the corresponding portrait preference association set is used as the preference recommendation content set of the corresponding user.

[0089] It should be noted that, in this document, relational terms such as first and second, etc., are used only to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply any actual relationship or order between these entities or operations. Moreover, the terms "comprises," "comprising," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that includes a list of elements includes not only those elements but also other elements not explicitly listed, or elements inherent to such process, method, article, or apparatus.

[0090] Finally, it should be noted that the above descriptions are merely preferred embodiments of the present invention and are not intended to limit the present invention. Although the present invention has been described in detail with reference to the aforementioned embodiments, those skilled in the art will be able to modify the technical solutions described in the aforementioned embodiments or substitute equivalents for some of the technical features. Any modifications, equivalent substitutions, and improvements made within the spirit and principles of the present invention shall be included within the scope of protection of the present invention.

Claims

1. An integrated e-commerce management method based on big data analysis, characterized in that: The method comprises the following steps: S1. Obtain the user behavior record of the user in the most recent preset time period to generate the user's preferred content information; combine the characteristic keywords of the corresponding products of the e-commerce store to obtain the portrait matching value of each e-commerce store based on the user behavior record; record the set of e-commerce stores whose portrait matching value based on the user behavior record is greater than the threshold as the portrait preference association set of the corresponding user; S2. Based on the product operation feature information set of each e-commerce store in the user's portrait preference association set, obtain the biased intervention feature set corresponding to each e-commerce store; and combine the historical after-sales information records of the corresponding user to generate the biased intervention feature analysis value corresponding to the biased intervention feature set of each e-commerce store based on the after-sales behavior portrait of the corresponding user; S3. Combine the historical after-sales information records of the corresponding user to generate a collaborative filtering risk value of the corresponding user based on the biased intervention feature set of the after-sales e-commerce store; S4. Extract the biased intervention feature analysis value corresponding to each element in the corresponding portrait preference association set based on the after-sales behavior portrait of the corresponding user, and the collaborative filtering risk value of the corresponding user based on the biased intervention feature set corresponding to the after-sales e-commerce store; and filter and optimize the portrait preference association set corresponding to the corresponding user based on the obtained collaborative filtering risk value and the obtained biased intervention feature analysis value, generate the preference recommendation content set of the corresponding user, and feed it back to the e-commerce account display terminal of the corresponding user.

2. The e-commerce comprehensive management method based on big data analysis according to claim 1 is characterized by: The user behavior record includes a keyword set consisting of keywords corresponding to titles of commodities browsed, clicked, and purchased by the user and the user's search keywords, and a commodity record array consisting of commodities browsed, clicked, and purchased by the user; The user's preference content information includes one or more preference content data groups, each preference content data group corresponds to a commodity category, and the preference content data group is recorded as [PC, PF], wherein PC represents the commodity category corresponding to the corresponding preference content data group; PF represents the frequency of commodity records belonging to PC in the corresponding preference content data group in the commodity record array in the user behavior record; The characteristic keywords of the products corresponding to the e-commerce stores are the keywords set by the administrator of the corresponding e-commerce stores for the product characteristics; When the portrait matching value of each e-commerce store based on the user behavior record is obtained, the portrait matching value of the i-th e-commerce store based on the user behavior record is recorded as Pm i , Wherein, GKB represents the keyword set in the user behavior record; GKF represents the set of characteristic keywords corresponding to the products of each element in the commodity record array in the user behavior record; both GKB and GKF are not empty sets; Gi represents the set of characteristic keywords corresponding to the products of the i-th e-commerce store; PF i It represents the frequency of product records corresponding to the product category of the corresponding product of the i-th e-commerce store in the user's preference content information; ZPF represents the sum of the corresponding product record frequencies in each preference content data group in the user's preference content information; count{} represents a function that counts the number of elements in a set; μ represents a preset normalization coefficient.

3. The e-commerce comprehensive management method based on big data analysis according to claim 1 is characterized by: The e-commerce store corresponding to the nth element in the user's profile preference association set is denoted as SN n , SN n The corresponding commodity operation characteristic information set is recorded as RCC n ; The RCC n It includes fraud information assessment value, transshipment information assessment curve and out-of-stock risk assessment value; RCC n The fraud information evaluation value in is recorded as QZn, QZn = CBn + r·STn / SAn; r represents the preset weight coefficient; CBn represents S n The percentage of negative reviews in the product reviews; SAn represents SN n The number of times the product was involved in the complaint platform; STn means SN n The number of successful user complaints in the ruling results of the relevant complaints of the internal products; Get RCC n When evaluating the transit information curve in the , obtain the corresponding e-commerce store SN in the historical data n The corresponding transshipment risk data pairs of the corresponding products at different transportation distances, wherein the first data in the transshipment risk data pairs is the transportation distance of the corresponding product, and the second data is the proportion of orders with abnormal product damage in all orders whose transportation distance between the user order location and the corresponding product delivery warehouse is less than the corresponding first data; the SN-based n The transshipment risk data of the products in the experiment are fitted with curves, and the corresponding function of the fitted curve is used as the RCC n The transport information assessment curve in ; Get RCC n When evaluating the out-of-stock risk in n The out-of-stock risk assessment value in is denoted as DHn. DHn=Vn / Cn·TPn, Where Cn represents SN n The average inventory of the product at different times; Vn represents SN n The average sales rate of the product at different times; TPn represents SN n The average value of the delivery delay deviations corresponding to each product order in; the delivery delay deviation indicates the interval between the courier pickup time of the logistics to which the corresponding product order belongs and the time when the corresponding user submits the product order; Get SN n The corresponding bias intervention feature set includes the fraud information assessment value, the transshipment information bias risk value and the out-of-stock risk assessment value; when obtaining the transshipment information bias risk value, obtain the distance between the delivery order location submitted by the corresponding user and the corresponding product delivery warehouse, denoted as L; substitute L into RCC n The corresponding function value in the transport information evaluation curve is used as SN n The corresponding biased risk value of the transport information in the biased intervention feature set.

4. The e-commerce integrated management method based on big data analysis according to claim 3 is characterized by: When generating the biased intervention feature analysis value corresponding to the biased intervention feature set of each e-commerce store based on the historical after-sales information record of the corresponding user, the biased intervention feature analysis value corresponding to the biased intervention feature set of the nth element in the user's portrait preference association set of the corresponding user's after-sales behavior portrait is recorded as BFAV n , <h2 style=";text-align:left;direction:ltr">BFAV<h2 style=";text-align:left;direction:ltr"> n <h2 style=";text-align:left;direction:ltr"> =Pm<h2 style=";text-align:left;direction:ltr"> in <h2 style=";text-align:left;direction:ltr"> / exp(r1·QZn+r2·ZYn+r3·DHn), Among them, QZn represents SN n The corresponding biased intervention feature set fraud information evaluation value; ZYn represents SN n The corresponding bias risk value of the transport information in the bias intervention feature set; DHn represents SN n The corresponding out-of-stock risk assessment value in the biased intervention feature set; Pm in Indicates SN n Portrait matching value based on user behavior records; r1 represents the proportion of orders with fraud information assessment values ​​greater than or equal to QZn in the corresponding biased intervention feature set in the historical after-sales information records; r2 represents the proportion of orders with transit information bias risk values ​​greater than or equal to ZYn in the corresponding biased intervention feature set in the historical after-sales information records; r3 represents the proportion of orders with out-of-stock risk assessment values ​​greater than or equal to DHn in the corresponding biased intervention feature set in the historical after-sales information records.

5. The e-commerce integrated management method based on big data analysis according to claim 4 is characterized by: When generating the collaborative filtering risk value of the corresponding user based on the biased intervention feature set of the after-sales e-commerce store, Obtain the biased intervention feature set corresponding to each after-sales order in the historical after-sales information record of the corresponding user and the portrait matching value of the e-commerce store to which the corresponding order belongs based on the user behavior record at the corresponding after-sales time point, and combine the calculation formula of the biased intervention feature analysis value corresponding to the biased intervention feature set of the nth element in the user's portrait preference association set based on the after-sales behavior portrait of the corresponding user to generate the biased intervention feature analysis value corresponding to each after-sales order in the historical after-sales information record of the corresponding user, and record the average value of the biased intervention feature analysis value corresponding to each after-sales order in the historical after-sales information record of the corresponding user as the collaborative filtering risk value of the corresponding user based on the biased intervention feature set corresponding to the after-sales e-commerce store.

6. The e-commerce integrated management method based on big data analysis according to claim 5 is characterized by: In the process of filtering and optimizing the corresponding user's portrait preference association set based on the obtained collaborative filtering risk value and the obtained biased intervention feature analysis value to generate the corresponding user's preference recommendation content set, The remaining set after removing the elements whose corresponding preference intervention feature analysis values ​​in the corresponding portrait preference association set are less than or equal to the collaborative filtering risk value is used as the preference recommendation content set for the corresponding user.

7. An integrated e-commerce management system based on big data analysis, characterized in that: The system includes the following modules: A preference feature association analysis module, which obtains the user's behavior records in the most recent preset time period and generates the user's preference content information; combines the characteristic keywords of the corresponding products of the e-commerce store to obtain the portrait matching value of each e-commerce store based on the user's behavior records; Record the set of e-commerce stores whose profile matching values ​​based on user behavior records are greater than a threshold as the profile preference association set of the corresponding user; An operation feature bias analysis module, which obtains a bias intervention feature set corresponding to each e-commerce store based on the product operation feature information set of each e-commerce store in the user's portrait preference association set; and generates a bias intervention feature analysis value corresponding to the bias intervention feature set of each e-commerce store based on the corresponding user's after-sales behavior portrait in combination with the historical after-sales information record of the corresponding user; A collaborative filtering risk analysis module, wherein the collaborative filtering risk analysis module generates a collaborative filtering risk value of the corresponding user based on a biased intervention feature set corresponding to the after-sales e-commerce store corresponding to the historical after-sales information record of the user; A preference filtering management module, which extracts the biased intervention feature analysis value corresponding to each element in the corresponding profile preference association set of the corresponding user's after-sales behavior portrait, and the collaborative filtering risk value of the corresponding user based on the biased intervention feature set corresponding to the after-sales e-commerce store; Based on the obtained collaborative filtering risk value and the obtained bias intervention feature analysis value, the portrait preference association set corresponding to the corresponding user is filtered and optimized to generate the preference recommendation content set of the corresponding user, and feed it back to the e-commerce account display terminal of the corresponding user.

8. The e-commerce integrated management system based on big data analysis according to claim 7 is characterized by: The operation characteristic bias analysis module includes a bias intervention characteristic set acquisition unit and a bias intervention characteristic analysis value generation unit; The biased intervention feature set acquisition unit acquires the biased intervention feature set corresponding to each e-commerce store based on the product operation feature information set of each e-commerce store in the user's portrait preference association set; The biased intervention feature analysis value generating unit generates a biased intervention feature analysis value corresponding to the biased intervention feature set of each e-commerce store based on the after-sales behavior portrait of the corresponding user in combination with the historical after-sales information records of the corresponding user.

9. The e-commerce integrated management system based on big data analysis according to claim 7 is characterized by: The preference filtering management module includes a preference filtering optimization unit and a preference feedback management unit. The preference filtering optimization unit extracts the biased intervention feature analysis value corresponding to each element in the corresponding profile preference association set based on the after-sales behavior portrait of the corresponding user, and the collaborative filtering risk value of the corresponding user based on the biased intervention feature set corresponding to the after-sales e-commerce store; Based on the obtained collaborative filtering risk value and the obtained biased intervention feature analysis value, the corresponding profile preference association set of the corresponding user is filtered and optimized to generate the preference recommendation content set of the corresponding user; The preference feedback management unit obtains the user preference recommendation content set generated by the preference filtering optimization unit, and feeds it back to the e-commerce account display terminal of the corresponding user.

Images