Cross-border e-commerce customer obtaining method and system based on intelligent retrieval

Abstract

The invention provides a cross-border e-commerce customer obtaining method and system based on intelligent retrieval, the method processes and analyzes user data based on a natural semantic recognition method, and the method comprises the following steps: intelligently retrieving enterprise target customers by adopting a search end, and performing related information interaction according to different search tags; according to the interaction result, carrying out data association merging on the search results of the plurality of search tags; performing data analysis after association and combination, and performing user activeness ranking; and according to a sorting result in the step S3, advertisement pushing with corresponding frequency and commodity selling in corresponding time periods are carried out on the users with different activeness. According to the method, the purchase desire of the user can be evaluated according to different target classification words of the user, promotion time, search keywords and selling time of different commodities are set according to shopping habits of the user, the probability of obtaining customers is increased, the customers purchasing for the first time are easier to obtain, the repeated purchase rate of old customers is increased, and the user experience is improved. And the flow casting cost is reduced.

Description

Technical Field

[0001] The present invention belongs to the technical field of e-commerce, and specifically relates to a cross-border e-commerce customer acquisition method and system based on intelligent retrieval. Background Art

[0002] With the development of Internet technology, especially the popularity of smart phones and mobile Internet, the development of cross-border e-commerce has been greatly promoted. Consumers can easily purchase global goods through a variety of online channels. As the technical basis for promoting economic integration and trade globalization, cross-border e-commerce has very important strategic significance. Cross-border e-commerce not only breaks through the barriers between countries and makes international trade borderless, but it is also causing great changes in the world's economic trade. As a kind of e-commerce trade, cross-border e-commerce needs to constantly attract consumers from new regions. Through effective order acquisition strategies, companies can increase the audience range, enter new markets, expand market coverage, and enhance global competitiveness. Obtaining orders is the key to increasing the revenue of cross-border e-commerce companies. By expanding the customer audience, cross-border e-commerce companies can achieve sustained growth in sales, promote business expansion, corporate visibility and influence, and gradually establish trust in the hearts of global consumers and form a customer base that they value.

[0003] Therefore, it is necessary to understand customer behavior and needs, optimize marketing strategies and improve conversion rates by analyzing customer data from different search terminals. However, in the prior art, the Chinese patent application document with application number CN202311202127.7 and the Chinese patent with application number CN202110835352.9 are both technical contents related to the management and operation of cross-border e-commerce and related systems. The prior art lacks the acquisition of purchase data information after searching for the same item with different search tags (keywords) through multiple clients or search terminals, and then determining the customer's activity and commonly used search tags and the difference in season or region of purchase of the same product, and finally controlling the advertising push and sales time of different items in cross-border e-commerce through the embodiment of activity, so as to achieve the technical effect of increasing the customer acquisition channels and scope of cross-border e-commerce. Summary of the invention

[0004] In view of the above-mentioned defects, the present invention provides a cross-border e-commerce customer acquisition method and system, device and medium based on intelligent retrieval. The present invention retrieves target customers of enterprises through different search tags of multiple search terminals, and uses the regions where multiple users in the new database after data matching and data association are located to traverse all user information documents in the new associated database C in a forward and reverse manner, and obtains the word frequencies of different document classification words respectively, and finally constructs a user activity evaluation index value based on word frequency, and then sorts the user activity according to the binary tree sorting method, and finally obtains the user activity ranking when different target classification words are used for classification, and then evaluates the user's desire to buy according to the different target classification words of the user, and then sets the promotion time, search keywords and sales time of different products according to the user's shopping habits, thereby increasing the probability of acquiring customers while making it easier to acquire first-time buyers, and increasing the repeat purchase rate of old customers, reducing the cost of investment.

[0005] The present invention provides the following technical solution: a cross-border e-commerce customer acquisition method based on intelligent retrieval, wherein the method processes and analyzes user data based on a natural semantic recognition method, and the method comprises the following steps:

[0006] S1: Use search-end intelligent retrieval of target customers and interact with relevant information based on different search tags;

[0007] S2: According to the interaction results, data association and merging of search results of multiple search tags are performed;

[0008] S3: Analyze the associated and merged data and sort the user activity;

[0009] S4: According to the ranking result of step S3, push advertisements with corresponding frequencies and sell goods in corresponding time periods to users with different activity levels.

[0010] Furthermore, in the step S1, the search end is used to intelligently search for target customers of the enterprise, and relevant information is interacted according to different search tags, including the following steps:

[0011] S11: Use the crawler method to obtain the data information of each user in the k-th search engine: the e-th user data subset U k,e : in is the registration ID information of the e-th user, The date the user purchased the product. The price of the product purchased by the user. The place where the goods purchased by the user are shipped. Membership record for users k=1,2,…,K; K is the total number of search engines; e=1,2,…,E; E is the number of user data subsets formed by the search results of each search engine; the search results of K search engines form the collection information set U={U 1,1 ,U 1,2 ,…,U k,e ,…,U K,E};

[0012] S12: Using natural semantic recognition method to identify the Hamming distance H of multiple user data subsets in K search results of K search engines:

[0013]

[0014] Among them, r∈{1,2,…,K}, h∈{1,2,…,E}; g rh =1,2,...,5; g ke =1,2,…,5; is the gth user in the hth user data subset in the rth search result rh User information The length of the character segment; is the gth user data subset in the eth user data subset in the kth search result ke User information The length of the character segment;

[0015] S13: Calculate the gth user data subset in the eth user data subset in the kth search result ke User information The label dimension and the gth user data subset in the hth user data subset in the rth search result rh User information The label dimension

[0016] S14: Calculate the interactive correlation information value of the mth search tag in multiple user data subsets in K search results

[0017]

[0018] Then, we obtain a set I of interactive correlation information values ​​of multiple search tags in multiple user data subsets in K databases:

[0019] Furthermore, the step S13 calculates the g-th user data subset in the e-th user data subset in the k-th search result. ke User information The label dimension The formula is as follows:

[0020]

[0021] Calculate the gth user data subset in the hth user data subset in the rth search result rh User information The label dimension The formula is as follows:

[0022]

[0023] Among them, la m is the mth search tag, m = 1, 2, …, M; k ≠ ​​r, e ≠ h.

[0024] Furthermore, the step S2 includes the following steps:

[0025] S21: Construct a search tag ranking model Y according to the interactive correlation information value set calculated in step S14:

[0026]

[0027] Where A is the linear equation formed The independent variable coefficient, B is the constant of the linear equation, p is a prime number, p=7; mod is the remainder function;

[0028] S22: Take the top five interactive correlation information values ​​in the results of the search tag ranking model Y from the largest to the smallest as the data integration keywords, and match and associate any user data subsets in every two search engines in the K search engines with b n As the nth data association label, n = 1, 2, 3, 4, 5; data association is combined to obtain As the bth user of the qth user in the new associated database C n User information; C = {c1, c2, ..., c q ,…,c Q}, q = 1, 2, ... Q; c q is the qth user information subset in the new associated database C, Q≤R×H; b1 to b5 all belong to {1,2,…,M} and are not equal to each other.

[0029] Furthermore, the S3 step includes the following steps:

[0030] S31: forwardly traverse all user information documents in the new associated database C, obtain the forward occurrence number J1 of the target classification word t in each subset, and calculate the forward word frequency TF(t,C) of the target classification word t:

[0031]

[0032] in, To calculate the new associated database C from the first user c1 to the Qth user c Q The number of all user information documents in; t = 1, 2, ..., T; T is the total number of target classification words;

[0033] S32: Reversely traverse all user information documents in the new associated database C, calculate the reverse occurrence number J2 of the target classification word t in each subset, and calculate the reverse document word frequency NTF(t,C) of the target classification word t:

[0034]

[0035] S33: Construct the calculation formula of the user activity evaluation index value Act(t,C) based on word frequency:

[0036]

[0037] S34: According to the user activity evaluation index value, all user information in the new associated database is traversed to sort the activity, and the advertising time, target sales area and listing time of products with different prices are selected according to the level of activity.

[0038] Furthermore, the step S34 includes the following steps:

[0039] S341: Take the user activity evaluation index value Act(t,C) of the target classification word t as a node, and compare it with the user activity evaluation index values ​​of any other two target classification words, take the user activity evaluation index value of the target classification word less than Act(t,C) as the left subtree node of the node, and take the user activity evaluation index value of the target classification word greater than Act(t,C) as the right subtree of the node, to form a binary tree;

[0040] S342: traverse all user information documents in the new associated database C, and gradually insert the user activity evaluation index value Act(t,C) of the target classification word t as a node into the binary tree constructed according to the rule of S341;

[0041] S343: repeat the steps S341-S342;

[0042] S344: Calculating Binary Tree Complexity

[0043] S345: Determine whether the binary tree complexity O(Act) is less than 0.35T. If so, form a sorted table of user activity evaluation index values ​​of different target classification words output from small to large; otherwise, repeat S341-S344.

[0044] The present invention also provides a cross-border e-commerce customer acquisition system based on intelligent retrieval using the above method, the system comprising an information search interaction module, a data association module, an activity ranking module and an advertisement push control module;

[0045] The information search interaction module is used to intelligently search for target customers of an enterprise using a search terminal, and to interact with relevant information according to different search tags;

[0046] The data association module is used to perform data association merging of search results of multiple search tags according to the interaction results;

[0047] The activity ranking module performs data analysis after association and merging, and performs user activity ranking;

[0048] The advertisement push control module is used to push advertisements with corresponding frequencies and sell goods in corresponding time periods to users with different activity levels according to the sorting results of the activity sorting module.

[0049] The beneficial effects of the present invention are:

[0050] 1. The cross-border e-commerce customer acquisition method based on intelligent retrieval provided by the present invention uses a crawler method to assist a natural semantic recognition method to obtain a user data subset U of each user in the search results of multiple search engines. k,e , forming a collection information set U, and then further identifying the Hamming distance H of multiple user data subsets in the K search results, calculating the label dimension of user information in the search results of multiple search engines, and using the interactive correlation information value of different search tags in multiple user data subsets in the search results of multiple search engines For each user data in the search results of multiple search engines, k,e The data is then simplified to facilitate efficient merging of relevant user information in the search results of multiple search engines, which is needed for subsequent data association merging.

[0051] 2. The present invention further calculates the interactive correlation information values ​​of the M search tags in the multiple user data subsets in the K search results, and forms an interactive correlation information value set I: This lays the foundation for building subsequent efficient data association and merging.

[0052] 3. The present invention constructs a search tag ranking model Then, the interactive correlation information value of the first five largest values ​​of the calculation results of the model is taken. and As the index for data search, the user data subset of each user in multiple search results is restructured and redundant data is eliminated, so as to realize data interactive association and merging, and obtain a new simplified database C. The new simplified database C is effectively organized and convenient for searching users' different preference search tags, peak purchase periods of different commodities, purchase rates of the commodity in different regions, etc., so as to facilitate the unified sorting and search of subsequent data.

[0053] 4. The present invention uses different target classification words to perform forward traversal and reverse traversal in sequence on the new simplified database C, and then further calculates the user activity evaluation index value. It can judge the user's activity for the product according to the size of the user's activity. Therefore, it further controls the corresponding frequency of advertising push and the sale of goods in the corresponding time period, which can increase the chance of acquiring customers, expand first-time purchasing customers, effectively improve the repeat purchase rate of old customers, and reduce the cost of investment. BRIEF DESCRIPTION OF THE DRAWINGS

[0054] The present invention will be described in more detail below based on embodiments and with reference to the accompanying drawings, wherein:

[0055] Figure 1 A schematic diagram of the process flow of the cross-border e-commerce customer acquisition method based on intelligent retrieval provided by the present invention;

[0056] Figure 2 A bar chart comparing the accuracy of the method provided by the present invention and the method of omitting step S2 and only using natural semantic recognition to perform data association tagging on search results of different search engines and then sorting the user activity;

[0057] Figure 3 A comparative line chart showing different methods for predicting user activity rating values;

[0058] Figure 4 A schematic diagram of the structure of a cross-border e-commerce customer acquisition system based on intelligent retrieval provided by the present invention; DETAILED DESCRIPTION

[0059] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. 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 creative work are within the scope of protection of the present invention.

[0060] like Figure 1As shown, it is a flow chart of the cross-border e-commerce customer acquisition method based on intelligent retrieval provided by the present invention. 1. The cross-border e-commerce customer acquisition method based on intelligent retrieval, the method processes and analyzes user data based on a natural semantic recognition method, and is characterized in that the method includes the following steps:

[0061] S1: Use search-end intelligent retrieval of target customers and interact with relevant information based on different search tags;

[0062] S2: According to the interaction results, data association and merging of search results of multiple search tags are performed;

[0063] S3: Analyze the associated and merged data and sort the user activity;

[0064] S4: According to the sorting results of step S3, push advertisements with corresponding frequencies and sell goods in corresponding time periods to users with different activity levels.

[0065] Different search engines have different emphases and engine preferences for search results generated by the same search tag, and the types of content that are limitedly displayed during the content crawling process are also different. Some focus more on localized information, while others focus more on global websites. This setting will cause information to be highlighted or omitted. In addition, different search engines update their cloud content at different speeds and frequencies. Especially in the field of cross-border e-commerce, the designed product information, prices, policies and other content change quickly. Therefore, as a preferred embodiment of the present invention, it is necessary to interact with relevant information on the results of corporate target customers intelligently retrieved by multiple search terminals with different search tags in turn. In step S1, the search terminal intelligently retrieves corporate target customers, and interacts with relevant information according to different search tags, including the following steps:

[0066] S11: Use a crawler method to obtain the data information of each user in the k-th search engine: the e-th user data subset U k,e : in is the registration ID information of the e-th user, The date the user purchased the product. The price of the product purchased by the user. The place where the goods purchased by the user are shipped. Membership record for users k=1,2,…,K; K is the total number of search engines; e=1,2,…,E; E is the number of user data subsets formed by the search results of each search engine; the search results of K search engines form the collection information set U={U 1,1 ,U 1,2 ,…,U k,e ,…,U K,E}; Each search engine forms a search result after the search is completed, that is, the kth search result is formed after the kth search engine is completed;

[0067] S12: Using natural semantic recognition method to identify the Hamming distance H of multiple user data subsets in K search results of K search engines:

[0068]

[0069] Among them, r∈{1,2,…,K}, h∈{1,2,…,E}; g rh =1,2,...,5; g ke =1,2,…,5; is the gth user in the hth user data subset in the rth search result rh User information The length of the character segment; is the gth user data subset in the eth user data subset in the kth search result ke User information The length of the character segment;

[0070] S13: Calculate the gth user data subset in the eth user data subset in the kth search result ke User information The label dimension and the gth user data subset in the hth user data subset in the rth search result rh User information The label dimension

[0071] S14: Calculate the interactive correlation information value of the mth search tag in multiple user data subsets in K search results

[0072]

[0073] Then, we obtain a set I of interactive correlation information values ​​of multiple search tags in multiple user data subsets in K databases:

[0074] Step S13 calculates the gth user data in the eth user data subset in the kth search result. ke User information The label dimension The formula is as follows:

[0075]

[0076] Calculate the gth user data subset in the hth user data subset in the rth search result rh User information The label dimension The formula is as follows:

[0077]

[0078] Among them, la m is the mth search tag, m = 1, 2, …, M; k ≠ ​​r, e ≠ h.

[0079] The S2 step includes the following steps:

[0080] S21: Based on the interactive correlation information value set calculated in step S14, a search tag ranking model Y is constructed:

[0081]

[0082] Where A is the linear equation formed The independent variable coefficient, B is the constant of the linear equation, p is a prime number, p=7; mod is the remainder function;

[0083] S22: Take the top five interactive correlation information values ​​in the results of the search tag ranking model Y from the largest to the smallest as the data integration keywords, and match and associate any user data subsets in every two search engines in the K search engines, that is, associate and merge the data of the e-th user data subset in the k-th search result with the h-th user data subset in the r-th database, and use b n As the nth data association label, n = 1, 2, 3, 4, 5, the gth user data subset in the kth search result is rh User registration information and the gth user data subset in the hth database rh User registration information The data association and merging are obtained As the bth user of the qth user in the new associated database C n User information; C = {c1, c2, ..., c q ,…,c Q}, q = 1, 2, ... Q; c q is the qth user information subset in the new associated database C, Q≤R×H; b1 to b5 all belong to {1,2,…,M} and are not equal to each other.

[0084] like Figure 2 As shown in the figure, the red column represents the accuracy of further sorting the user activity of the new association database C obtained by the method of the present invention; the blue column represents the user information obtained by not using the association merging of each user data subset after omitting steps S21-S22 and not further sorting the search tags. Perform label setting, directly using the interactive correlation information values ​​in multiple user data subsets obtained in step S14 The green column is the accuracy of sorting user activity by using the first five search terms in the crawler information obtained by natural semantic recognition as search labels, performing data association, and obtaining a new associated database. Figure 2 It can be seen that the method provided by the present invention can more accurately sort user activity. After matching and associating different user data subsets through the search results of multiple search engines and obtaining data association tags, a new association database can obtain more comprehensive and diverse information, ensuring that the information obtained is more extensive and authoritative. At the same time, it avoids the simplification or bias of information, prevents information from being over-filtered, makes the results more objective and diverse, obtains update dynamics more comprehensively, thereby improving the accuracy and timeliness of information, and can help sellers better adapt to the local needs and trends of various markets, thereby optimizing marketing strategies.

[0085] Furthermore, step S3 includes the following steps:

[0086] S31: Forward traverse all user information documents in the new associated database C, obtain the forward occurrence number J1 of the target classification word t in each subset, and calculate the forward word frequency TF(t,C) of the target classification word t:

[0087]

[0088] in, To calculate the new associated database C from the first user c1 to the Qth user c Q The number of all user information documents in; t = 1, 2, ..., T; T is the total number of target classification words;

[0089] S32: Reversely traverse all user information documents in the new associated database C, calculate the reverse occurrence number J2 of the target classification word t in each subset, and calculate the reverse document word frequency NTF(T,C) of the target classification word t (that is, the frequency of the target classification word t appearing when reversely traversing the user information documents).

[0090]

[0091] S33: Construct the calculation formula of the user activity evaluation index value Act(t,C) based on word frequency:

[0092]

[0093] S34: According to the user activity evaluation index value, all user information in the new associated database is traversed to sort the activity, and the advertising time, target sales area and listing time of products with different prices are selected according to the level of activity.

[0094] After obtaining the user activity evaluation index value based on word frequency calculation, we can determine the influence of different user registration addresses, different seasons or time periods of each year, different product pricing, and different shipping locations in the corpus of K search engine search results on the intensity of user purchasing desire based on the size of user activity. When different product pricing, target area promotion, product shipping location, and shelf sales period are selectively set for the products to be sold, users will be more active in the product and have a stronger desire to buy it. Therefore, it can increase the chance of acquiring customers, make it easier to acquire first-time buyers, increase the repeat purchase rate of old customers, and reduce the cost of investment.

[0095] Step S34 includes the following steps:

[0096] S341: Take the user activity evaluation index value Act(t,C) of the target classification word t as a node, and compare it with the user activity evaluation index values ​​of any other two target classification words, take the user activity evaluation index value of the target classification word less than Act(t,C) as the left subtree node of the node, and take the user activity evaluation index value of the target classification word greater than Act(t,C) as the right subtree of the node, to form a binary tree;

[0097] S342: traverse all user information documents in the new associated database C, and gradually insert the user activity evaluation index value Act(t,C) of the target classification word t as a node into the binary tree constructed according to the rules of S341;

[0098] S343: repeat steps S341-S342;

[0099] S344: Calculating Binary Tree Complexity

[0100] S345: Determine whether the binary tree complexity O(Act) is less than 0.35T. If so, form a sorted table of user activity evaluation index values ​​of different target classification words output from small to large; otherwise, repeat S341-S344.

[0101] By traversing the new associated database forward and backward, we can obtain the forward word frequency TF(t,C) and the reverse word frequency NTF(t,C) respectively, which can effectively determine the importance of the target classification word t in the subset of the associated database C from the context, and then further calculate the user activity evaluation index value Act(t,C) by coupling the two to more accurately analyze the data in the new associated database, avoiding the influence of the target classification words that appear frequently in the entire new associated database C but have no special importance to the user activity evaluation on its calculation weight, effectively improving the accuracy and robustness of the user activity calculation. Figure 3 As shown, the user activity evaluation value is predicted by the method provided by the present invention and the CNN and GA methods respectively. It can be seen that the method provided by the present invention is closer to the user activity evaluation value and the prediction result is more accurate.

[0102] The present invention also provides a cross-border e-commerce customer acquisition system based on intelligent retrieval using the above method, such as Figure 4 As shown, the system provided by the present invention includes an information search interaction module, a data association module, an activity ranking module and an advertisement push control module;

[0103] The information search interaction module is used to intelligently search for target customers of enterprises through the search end and interact with relevant information according to different search tags;

[0104] A data association module, used to associate and merge the search results of multiple search tags according to the interaction results;

[0105] The activity ranking module analyzes the associated and merged data and ranks the user activity;

[0106] The advertisement push control module is used to push advertisements of corresponding frequencies and sell goods in corresponding time periods to users with different activity levels according to the sorting results of the activity sorting module.

[0107] The cross-border e-commerce customer acquisition method based on intelligent retrieval provided in this application may be in the form of a computer program product implemented on one or more storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing program code. Machine-readable storage media include permanent and non-permanent, removable and non-removable media, and information storage can be implemented by any method or technology. Information can be computer-readable instructions, data structures, modules of programs, or other data. Examples of machine-readable storage media include but are not limited to: phase change memory (PRAM), static random access memory (SRAM), dynamic random access memory (DRAM), other types of random access memory (RAM), read-only memory (ROM), electrically erasable programmable read-only memory (EEPROM), flash memory or other memory technology, read-only compact disk read-only memory (CD-ROM), digital versatile disc (DVD) or other optical storage, magnetic cassettes, magnetic tape magnetic disk storage or other magnetic storage devices or any other non-transmission media that can be used to store information that can be accessed by a computing device.

[0108] It should be noted that the serial numbers of the above-mentioned embodiments of the present invention are only for description and do not represent the advantages and disadvantages of the embodiments. In this article, relational terms such as "first" and "second" are only used to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply any such actual relationship or order between these entities or operations. And the terms "include", "comprise" or any other variants thereof in this article are intended to cover non-exclusive inclusion, so that a process, device, article or method including a series of elements includes not only those elements, but also other elements not explicitly listed, or also includes elements inherent to such process, device, article or method. In the absence of further restrictions, an element defined by the sentence "including a..." does not exclude the existence of other identical elements in the process, device, article or method including the element.

[0109] The above are only specific embodiments of the present disclosure, so that those skilled in the art can understand or implement the present disclosure. Various modifications to these embodiments will be apparent to those skilled in the art, and the general principles defined herein can be implemented in other embodiments without departing from the spirit or scope of the present disclosure. Therefore, the present disclosure will not be limited to these embodiments herein, but will conform to the widest scope consistent with the principles and novel features disclosed herein.

Claims

1. A cross-border e-commerce customer acquisition method based on intelligent retrieval, wherein the method processes and analyzes user data based on a natural semantic recognition method, and is characterized in that: The method comprises the following steps: S1: Use search-end intelligent retrieval of target customers and interact with relevant information based on different search tags; S2: According to the interaction results, data association and merging of search results of multiple search tags are performed; S3: Analyze the associated and merged data and sort the user activity; S4: According to the ranking result of step S3, push advertisements with corresponding frequencies and sell goods in corresponding time periods to users with different activity levels.

2. The cross-border e-commerce customer acquisition method based on intelligent retrieval according to claim 1 is characterized in that: In the step S1, the search end is used to intelligently search for target customers of the enterprise, and relevant information is interacted according to different search tags, including the following steps: S11: Use the crawler method to obtain the data information of each user in the k-th search engine: the e-th user data subset U k,e : in is the registration ID information of the e-th user, The date the user purchased the product. The price of the product purchased by the user. The place of shipment of the goods purchased by the user. Membership record for users k=1,2,…,K; K is the total number of search engines; e=1,2,…,E; E is the number of user data subsets formed by the search results of each search engine; the search results of K search engines form the collection information set U={U 1,1 ,U 1,2 ,…,U k,e ,…,U K,E }; S12: Use natural semantic recognition method to identify the Hamming distance H of multiple user data subsets in K search results of K search engines: Among them, r∈{1,2,…,K}, h∈{1,2,…,E}; g rh =1,2,...,5; g ke =1,2,…,5; is the gth user in the hth user data subset in the rth search result rh User information The length of the character segment; is the gth user data subset in the eth user data subset in the kth search result ke User information The length of the character segment; S13: Calculate the gth user data subset in the eth user data subset in the kth search result ke User information The label dimension and the gth user data subset in the hth user data subset in the rth search result rh User information The label dimension S14: Calculate the interactive correlation information value of the mth search tag in multiple user data subsets in K search results Then, we obtain a set I of interactive correlation information values ​​of multiple search tags in multiple user data subsets in K databases:

3. The cross-border e-commerce customer acquisition method based on intelligent retrieval according to claim 22 is characterized in that: The step S13 calculates the gth user data in the eth user data subset in the kth search result. ke User information The label dimension The formula is as follows: Calculate the gth user data subset in the hth user data subset in the rth search result rh User information The label dimension The formula is as follows: Among them, la m is the mth search tag, m = 1, 2, …, M; k ≠ ​​r, e ≠ h.

4. The cross-border e-commerce customer acquisition method based on intelligent retrieval according to claim 2 is characterized in that: The S2 step includes the following steps: S21: Construct a search tag ranking model Y according to the interactive correlation information value set calculated in step S14: Where A is the linear equation formed The independent variable coefficient, B is the constant of the linear equation, p is a prime number, p=7; mod is the remainder function; S22: Take the top five interactive correlation information values ​​in the results of the search tag ranking model Y from the largest to the smallest as the data integration keywords, and match and associate any user data subsets in every two search engines in the K search engines with b n As the nth data association label, n = 1, 2, 3, 4, 5; data association is combined to obtain As the bth user of the qth user in the new associated database C n User information; C = {c1, c2, ..., c q ,…,c Q }, q = 1, 2, ... Q; c q is the qth user information subset in the new associated database C, Q≤R×H; b1 to b5 all belong to {1,2,…,M} and are not equal to each other.

5. The cross-border e-commerce customer acquisition method based on intelligent retrieval according to claim 4 is characterized in that: The S3 step includes the following steps: S31: forwardly traverse all user information documents in the new associated database C, obtain the forward occurrence number J1 of the target classification word t in each subset, and calculate the forward word frequency TF(t,C) of the target classification word t: in, To calculate the new associated database C from the first user c1 to the Qth user c Q The number of all user information documents in; t = 1, 2, ..., T; T is the total number of target classification words; S32: Reversely traverse all user information documents in the new associated database C, calculate the reverse occurrence number J2 of the target classification word t in each subset, and calculate the reverse document term frequency NTF(t,C) of the target classification word t: S33: Construct the calculation formula of the user activity evaluation index value Act(t,C) based on word frequency: S34: According to the user activity evaluation index value, all user information in the new associated database is traversed to sort the activity, and the advertising time, target sales area and listing time of products with different prices are selected according to the level of activity.

6. The cross-border e-commerce customer acquisition method based on intelligent retrieval according to claim 5 is characterized in that: The step S34 comprises the following steps: S341: Take the user activity evaluation index value Act(t,C) of the target classification word t as a node, and compare it with the user activity evaluation index values ​​of any other two target classification words, take the user activity evaluation index value of the target classification word less than Act(t,C) as the left subtree node of the node, and take the user activity evaluation index value of the target classification word greater than Act(t,C) as the right subtree of the node, to form a binary tree; S342: traverse all user information documents in the new associated database C, and gradually insert the user activity evaluation index value Act(t,C) of the target classification word t as a node into the binary tree constructed according to the rule of S341; S343: repeat the steps S341-S342; S344: Calculating Binary Tree Complexity S345: Determine whether the binary tree complexity O(Act) is less than 0.35T. If so, form a sorted table of user activity evaluation index values ​​of different target classification words output from small to large; otherwise, repeat S341-S344.

7. A cross-border e-commerce customer acquisition system based on intelligent retrieval using the method according to any one of claims 1 to 6, characterized in that: The system includes an information search interaction module, a data association module, an activity ranking module and an advertisement push control module; The information search interaction module is used to intelligently search for target customers of an enterprise using a search terminal, and to interact with relevant information according to different search tags; The data association module is used to perform data association merging of search results of multiple search tags according to the interaction results; The activity ranking module performs data analysis after association and merging, and performs user activity ranking; The advertisement push control module is used to push advertisements with corresponding frequencies and sell goods in corresponding time periods to users with different activity levels according to the sorting results of the activity sorting module.

Images