Marketing information push optimization method based on artificial intelligence

Abstract

The invention relates to the field of electronic commerce, and discloses a marketing information push optimization method based on artificial intelligence, which comprises the following steps: acquiring a historical product webpage set browsed by a user in a set time period; obtaining a description document of each historical product webpage, and screening out a keyword set of each historical product webpage from the description document; screening out a common keyword set from all the keyword sets, and inputting the keyword sets and the common keyword set into a pre-trained convolutional neural network to obtain interested coefficients of the user for each historical product webpage; screening out potential interested product webpages from the historical product webpage set according to the interested coefficients, and pushing the potential interested product webpages as marketing recommendation information to the user; according to the method, the interest intensity of the user on the product can be evaluated more accurately, so that the pushing content is optimized, irrelevant or repeated product pushing is avoided, and resource waste is avoided.

Description

Technical Field

[0001] The present invention relates to the field of e-commerce, and more specifically, to an artificial intelligence-based marketing information push optimization method. Background Art

[0002] In the field of e-commerce and online marketing, accurate push notifications have become a core technology for improving user experience and conversion rates; however, existing marketing information push methods mostly rely on users' historical browsing records for recommendations; this approach has significant flaws, especially when the system simply pushes based on multiple products that the user has browsed, which can easily lead to resource waste and reduce the accuracy of the push notifications; for example, if a user has browsed multiple products, the system will push all of these products to the user, which will not only cause excessive consumption of resources, but may also cause user disgust due to the redundancy and irrelevance of the pushed content.

[0003] In addition, traditional methods fail to fully tap into users' potential interests. Pushed content is often simply matched based on users' browsing behavior, but fails to be screened based on the users' actual interest intensity, further affecting the accuracy of recommendations. Pushing too much batch content not only reduces marketing effectiveness, but may also make users feel bored, thus affecting their purchasing decisions and platform loyalty.

[0004] Therefore, how to optimize the push method based on historical browsing records, avoid waste of resources, and improve user experience while delivering accurate push information has become a technical problem that needs to be solved urgently in the current e-commerce and online marketing fields. Summary of the invention

[0005] In order to overcome the above-mentioned defects of the prior art, an embodiment of the present invention provides a marketing information push optimization method based on artificial intelligence.

[0006] To achieve the above object, the present invention provides the following technical solutions:

[0007] In a first aspect, a marketing information push optimization method based on artificial intelligence is provided, the method comprising:

[0008] Obtaining a set of historical product web pages browsed by a user within a given period of time, wherein the set of historical product web pages includes M historical product web pages, where M is an integer greater than zero;

[0009] Obtain a description document of each historical product webpage, and use the TF-IDF method to filter out a keyword set of each historical product webpage from the description document to obtain R keyword sets, wherein the keyword set includes N keywords, and R and N are integers greater than zero;

[0010] A common keyword set is selected from all keyword sets, and the keyword set and the common keyword set are input into a pre-trained convolutional neural network to obtain the user's interest coefficient for each historical product web page; the common keyword set includes S common keywords, where S is an integer greater than zero;

[0011] Potentially interested product web pages are screened out from the historical product web page collection according to the size of the interest coefficient, and the potential interested product web pages are pushed to the user as marketing recommendation information.

[0012] Furthermore, the keyword set of each historical product webpage is filtered out from the description document using the TF-IDF method, including:

[0013] a1: extract the description document of the i-th historical product web page in the historical product web page set, where i is an integer greater than zero;

[0014] a2: Perform word segmentation on the description document of the i-th historical product webpage to obtain G phrases, where G is an integer greater than zero;

[0015] a3: extract the jth word group from the G word groups, and calculate the first TF value of the jth word group using the TF-IDF method, where j is an integer greater than zero;

[0016] a4: Compare the first TF value of the j-th word group with the preset first TF threshold. If the first TF value is greater than or equal to the first TF threshold, mark the j-th word group as a keyword, set j=j+1, and return to step a3; if the first TF value is less than the first TF threshold, mark the j-th word group as a non-keyword, set j=j+1, and return to step a3;

[0017] a5: Repeat the above steps a3 to a4 until the loop ends when G=j, and N keywords are obtained. The N keywords are used as the keyword set of the i-th historical product webpage, and i=i+1, and then return to step a1;

[0018] a6: Repeat the above steps a1 to a5 until the loop ends when M=i, and obtain the keyword set of each historical product web page.

[0019] Furthermore, the screening out of the common keyword set from all keyword sets includes:

[0020] Get R keyword sets, and calculate the second TF value of each keyword in each keyword set relative to all keyword sets;

[0021] The second TF value is compared with a preset second TF threshold value. If the second TF value is greater than or equal to the second TF threshold value, the corresponding keyword is marked as a common keyword; if the second TF value is less than the second TF threshold value, the corresponding keyword is marked as a non-common keyword;

[0022] All keywords whose second TF values ​​are greater than or equal to the second TF threshold are counted to obtain S common keywords, and the S common keywords are used as a common keyword set.

[0023] Furthermore, the training method of the pre-trained convolutional neural network is as follows:

[0024] Acquire historical interest coefficient training data, and divide the historical keyword training data into a training set and a test set, wherein the historical interest coefficient training data includes feature data and its corresponding interest coefficient;

[0025] Wherein, the characteristic data includes a keyword set and a common keyword set;

[0026] Construct a convolutional neural network for performing regression tasks, use the feature data in the training set as the input of the convolutional neural network, use the coefficient of interest in the training set as the output of the convolutional neural network, train the convolutional neural network, and obtain an initial convolutional neural network;

[0027] The test set is used to verify the model of the initial convolutional neural network, and the initial convolutional neural network with a value less than or equal to the preset test error threshold is output as the pre-trained convolutional neural network.

[0028] Furthermore, the method for generating the coefficient of interest is as follows:

[0029] Using a preset similarity measurement model, calculate the similarity between the keyword set and the common keyword set;

[0030] The similarity measurement model is expressed as follows:

[0031]

[0032] Where: Sim(G,H) represents similarity, G represents the set of keywords in vectorized form, and H represents the set of common keywords in vectorized form;

[0033] The similarity is taken as the user's interest coefficient for each historical product web page.

[0034] Furthermore, the step of screening out potential product web pages of interest from the historical product web page collection includes:

[0035] Sort the interest coefficient of each historical product web page from large to small;

[0036] The historical product webpage that corresponds to the first ranking is used as the potential product webpage of interest.

[0037] In a second aspect, a marketing information push optimization system based on artificial intelligence is provided, which is implemented based on any one of the above-mentioned marketing information push optimization methods based on artificial intelligence, and the system includes:

[0038] A data acquisition module, used to acquire a set of historical product web pages browsed by a user within a given period of time, wherein the set of historical product web pages includes M historical product web pages, where M is an integer greater than zero;

[0039] An information screening module, used to obtain a description document of each historical product webpage, and use the TF-IDF method to screen out a keyword set of each historical product webpage from the description document to obtain R keyword sets, wherein the keyword set includes N keywords, and R and N are integers greater than zero;

[0040] A demand analysis module is used to filter out a common keyword set from all keyword sets, input the keyword set and the common keyword set into a pre-trained convolutional neural network, and obtain the user's interest coefficient for each historical product web page; the common keyword set includes S common keywords, where S is an integer greater than zero;

[0041] The marketing push module is used to filter out potential interest product web pages from the historical product web page collection according to the interest coefficient, and push the potential interest product web pages to users as marketing recommendation information.

[0042] According to a third aspect, an electronic device is provided, comprising a memory, a processor, and a computer program stored in the memory and running on the processor, wherein when the processor executes the computer program, the artificial intelligence-based marketing information push optimization method described in any one of the above items is implemented.

[0043] In a fourth aspect, a computer-readable storage medium is provided, on which a computer program is stored. When the computer program is executed, any of the above-mentioned artificial intelligence-based marketing information push optimization methods is implemented.

[0044] Compared with the prior art, the present invention has the following beneficial effects:

[0045] The present application discloses an artificial intelligence-based marketing information push optimization method, comprising: obtaining a set of historical product web pages browsed by a user within a given period of time; obtaining a description document of each historical product web page, and filtering out a keyword set of each historical product web page from the description document; filtering out a common keyword set from all keyword sets, and inputting the keyword set and the common keyword set into a pre-trained convolutional neural network to obtain the user's interest coefficient for each historical product web page; filtering out potential interest product web pages from the historical product web page set according to the size of the interest coefficient, and pushing the potential interest product web pages to the user as marketing recommendation information; based on the above technical features, the present invention introduces the calculation of the interest coefficient and TF-IDF keyword analysis, and this solution can more accurately evaluate the user's interest intensity in the product, thereby optimizing the push content, avoiding irrelevant or repeated product pushes, and avoiding waste of resources. BRIEF DESCRIPTION OF THE DRAWINGS

[0046] Figure 1 A flowchart of a marketing information push optimization method based on artificial intelligence provided by the present invention;

[0047] Figure 2 A schematic diagram of the module structure of a marketing information push optimization system based on artificial intelligence provided by the present invention;

[0048] Figure 3 The present invention provides a schematic structural diagram of an electronic device. DETAILED DESCRIPTION

[0049] 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.

[0050] Example 1

[0051] See also Figure 1 As shown, this embodiment discloses a marketing information push optimization method based on artificial intelligence, and the method includes:

[0052] S101: Acquire a set of historical product web pages browsed by a user within a predetermined period of time, wherein the set of historical product web pages includes M historical product web pages, where M is an integer greater than zero;

[0053] It should be noted that: the predetermined time period is set artificially in advance by technicians. For example, it can be one hour, one day, or one week, etc. Further explanation is that by selecting the predetermined time period, the potential needs of users at different stages can be effectively analyzed, which is conducive to realizing product matching for users at different stages; among them, the M historical product web pages are obtained by collecting user historical browsing logs;

[0054] S102: Obtain the description document of each historical product web page, and use the TF-IDF method to screen out the keyword set of each historical product web page from the description document, obtaining R keyword sets. Each keyword set contains N keywords, where R and N are integers greater than zero;

[0055] It should be understood that: at present, in order to ensure that users can more easily find relevant products when using the search engine or the internal search function of the website, e-commerce enterprises will specify multiple relevant keywords for each product page, and at the same time, carry out relevant page layout and introduction according to these keywords. In this way, on the one hand, it can accurately summarize the product characteristics, and on the other hand, it can facilitate users' accurate search;

[0056] In implementation, the use of the TF-IDF method to screen out the keyword set of each historical product web page from the description document includes:

[0057] a1: Extract the description document of the i-th historical product web page in the set of historical product web pages, where i is an integer greater than zero;

[0058] a2: Perform word segmentation on the description document of the i-th historical product web page to obtain G phrases, where G is an integer greater than zero;

[0059] It should be noted that: before using the TF-IDF method to screen out the keyword set of each historical product web page from the description document, word segmentation will be performed on the description document and meaningless characters will be removed. The word segmentation is implemented based on existing word segmentation means, such as: using jieba word segmentation software, etc.; among them, the meaningless characters include meaningless words and phrases (including but not limited to repeated words, stop words removal, onomatopoeia, and imitation words, etc.) and meaningless symbols (such as: "ah", "de", ",", and ".", etc.);

[0060] a3: Extract the j-th phrase from the G phrases, and use the TF-IDF method to calculate the first TF value of the j-th phrase, where j is an integer greater than zero;

[0061] It should be understood that the TF-IDF method is a commonly used text mining technology, which is used to measure the importance of a word in a document and the entire document set; it combines the two parts of word frequency (TF value) and inverse document frequency (IDF value), and is widely used in information retrieval, text classification, keyword extraction and other fields; among them, the TF value refers to the frequency of a word appearing in a document, which is used to measure the relative importance of a word appearing in a document; the higher the TF value, the higher the frequency of the word in the document, and the greater its contribution to the content of the document;

[0062] It is understandable that in the SEO optimization scenario of product web pages, each product page is typed and laid out based on multiple keywords, so that the multiple keywords of each product page can reflect the characteristics of the corresponding product; for example, when a product page is a running shoe, and the keywords of the product page are "anti-odor" and "breathable", it means that "anti-odor" and "breathable" are the characteristics of the product (running shoes) on the product page;

[0063] It can be further understood that: by analyzing and extracting keywords from the user's historical product web pages, it can help to accurately extract the user's potential needs, thereby providing important data support and decision-making basis for the accurate push of subsequent marketing recommendation information, and then facilitating the implementation of personalized marketing information push to improve user satisfaction and conversion rate;

[0064] a4: Compare the first TF value of the j-th word group with the preset first TF threshold. If the first TF value is greater than or equal to the first TF threshold, mark the j-th word group as a keyword, set j=j+1, and return to step a3; if the first TF value is less than the first TF threshold, mark the j-th word group as a non-keyword, set j=j+1, and return to step a3;

[0065] a5: Repeat the above steps a3 to a4 until the loop ends when G=j, and N keywords are obtained. The N keywords are used as the keyword set of the i-th historical product webpage, and i=i+1, and then return to step a1;

[0066] a6: Repeat the above steps a1 to a5 until the loop ends when M=i, and obtain the keyword set of each historical product web page.

[0067] S103: Filter out a common keyword set from all keyword sets, input the keyword set and the common keyword set into a pre-trained convolutional neural network, and obtain the user's interest coefficient for each historical product web page; the common keyword set includes S common keywords, where S is an integer greater than zero;

[0068] In implementation, the method of screening out a common keyword set from all keyword sets includes:

[0069] Get R keyword sets, and calculate the second TF value of each keyword in each keyword set relative to all keyword sets;

[0070] The second TF value is compared with a preset second TF threshold value. If the second TF value is greater than or equal to the second TF threshold value, the corresponding keyword is marked as a common keyword; if the second TF value is less than the second TF threshold value, the corresponding keyword is marked as a non-common keyword;

[0071] Count all keywords whose second TF values ​​are greater than or equal to the second TF threshold to obtain S common keywords, and use the S common keywords as a common keyword set;

[0072] It should be noted that: unlike the first TF value, which is used to reveal the frequency of occurrence of a phrase in a description document, the second TF value is used to reveal the frequency of occurrence of a keyword in all keyword sets;

[0073] It can be understood that: when the second TF value of one or more common keywords is larger, it can more indirectly reflect the user's potential product needs at the current stage; for example, when a user browses multiple historical product webpages of running shoes, and "anti-odor" and "breathable" appear in multiple historical product webpages, it means that the user currently needs a "anti-odor" and "breathable" running shoe product; therefore, common keyword mining can help accurately locate the user's potential needs at the current stage, which is conducive to the accurate push of subsequent marketing recommendation information;

[0074] In implementation, the training method of the pre-trained convolutional neural network is as follows:

[0075] Acquire historical interest coefficient training data, and divide the historical keyword training data into a training set and a test set, wherein the historical interest coefficient training data includes feature data and its corresponding interest coefficient;

[0076] Wherein, the characteristic data includes a keyword set and a common keyword set;

[0077] It should be noted that the characteristic data and interest coefficients in the historical interest coefficient training data are obtained by technicians based on historical records and experimental data collection records;

[0078] The method for generating the coefficient of interest is as follows:

[0079] Using a preset similarity measurement model, calculate the similarity between the keyword set and the common keyword set;

[0080] Specifically, the expression of the similarity measurement model is as follows:

[0081]

[0082] Where: Sim(G,H) represents similarity, G represents the set of keywords in vectorized form, and H represents the set of common keywords in vectorized form;

[0083] The similarity is used as the user's interest coefficient for each historical product web page;

[0084] Construct a convolutional neural network for performing regression tasks, use the feature data in the training set as the input of the convolutional neural network, use the coefficient of interest in the training set as the output of the convolutional neural network, train the convolutional neural network, and obtain an initial convolutional neural network;

[0085] The test set is used to verify the model of the initial convolutional neural network, and the initial convolutional neural network with a value less than or equal to the preset test error threshold is output as the pre-trained convolutional neural network.

[0086] S104: Filtering potential product web pages of interest from the historical product web page set according to the interest coefficient, and pushing the potential product web pages of interest to the user as marketing recommendation information;

[0087] In implementation, the step of selecting potential product web pages of interest from the historical product web page collection includes:

[0088] Sort the interest coefficient of each historical product web page from large to small;

[0089] The historical product webpage that is ranked first is used as the potential product webpage of interest;

[0090] By optimizing the marketing information push based on historical browsing records, the aim is to solve the problems of resource waste and lack of accuracy in traditional push methods. In existing marketing push methods, product recommendations usually rely on users' browsing history, but this method often fails to effectively capture changes in users' interests, resulting in repeated and redundant push content, which in turn affects the accuracy of push and user experience. By introducing the calculation of interest coefficients and TF-IDF keyword analysis, this solution can more accurately evaluate the intensity of users' interest in products, thereby optimizing push content, avoiding irrelevant or repeated product push, and avoiding resource waste.

[0091] Example 2

[0092] See also Figure 2 As shown, based on the same inventive concept, this embodiment discloses a marketing information push optimization system based on artificial intelligence. For details not provided in this embodiment, please refer to the description of the relevant parts in Embodiment 1. The system includes:

[0093] The data acquisition module 210 is used to acquire a set of historical product web pages browsed by a user within a predetermined period of time, wherein the set of historical product web pages includes M historical product web pages, where M is an integer greater than zero;

[0094] An information screening module 220 is used to obtain a description document of each historical product webpage, and screen a keyword set of each historical product webpage from the description document using a TF-IDF method to obtain R keyword sets, wherein the keyword set includes N keywords, and R and N are integers greater than zero;

[0095] The demand analysis module 230 is used to filter out a common keyword set from all keyword sets, input the keyword set and the common keyword set into a pre-trained convolutional neural network, and obtain the user's interest coefficient for each historical product web page; the common keyword set includes S common keywords, where S is an integer greater than zero;

[0096] The marketing push module 240 is used to filter out potential interest product web pages from the historical product web page set according to the interest coefficient, and push the potential interest product web pages to the user as marketing recommendation information.

[0097] Example 3

[0098] See also Figure 3 As shown, this embodiment discloses an electronic device, including a memory, a processor, and a computer program stored in the memory and running on the processor, and when the processor executes the computer program, it implements any one of the artificial intelligence-based marketing information push optimization methods provided by the above methods.

[0099] Since the electronic device introduced in this embodiment is an electronic device used to implement the marketing information push optimization method based on artificial intelligence in the embodiment of this application, based on the marketing information push optimization method based on artificial intelligence introduced in the embodiment of this application, the technical personnel of this field can understand the specific implementation of the electronic device of this embodiment and its various variations, so how the electronic device implements the method in the embodiment of this application is not introduced in detail here. As long as the technical personnel of this field implement the electronic device used by the marketing information push optimization method based on artificial intelligence in the embodiment of this application, it belongs to the scope of protection of this application.

[0100] Example 4

[0101] This embodiment discloses a computer-readable storage medium, including a memory, a processor, and a computer program stored in the memory and running on the processor. When the processor executes the computer program, it implements any one of the artificial intelligence-based marketing information push optimization methods provided by the above methods.

[0102] The above formulas are all dimensionless and numerical calculations. The formula is a formula for the most recent real situation obtained by collecting a large amount of data and performing software simulation. The preset parameters, weights and thresholds in the formula are set by technicians in this field according to the situation.

[0103] The above embodiments can be implemented in whole or in part by software, hardware, firmware or any other combination. When implemented by software, the above embodiments can be implemented in whole or in part in the form of a computer program product. The computer program product includes one or more computer instructions or computer programs. When the computer instructions or computer programs are loaded or executed on a computer, the process or function described in the embodiment of the present invention is generated in whole or in part. The computer can be a general-purpose computer, a special-purpose computer, a computer network or other programmable device. The computer instructions can be stored in a computer-readable storage medium, or transmitted from one computer-readable storage medium to another computer-readable storage medium. For example, the computer instructions can be transmitted from one website site, computer, server or data center to another website site, computer, server or data center through a wired network or a wireless network. The computer-readable storage medium can be any available medium that can be accessed by a computer or a data storage device such as a server or data center that contains one or more available media sets. The available medium can be a magnetic medium (for example, a floppy disk, a hard disk, a tape), an optical medium (for example, a DVD) or a semiconductor medium. The semiconductor medium can be a solid-state hard disk.

[0104] Those skilled in the art will appreciate that the units and algorithm steps of each example described in conjunction with the embodiments disclosed in the present invention can be implemented in electronic hardware or a combination of computer software and electronic hardware. Whether these functions are performed in hardware or software depends on the specific application and design constraints of the technical solution. Professional and technical personnel can use different methods to implement the described functions for each specific application, but such implementation should not be considered to be beyond the scope of the present invention.

[0105] Those skilled in the art can clearly understand that, for the convenience and brevity of description, the specific working processes of the systems, devices and units described above can refer to the corresponding processes in the aforementioned method embodiments and will not be repeated here.

[0106] In the several embodiments provided by the present invention, it should be understood that the disclosed systems, devices and methods can be implemented in other ways. For example, the device embodiments described above are only schematic, for example, the division of the units is only one, and there may be other division methods during implementation, such as multiple units or components can be combined or integrated into another system, or some features can be ignored or not executed. Another point is that the mutual coupling or direct coupling or communication connection shown or discussed can be through some interfaces, indirect coupling or communication connection of devices or units, which can be electrical, mechanical or other forms.

[0107] The units described as separate components may or may not be physically separated, and the components shown as units may or may not be physical units, that is, they may be located in one place or distributed on multiple network units. Some or all of the units may be selected as needed to achieve the purpose of the solution of this embodiment.

[0108] In addition, each functional unit in each embodiment of the present invention may be integrated into one processing unit, or each unit may exist physically separately, or two or more units may be integrated into one unit.

[0109] The above is only a specific embodiment of the present invention, but the protection scope of the present invention is not limited thereto. Any person skilled in the art can easily think of changes or substitutions within the technical scope disclosed by the present invention, which should be included in the protection scope of the present invention. Therefore, the protection scope of the present invention should be based on the protection scope of the claims.

[0110] Finally: The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc. made within the spirit and principles of the present invention should be included in the protection scope of the present invention.

Claims

1. A marketing information push optimization method based on artificial intelligence, characterized in that: The method comprises: Obtaining a set of historical product web pages browsed by a user within a given period of time, wherein the set of historical product web pages includes M historical product web pages, where M is an integer greater than zero; Obtain a description document of each historical product webpage, and use the TF-IDF method to filter out a keyword set of each historical product webpage from the description document to obtain R keyword sets, wherein the keyword set includes N keywords, and R and N are integers greater than zero; A common keyword set is selected from all keyword sets, and the keyword set and the common keyword set are input into a pre-trained convolutional neural network to obtain the user's interest coefficient for each historical product web page; the common keyword set includes S common keywords, where S is an integer greater than zero; Potentially interested product web pages are screened out from the historical product web page collection according to the size of the interest coefficient, and the potential interested product web pages are pushed to the user as marketing recommendation information.

2. The marketing information push optimization method based on artificial intelligence according to claim 1 is characterized in that: The keyword set of each historical product webpage is screened out from the description document using the TF-IDF method, including: a1: extract the description document of the i-th historical product web page in the historical product web page set, where i is an integer greater than zero; a2: Perform word segmentation on the description document of the i-th historical product webpage to obtain G phrases, where G is an integer greater than zero; a3: extract the jth word group from the G word groups, and calculate the first TF value of the jth word group using the TF-IDF method, where j is an integer greater than zero; a4: Compare the first TF value of the j-th word group with the preset first TF threshold. If the first TF value is greater than or equal to the first TF threshold, mark the j-th word group as a keyword, set j=j+1, and return to step a3; if the first TF value is less than the first TF threshold, mark the j-th word group as a non-keyword, set j=j+1, and return to step a3; a5: Repeat the above steps a3 to a4 until the loop ends when G=j, and N keywords are obtained. The N keywords are used as the keyword set of the i-th historical product webpage, and i=i+1, and then return to step a1; a6: Repeat the above steps a1 to a5 until the loop ends when M=i, and obtain the keyword set of each historical product web page.

3. The marketing information push optimization method based on artificial intelligence according to claim 2 is characterized in that: The method of screening out a common keyword set from all keyword sets includes: Get R keyword sets, and calculate the second TF value of each keyword in each keyword set relative to all keyword sets; The second TF value is compared with a preset second TF threshold value. If the second TF value is greater than or equal to the second TF threshold value, the corresponding keyword is marked as a common keyword; if the second TF value is less than the second TF threshold value, the corresponding keyword is marked as a non-common keyword; All keywords whose second TF values ​​are greater than or equal to the second TF threshold are counted to obtain S common keywords, and the S common keywords are used as a common keyword set.

4. The marketing information push optimization method based on artificial intelligence according to claim 3 is characterized in that: The training method of the pre-trained convolutional neural network is as follows: Acquire historical interest coefficient training data, and divide the historical keyword training data into a training set and a test set, wherein the historical interest coefficient training data includes feature data and its corresponding interest coefficient; Wherein, the characteristic data includes a keyword set and a common keyword set; Construct a convolutional neural network for performing regression tasks, use the feature data in the training set as the input of the convolutional neural network, use the coefficient of interest in the training set as the output of the convolutional neural network, train the convolutional neural network, and obtain an initial convolutional neural network; The test set is used to verify the model of the initial convolutional neural network, and the initial convolutional neural network with a value less than or equal to the preset test error threshold is output as the pre-trained convolutional neural network.

5. The marketing information push optimization method based on artificial intelligence according to claim 4 is characterized in that: The method for generating the coefficient of interest is as follows: Using a preset similarity measurement model, calculate the similarity between the keyword set and the common keyword set; The similarity measurement model is expressed as follows: Where: Sim(G,H) represents similarity, G represents the set of keywords in vectorized form, and H represents the set of common keywords in vectorized form; The similarity is taken as the user's interest coefficient for each historical product web page.

6. The marketing information push optimization method based on artificial intelligence according to claim 5 is characterized in that: The step of selecting potential product web pages of interest from the historical product web page collection includes: Sort the interest coefficient of each historical product web page from large to small; The historical product webpage that corresponds to the first ranking is used as the potential product webpage of interest.

7. A marketing information push optimization system based on artificial intelligence, characterized in that: The method is implemented based on the artificial intelligence-based marketing information push optimization method according to any one of claims 1 to 6, and the system comprises: A data acquisition module, used to acquire a set of historical product web pages browsed by a user within a given period of time, wherein the set of historical product web pages includes M historical product web pages, where M is an integer greater than zero; An information screening module, used to obtain a description document of each historical product webpage, and use the TF-IDF method to screen out a keyword set of each historical product webpage from the description document to obtain R keyword sets, wherein the keyword set includes N keywords, and R and N are integers greater than zero; A demand analysis module is used to filter out a common keyword set from all keyword sets, input the keyword set and the common keyword set into a pre-trained convolutional neural network, and obtain the user's interest coefficient for each historical product web page; the common keyword set includes S common keywords, where S is an integer greater than zero; The marketing push module is used to filter out potential interest product web pages from the historical product web page collection according to the interest coefficient, and push the potential interest product web pages to users as marketing recommendation information.

8. An electronic device comprising a memory, a processor, and a computer program stored in the memory and running on the processor, characterized in that: When the processor executes the computer program, the marketing information push optimization method based on artificial intelligence described in any one of claims 1 to 6 is implemented.

9. A computer-readable storage medium, characterized in that: The computer-readable storage medium stores a computer program, and when the computer program is executed, the artificial intelligence-based marketing information push optimization method according to any one of claims 1 to 6 is implemented.

Images