Advertisement delivery optimization system and method based on artificial intelligence

Abstract

The present invention discloses an artificial intelligence-based advertising delivery optimization system and method, and relates to the technical field of advertising delivery optimization. The present invention is authorized by a mobile terminal user to collect historical operation data and real-time operation data in a mobile terminal application software; analyzes the indicators of the historical operation data corresponding to the jump operation between the mobile terminal user's application software, and generates a feature combination of the terminal user's historical operation data; and constructs an application software advertising analysis model according to the feature combination of the historical operation data; based on the real-time operation data of the mobile terminal user in the mobile terminal application software, the subjective true value of the real-time operation data is analyzed by the application software advertising analysis model, and according to the application software control model, a control instruction is output for the jump operation of the mobile terminal application software, and the mobile terminal is controlled. Improve the comfort level of the mobile terminal user's application software use.

Description

Technical Field

[0001] The present invention relates to the technical field of advertising delivery optimization, and in particular to an advertising delivery optimization system and method based on artificial intelligence. Background Art

[0002] Mobile Internet is a combination of Internet and mobile communication. As an extension and development of traditional Internet, it will bring several major impacts: from real-time to anytime time concept, from online to on-site space concept, from functional to intelligent terminal concept. This has also triggered a quiet change in advertising communication, which is specifically reflected in: advertising subjects from single to multiple, advertising information from unbalanced to complete, advertising media from split screen to cross-screen, advertising objects from niche to elite, and advertising effects from precise to matching.

[0003] In the era of mobile Internet, using mobile phone software to publish advertisements is a normal business model; however, the means of some software pop-up advertisements to induce clicks are constantly "upgraded", making it increasingly difficult for users to defend themselves; when the software is started, a slight shake will cause the screen to immediately jump to a third-party shopping software; when the software is used, if you leave it for a while, the advertisement will pop up again when you return; if you try to click the close button, you will still be redirected to the advertisement; the above-mentioned bad advertising model brings a bad user experience to mobile terminal users, which can easily make mobile terminal users resist the generation of advertisements, and is not conducive to the spread of shopping advertisements.

[0004] Therefore, the present invention discloses an artificial intelligence-based advertising delivery optimization system and method to solve the above problems. Summary of the invention

[0005] The purpose of the present invention is to provide an artificial intelligence-based advertising optimization system and method to solve the problems raised in the prior art.

[0006] To achieve the above object, the present invention provides the following technical solution: an advertising delivery optimization method based on artificial intelligence, the method comprising the following steps:

[0007] S1: The mobile terminal user authorizes the collection of historical operation data and real-time operation data in the mobile terminal application software; the historical operation data includes the historical operation data of the shopping software and the historical operation data of the non-shopping software; the historical operation data of the non-shopping software includes the historical pre-jump page data and the mobile terminal reverse data; the historical operation data of the shopping software includes the historical daily browsing data and the historical post-jump browsing data;

[0008] S2: constructing an application software control model, wherein the application software control model is used to manage and control the jump operation of the mobile terminal application software;

[0009] S3: extracting the historical operation data of the application software, analyzing the indicators of the historical operation data corresponding to the jump operation between the application software of the mobile terminal user, generating the feature combination of the historical operation data of the terminal user; and constructing the application software advertising analysis model based on the feature combination of the historical operation data;

[0010] S4: Based on the real-time operation data of mobile terminal users in the mobile terminal application software, the subjective true value of the real-time operation data is analyzed by the application software advertising analysis model, and according to the application software control model, a control instruction is output for the jump operation of the mobile terminal application software, and the control instruction is sent to the mobile terminal system execution end to control the mobile terminal.

[0011] In S1, the historical daily browsing data is the data corresponding to the mobile terminal user entering the shopping software without jumping and performing shopping browsing.

[0012] According to the above scheme, in S2, the application software control model is arranged inside the mobile terminal system software, and the mobile terminal user performs authorization processing. During the use of the mobile terminal application software by the mobile terminal user, the real-time operation data of the mobile terminal application software is monitored. When the mobile terminal application software performs a jump operation between software, jump operation management control is provided according to the application software control model; the jump operation management control includes consent to jump and jump cancellation.

[0013] According to the above scheme, S3 includes the following:

[0014] S301: extracting the timestamp of the page data before the historical jump, the system sets a time range for extracting historical daily browsing data, and extracting historical daily browsing data of the historical years according to the extraction time range; the historical daily browsing data includes the product details page data and corresponding browsing time data of a single product in the historical browsing records of the shopping software;

[0015] There are certain periodic changes in the shopping habits of mobile terminal users. In the rapidly developing information age, there are shopping activities at the same time nodes, which further increases the periodicity of the shopping habits of mobile terminal users. The present invention sets a time range for extracting historical daily browsing data through the system, and extracts historical daily browsing data of historical years according to the extraction time range; the historical daily browsing data contains the shopping tendencies of mobile terminal users, and the ones that form shopping orders are direct shopping tendencies, and the ones that do not form shopping orders are indirect shopping tendencies. The historical daily browsing data provides data support for subsequent analysis, thereby improving the accuracy of the system.

[0016] S302: Analyze the similarity between the page data before the historical jump and the product details page data of a single product to form a page similarity value SF; generate a mobile terminal twist angle RE according to the mobile terminal twist data; generate a subjective true value STV of the jump operation according to the browsing data after the historical jump and the browsing time data corresponding to the product details page data of a single product;

[0017] S303: Based on the page similarity value, the mobile terminal torsion angle and the subjective true value, a historical operation data feature combination is formed, which is recorded as DF i =(SF i , R.E. i , STV i ); Each historical jump operation corresponds to a historical running data feature combination; where DF i represents the i-th historical operation data feature combination; SF i Represents the page similarity value in the i-th historical operation data feature combination; RE i STV represents the twist angle of the mobile terminal in the i-th historical operation data feature combination; i represents the subjective true value in the i-th historical operation data feature combination; i∈[1,I]; I represents the total number of historical operation data feature combinations;

[0018] S304: Utilize the combination of historical operation data features, take the page similarity value and the mobile terminal twisting angle as independent variables, and take the subjective true value as the dependent variable, to establish an application software advertising analysis model: STV=α1×SF+α2×RE+α3; where α1, α2 and α3 represent fitting coefficients, and use the least squares method to calculate and solve α1, α2 and α3 in the application software advertising analysis model.

[0019] Building an application software advertising analysis model can effectively combine historical operation data features for quantitative analysis, providing effective data support for the subjective true value of subsequent calculation jump operations of real-time operation data.

[0020] According to the above solution, in S302, the calculation of page similarity values ​​includes the following contents:

[0021] The text content in the historical pre-jump page data and the product detail page data is extracted using a text extraction technology; a text feature vector group of the historical pre-jump page data and the product detail page data is formed respectively, and a first similarity FS is calculated based on the text feature vector group of the historical pre-jump page data and the product detail page data;

[0022] ;

[0023] Among them, PBG nRepresents the nth text feature vector of the page data before the historical jump, HD （m，n） Represents the nth text feature vector group of the mth product detail page data; n∈[1,N]; the number of text feature vectors in the Nth text feature vector group; m∈[1,M]; M represents the number of product detail page data;

[0024] Convert the images in the page data before the historical jump and the product details page data into the same color space to generate a color histogram of each image; calculate a second similarity SS based on the color histogram of the image in the page data before the historical jump and the product details page data; the second similarity is equal to the average value of the histogram intersection of the color histogram of the page data image before the historical jump and the color histogram of each product details page data image;

[0025] Calculate the page similarity value SF according to the first similarity FS and the second similarity SS:

[0026] SF = A1 × FS + A2 × SS;

[0027] Wherein, A1 and A2 are both similarity coefficients, and the similarity coefficients are system preset constant page similarity values;

[0028] The attractiveness of shopping software advertisements is mainly reflected in text and images. The present invention can further improve the accuracy and robustness of system analysis by analyzing the above two aspects.

[0029] The mobile terminal torsion angle RE is collected by a sensor in the mobile terminal;

[0030] The browsing time corresponding to the product details page data that has not formed a shopping order is extracted, and the average value is calculated and recorded as the browsing time intervention value; the subjective true value STV is equal to the quotient of the page browsing time after the jump and the browsing time intervention value; wherein, if the subjective true value is greater than 1, the jump is marked as a subjective jump; if the subjective true value is less than or equal to 1, the jump is marked as a non-subjective jump.

[0031] The indirect shopping tendency is when a shopping order is not formed. The corresponding mobile terminal users are less interested in this part of the product detail page data than the product detail page data that forms a shopping order. By calculating the browsing time corresponding to the product detail page data that does not form a shopping order as the browsing time intervention value, it can more effectively distinguish between indirect shopping tendencies and direct shopping tendencies, and further quantify the shopping tendencies, clearly analyze the historical operation data, provide data support for building an application software advertising analysis model, and further improve the accuracy of system analysis and judgment.

[0032] According to the above scheme, in S4, the pre-jump page data and mobile terminal twisting data in the real-time running data are substituted into the application software advertising analysis model to generate a subjective true value STV; if the subjective true value STV is greater than 1, the jump is determined to be a subjective jump, and the jump is not controlled; if the subjective true value STV is less than or equal to 1, the jump is determined to be a non-subjective jump, and a jump inhibition instruction is generated; the jump inhibition instruction is sent to the mobile terminal system execution end to prevent the mobile terminal from performing this jump operation.

[0033] The subjective true value of the jump operation can be quantified based on real-time operation data and application software advertising analysis models, which can effectively eliminate shopping software advertisements with low shopping tendencies among mobile terminal users' indirect shopping tendencies, effectively avoid application software jumps caused by non-subjective consciousness of mobile terminal users, and improve the comfort level of mobile terminal users in using application software.

[0034] Another aspect of the present application is an artificial intelligence-based advertising delivery optimization system, which is applied to the above-mentioned artificial intelligence-based advertising delivery optimization method, and includes an authorization data acquisition module, an application software control model construction module, an advertising analysis model construction module, and an application software control module;

[0035] The authorization data collection module is used to collect historical operation data and real-time operation data in the mobile terminal application software with authorization from the mobile terminal user;

[0036] The application software control model building module is used to build an application software control model, and the application software control model is used to manage and control the jump operation of the mobile terminal application software;

[0037] The advertising analysis model building module is used to extract the historical operation data of the application software, analyze the indicators of the historical operation data corresponding to the jump operation between the application software of the mobile terminal user, generate the historical operation data feature combination of the terminal user; and build the application software advertising analysis model according to the historical operation data feature combination;

[0038] The application software control module is used to output control instructions for the jump operation of the mobile terminal application software based on the real-time operation data of the mobile terminal user in the mobile terminal application software, through the analysis of the subjective true value of the real-time operation data by the application software advertising analysis model, and according to the application software control model, and send the control instructions to the mobile terminal system execution end to control the mobile terminal.

[0039] According to the above scheme, the historical operation data includes shopping software historical operation data and non-shopping software historical operation data; the non-shopping software historical operation data includes historical pre-jump page data and mobile terminal twist data; the shopping software historical operation data includes historical daily browsing data and historical post-jump browsing data.

[0040] According to the above scheme, the application software control model is arranged inside the mobile terminal system software, and the mobile terminal user performs authorization processing. During the use of the mobile terminal application software by the mobile terminal user, the real-time operation data of the mobile terminal application software is monitored. When the mobile terminal application software performs a jump operation between software, jump operation management control is provided according to the application software control model; the jump operation management control includes consent to jump and jump cancellation.

[0041] According to the above scheme, the advertising analysis model building module includes a historical daily browsing data splitting unit, a historical operation data indicator analysis unit and a model building unit;

[0042] The historical daily browsing data splitting unit is used to extract the timestamp of the page data before the historical jump. The system sets a time range for extracting the historical daily browsing data, and extracts the historical daily browsing data of the historical years according to the extraction time range; the historical daily browsing data includes the product details page data and the corresponding browsing time data of a single product in the historical browsing records of the shopping software;

[0043] The historical operation data indicator analysis unit is used to analyze the similarity between the historical page data before the jump and the product details page data of a single product to form a page similarity value SF; generate a mobile terminal twist angle RE according to the mobile terminal twist data; generate a subjective true value STV of the jump operation according to the browsing data after the historical jump and the browsing time data corresponding to the product details page data of a single product, and form a historical operation data feature combination according to the page similarity value, the mobile terminal twist angle and the subjective true value;

[0044] The model building unit is used to utilize the historical operation data feature combination, take the page similarity value and the mobile terminal twist angle as independent variables, take the subjective true value as the dependent variable, establish the application software advertising analysis model, and use the least squares method to calculate and solve the fitting coefficient in the application software advertising analysis model.

[0045] According to the above solution, the application software control module includes a subjective truth value calculation unit and an instruction generation and transmission unit;

[0046] The subjective truth value calculation unit is used to substitute the pre-jump page data and the mobile terminal twist data in the real-time operation data into the application software advertisement analysis model to generate a subjective truth value;

[0047] The instruction generation and transmission unit generates a jump inhibition instruction according to the subjective true value; and sends the jump inhibition instruction to the mobile terminal system execution end to prevent the current jump operation of the mobile terminal.

[0048] Compared with the prior art, the beneficial effects of the present invention are as follows: there are certain periodic changes in the shopping habits of mobile terminal users. In the rapidly developing information age, there are shopping activities at the same time nodes, which further improves the periodicity of the shopping habits of mobile terminal users; the present invention sets a time range for extracting historical daily browsing data through the system, and extracts historical daily browsing data of historical years according to the extraction time range; there are shopping tendencies of mobile terminal users in the historical daily browsing data, and those that form shopping orders are direct shopping tendencies, and those that do not form shopping orders are indirect shopping tendencies. The historical daily browsing data provides data support for subsequent analysis, thereby improving the accuracy of the system; constructing an application software advertising analysis model can effectively combine historical operation data features for quantitative analysis, and provide effective data support for the subjective true value of the subsequent calculation jump operation of real-time operation data; the attractiveness of shopping software advertisements is mainly reflected in text and images, and the present invention uses the above The two-aspect analysis can further improve the accuracy and robustness of the system analysis; the indirect shopping tendency is that the shopping order is not formed, and the corresponding mobile terminal users are less interested in this part of the product detail page data than the product detail page data that forms a shopping order. The browsing time corresponding to the product detail page data that does not form a shopping order is calculated as the browsing time intervention value, which can more effectively distinguish between indirect shopping tendencies and direct shopping tendencies, and further quantify the shopping tendencies, clearly analyze the historical operation data, and provide data support for building an application software advertising analysis model, further improving the accuracy of system analysis and judgment; according to the real-time operation data and the application software advertising analysis model, the subjective true value of the jump operation can be quantified, which can effectively eliminate the shopping software advertisements with lower shopping tendencies among the indirect shopping tendencies of mobile terminal users, effectively avoid the situation where the application software jumps due to the non-subjective consciousness of the mobile terminal users, and improve the comfort level of mobile terminal users in using application software. BRIEF DESCRIPTION OF THE DRAWINGS

[0049] 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:

[0050] Figure 1 It is a flow chart of the method for optimizing advertisement delivery based on artificial intelligence of the present invention;

[0051] Figure 2 It is a structural schematic diagram of the artificial intelligence-based advertising delivery optimization system of the present invention. DETAILED DESCRIPTION

[0052] 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.

[0053] See also Figure 1 The present invention provides a technical solution: an advertising delivery optimization method based on artificial intelligence, the method comprising the following steps:

[0054] S1: The mobile terminal user authorizes the collection of historical operation data and real-time operation data in the mobile terminal application software; the historical operation data includes the historical operation data of shopping software and the historical operation data of non-shopping software; the historical operation data of non-shopping software includes the historical pre-jump page data and mobile terminal reversal data; the historical operation data of shopping software includes the historical daily browsing data and the historical post-jump browsing data;

[0055] S2: Building an application software control model, which is used to manage and control the jump operation of the mobile terminal application software;

[0056] In S2, the application software control model is arranged inside the mobile terminal system software, and the mobile terminal user performs authorization processing. When the mobile terminal user uses the mobile terminal application software, the real-time operation data of the mobile terminal application software is monitored. When the mobile terminal application software performs a jump operation between software, jump operation management control is provided according to the application software control model; the jump operation management control includes consent to jump and jump cancellation.

[0057] S3: extracting the historical operation data of the application software, analyzing the indicators of the historical operation data corresponding to the jump operation between the application software of the mobile terminal user, generating the feature combination of the historical operation data of the terminal user; and constructing the application software advertising analysis model based on the feature combination of the historical operation data;

[0058] In S3, include the following:

[0059] S301: extract the timestamp of the page data before the historical jump, the system sets the time range for extracting the historical daily browsing data, and extracts the historical daily browsing data of the historical years according to the extraction time range; the historical daily browsing data includes the product details page data and the corresponding browsing time data of a single product in the historical browsing records of the shopping software;

[0060] S302: Analyze the similarity between the page data before the historical jump and the product details page data of a single product to form a page similarity value SF; generate a mobile terminal twist angle RE according to the mobile terminal twist data; generate a subjective true value STV of the jump operation according to the browsing data after the historical jump and the browsing time data corresponding to the product details page data of a single product;

[0061] S303: Based on the page similarity value, the mobile terminal torsion angle and the subjective true value, a historical operation data feature combination is formed, which is recorded as DF i =(SF i , R.E. i , STV i ); Each historical jump operation corresponds to a historical running data feature combination; where DF i represents the i-th historical operation data feature combination; SF i Represents the page similarity value in the i-th historical operation data feature combination; RE i STV represents the twist angle of the mobile terminal in the i-th historical operation data feature combination; i represents the subjective true value in the i-th historical operation data feature combination; i∈[1,I]; I represents the total number of historical operation data feature combinations;

[0062] S304: Utilize the combination of historical operation data features, take the page similarity value and the mobile terminal twisting angle as independent variables, and take the subjective true value as the dependent variable, to establish an application software advertising analysis model: STV=α1×SF+α2×RE+α3; where α1, α2 and α3 represent fitting coefficients, and use the least squares method to calculate and solve α1, α2 and α3 in the application software advertising analysis model.

[0063] In S302, the calculation of page similarity values ​​includes the following:

[0064] The text content in the historical pre-jump page data and the product detail page data is extracted using a text extraction technology; a text feature vector group of the historical pre-jump page data and the product detail page data is formed respectively, and a first similarity FS is calculated based on the text feature vector group of the historical pre-jump page data and the product detail page data;

[0065] ;

[0066] Among them, PBG n Represents the nth text feature vector of the page data before the historical jump, HD （m，n） Represents the nth text feature vector group of the mth product detail page data; n∈[1,N]; the number of text feature vectors in the Nth text feature vector group; m∈[1,M]; M represents the number of product detail page data;

[0067] The images in the page data before the historical jump and the product details page data are converted into the same color space to generate color histograms of each image; the second similarity SS is calculated according to the color histograms of the images in the page data before the historical jump and the product details page data; the second similarity is equal to the average value of the histogram intersection of the color histogram of the page data image before the historical jump and the color histogram of each product details page data image;

[0068] Calculate the page similarity value SF according to the first similarity FS and the second similarity SS:

[0069] SF = A1 × FS + A2 × SS;

[0070] Among them, A1 and A2 are similarity coefficients, and the similarity coefficients are system preset constants;

[0071] The mobile terminal torsion angle RE is collected by a sensor in the mobile terminal;

[0072] The browsing time corresponding to the product details page data that has not formed a shopping order is extracted, and the average value is calculated and recorded as the browsing time intervention value; the subjective true value STV is equal to the quotient of the page browsing time after the jump and the browsing time intervention value; wherein, if the subjective true value is greater than 1, the jump is marked as a subjective jump; if the subjective true value is less than or equal to 1, the jump is marked as a non-subjective jump.

[0073] Embodiment 1: In this embodiment, the browsing time set corresponding to the product detail page data in the historical daily browsing data is {1, 2, 2, 3, 3, 4, 5, 6, 7, 9, 15, 16, 20}; for the same mobile terminal user, the browsing time of the product detail page data that generates a shopping order is generally greater than the browsing time of the product detail page data that does not generate a shopping order; the browsing time is in minutes;

[0074] Therefore, in this embodiment, the browsing time set of the product detail page data that generates a shopping order is {7, 15, 16, 20}; the browsing time set of the product detail page data that does not generate a shopping order is {1, 2, 2, 3, 3, 4, 5, 6, 9};

[0075] Browsing time intervention value BTIV = (1+2+2+3+4+5+6+9) ÷ 8 = 4;

[0076] If the page browsing time after the jump is 5, that is, the subjective true value STV = 5 ÷ 4 = 1.25 > 1; mark this jump as a subjective jump;

[0077] If the page browsing time after the jump is 2, that is, the subjective true value STV = 2 ÷ 4 = 0.5 is less than 1; mark this jump as a non-subjective jump;

[0078] S4: Based on the real-time operation data of mobile terminal users in the mobile terminal application software, the subjective true value of the real-time operation data is analyzed by the application software advertising analysis model, and according to the application software control model, a control instruction is output for the jump operation of the mobile terminal application software, and the control instruction is sent to the mobile terminal system execution end to control the mobile terminal.

[0079] In S4, the pre-jump page data and mobile terminal twist data in the real-time running data are substituted into the application software advertising analysis model to generate a subjective true value STV; if the subjective true value STV is greater than 1, the jump is determined to be a subjective jump and the jump is not controlled; if the subjective true value STV is less than or equal to 1, the jump is determined to be a non-subjective jump and a jump inhibition instruction is generated; the jump inhibition instruction is sent to the mobile terminal system execution end to prevent the mobile terminal from performing this jump operation.

[0080] See also Figure 2 , the present invention provides a technical solution: an artificial intelligence-based advertising delivery optimization system, the system includes an authorization data acquisition module, an application software control model construction module, an advertising analysis model construction module and an application software control module;

[0081] The authorization data collection module is used to collect historical operation data and real-time operation data in the mobile terminal application software through authorization by the mobile terminal user;

[0082] The application software control model building module is used to build an application software control model, and the application software control model is used to manage and control the jump operation of the mobile terminal application software;

[0083] Advertisement analysis model building module, used to extract application software historical operation data, analyze the indicators of historical operation data corresponding to the jump operation between application software of mobile terminal users, generate the feature combination of terminal users' historical operation data; and build application software advertisement analysis model based on the feature combination of historical operation data;

[0084] The application software control module is used to output control instructions for the jump operation of the mobile terminal application software based on the real-time operation data of the mobile terminal users in the mobile terminal application software, through the analysis of the subjective true value of the real-time operation data by the application software advertising analysis model, and according to the application software control model, send the control instructions to the mobile terminal system execution end to control the mobile terminal.

[0085] The historical operation data includes the historical operation data of shopping software and the historical operation data of non-shopping software; the historical operation data of non-shopping software includes the historical pre-jump page data and mobile terminal reversal data; the historical operation data of shopping software includes the historical daily browsing data and the historical post-jump browsing data.

[0086] The application software control model is arranged inside the mobile terminal system software. The mobile terminal user performs authorization processing. When the mobile terminal user uses the mobile terminal application software, the real-time operation data of the mobile terminal application software is monitored. When the mobile terminal application software jumps between software, jump operation management and control are provided according to the application software control model; jump operation management and control include consent to jump and jump cancellation.

[0087] The advertising analysis model building module includes a historical daily browsing data splitting unit, a historical operation data indicator analysis unit and a model building unit;

[0088] The historical daily browsing data splitting unit is used to extract the timestamp of the page data before the historical jump. The system sets the time range for extracting the historical daily browsing data, and extracts the historical daily browsing data of the historical years according to the extraction time range; the historical daily browsing data includes the product details page data and corresponding browsing time data of a single product in the historical browsing records of the shopping software;

[0089] The historical operation data indicator analysis unit is used to analyze the similarity between the historical page data before the jump and the product details page data of a single product to form a page similarity value SF; generate a mobile terminal twist angle RE according to the mobile terminal twist data; generate a subjective true value STV of the jump operation according to the browsing data after the historical jump and the browsing time data corresponding to the product details page data of a single product, and form a historical operation data feature combination according to the page similarity value, the mobile terminal twist angle and the subjective true value;

[0090] The model building unit is used to utilize the combination of historical operation data features, take the page similarity value and the mobile terminal torsion angle as independent variables, and take the subjective true value as the dependent variable, to establish an application software advertising analysis model, and use the least squares method to calculate and solve the fitting coefficients in the application software advertising analysis model.

[0091] The application software control module includes a subjective true value calculation unit and an instruction generation and transmission unit;

[0092] The subjective truth value calculation unit is used to substitute the pre-jump page data and the mobile terminal twist data in the real-time operation data into the application software advertisement analysis model to generate a subjective truth value;

[0093] The instruction generation and transmission unit generates a jump inhibition instruction according to the subjective true value; and sends the jump inhibition instruction to the execution end of the mobile terminal system to prevent the current jump operation of the mobile terminal.

[0094] It should be noted that, in this article, relational terms such as first and second, etc. 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. Moreover, the terms "include", "comprise" or any other variants thereof are intended to cover non-exclusive inclusion, so that a process, method, article or device 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, method, article or device.

[0095] It will be apparent to those skilled in the art that the invention is not limited to the details of the exemplary embodiments described above and that the invention can be implemented in other specific forms without departing from the spirit or essential features of the invention. Therefore, the embodiments should be considered exemplary and non-limiting in all respects, and the scope of the invention is defined by the appended claims rather than the foregoing description, and it is intended that all variations falling within the meaning and scope of the equivalent elements of the claims be included in the invention. Any reference numeral in a claim should not be considered as limiting the claim to which it relates.

Claims

1. An artificial intelligence-based advertising optimization method, characterized in that: The method comprises the following steps: S1: The mobile terminal user authorizes the collection of historical operation data and real-time operation data in the mobile terminal application software; the historical operation data includes the historical operation data of the shopping software and the historical operation data of the non-shopping software; the historical operation data of the non-shopping software includes the historical pre-jump page data and the mobile terminal reverse data; the historical operation data of the shopping software includes the historical daily browsing data and the historical post-jump browsing data; S2: constructing an application software control model, wherein the application software control model is used to manage and control the jump operation of the mobile terminal application software; S3: extracting the historical operation data of the application software, analyzing the indicators of the historical operation data corresponding to the jump operations between the application software of the mobile terminal user, and generating the characteristic combination of the historical operation data of the terminal user; And build an application software advertising analysis model based on the combination of historical operation data characteristics; S4: Based on the real-time operation data of the mobile terminal user in the mobile terminal application software, the subjective true value of the real-time operation data is analyzed by the application software advertising analysis model, and according to the application software control model, a control instruction is output for the jump operation of the mobile terminal application software, and the control instruction is sent to the mobile terminal system execution end to control the mobile terminal; In S3, include the following: S301: extracting the timestamp of the page data before the historical jump, the system sets a time range for extracting historical daily browsing data, and extracting historical daily browsing data of the historical years according to the extraction time range; the historical daily browsing data includes the product details page data and corresponding browsing time data of a single product in the historical browsing records of the shopping software; S302: Analyze the similarity between the page data before the historical jump and the product details page data of a single product to form a page similarity value SF; generate a mobile terminal twist angle RE according to the mobile terminal twist data; generate a subjective true value STV of the jump operation according to the browsing data after the historical jump and the browsing time data corresponding to the product details page data of a single product; S303: Based on the page similarity value, the mobile terminal torsion angle and the subjective true value, a historical operation data feature combination is formed, which is recorded as DF i =(SF i , R.E. i , STV i ); Each historical jump operation corresponds to a historical operation data feature combination; in DF i represents the i-th historical operation data feature combination; SF i Represents the page similarity value in the i-th historical operation data feature combination; RE i STV represents the twist angle of the mobile terminal in the i-th historical operation data feature combination; i represents the subjective true value in the i-th historical operation data feature combination; i∈[1,I]; I represents the total number of historical operation data feature combinations; S304: Utilize the combination of historical operation data features, take the page similarity value and the mobile terminal twisting angle as independent variables, and take the subjective true value as the dependent variable, to establish an application software advertising analysis model: STV = α1×SF+α2×RE+α3; wherein α1, α2 and α3 represent fitting coefficients, and use the least squares method to calculate and solve α1, α2 and α3 in the application software advertising analysis model.

2. The method for optimizing advertising delivery based on artificial intelligence according to claim 1, characterized in that: In S2, the application software control model is arranged inside the mobile terminal system software, and the mobile terminal user performs authorization processing. When the mobile terminal user uses the mobile terminal application software, the real-time operation data of the mobile terminal application software is monitored. When the mobile terminal application software performs a jump operation between software, the jump operation management control is provided according to the application software control model; The jump operation management control includes jump approval and jump cancellation.

3. The method for optimizing advertising delivery based on artificial intelligence according to claim 2, characterized in that: In S302, the calculation of page similarity values ​​includes the following: Use text extraction technology to extract text content from the page data before the historical jump and the product details page data; Forming text feature vector groups of the historical pre-jump page data and the product details page data respectively, and calculating a first similarity FS according to the text feature vector groups of the historical pre-jump page data and the product details page data; Among them, PBG n Represents the nth text feature vector of the page data before the historical jump, HD (m，n) Represents the nth text feature vector group of the mth product detail page data; n∈[1,N]; the number of text feature vectors in the Nth text feature vector group; m∈[1,M]; M represents the number of product detail page data; Convert the images in the page data before the historical jump and the product details page data into the same color space to generate a color histogram of each image; calculate a second similarity SS based on the color histogram of the image in the page data before the historical jump and the product details page data; the second similarity is equal to the average value of the histogram intersection of the color histogram of the page data image before the historical jump and the color histogram of each product details page data image; Calculate the page similarity value SF according to the first similarity FS and the second similarity SS: SF = A1 × FS + A2 × SS; Wherein, A1 and A2 are similarity coefficients, and the similarity coefficients are system preset constants; The mobile terminal torsion angle RE is collected by a sensor in the mobile terminal; The browsing time corresponding to the product details page data that has not formed a shopping order is extracted, and the average value is calculated and recorded as the browsing time intervention value; the subjective true value STV is equal to the quotient of the page browsing time after the jump and the browsing time intervention value; wherein, if the subjective true value is greater than 1, the jump is marked as a subjective jump; if the subjective true value is less than or equal to 1, the jump is marked as a non-subjective jump.

4. The method for optimizing advertising delivery based on artificial intelligence according to claim 3, characterized in that: In S4, the pre-jump page data and mobile terminal twist data in the real-time running data are substituted into the application software advertising analysis model to generate a subjective true value STV; if the subjective true value STV is greater than 1, the jump is determined to be a subjective jump, and the jump is not controlled; if the subjective true value STV is less than or equal to 1, the jump is determined to be a non-subjective jump, and a jump inhibition instruction is generated; the jump inhibition instruction is sent to the mobile terminal system execution end to prevent the mobile terminal from performing this jump operation.

5. An artificial intelligence-based advertising delivery optimization system, wherein the system is applied to the artificial intelligence-based advertising delivery optimization method according to any one of claims 1 to 4, and is characterized in that: The system includes an authorization data acquisition module, an application software control model construction module, an advertisement analysis model construction module and an application software control module; The authorization data collection module is used to collect historical operation data and real-time operation data in the mobile terminal application software with authorization from the mobile terminal user; The application software control model building module is used to build an application software control model, and the application software control model is used to manage and control the jump operation of the mobile terminal application software; The advertising analysis model building module is used to extract the historical operation data of the application software, analyze the indicators of the historical operation data corresponding to the jump operation between the application software of the mobile terminal user, and generate the historical operation data feature combination of the terminal user; And build an application software advertising analysis model based on the combination of historical operation data characteristics; The application software control module is used to output control instructions for the jump operation of the mobile terminal application software based on the real-time operation data of the mobile terminal user in the mobile terminal application software, through the analysis of the subjective true value of the real-time operation data by the application software advertising analysis model, and according to the application software control model, and send the control instructions to the mobile terminal system execution end to control the mobile terminal.

6. The artificial intelligence-based advertising optimization system according to claim 5, characterized in that: The historical operation data includes shopping software historical operation data and non-shopping software historical operation data; the non-shopping software historical operation data includes historical pre-jump page data and mobile terminal twist data; the shopping software historical operation data includes historical daily browsing data and historical post-jump browsing data.

7. The artificial intelligence-based advertising optimization system according to claim 5, characterized in that: The application software control model is arranged inside the mobile terminal system software, and the mobile terminal user performs authorization processing. When the mobile terminal user uses the mobile terminal application software, the real-time operation data of the mobile terminal application software is monitored. When the mobile terminal application software performs a jump operation between software, the jump operation management control is provided according to the application software control model; The jump operation management control includes jump approval and jump cancellation.

8. The artificial intelligence-based advertising optimization system according to claim 5, characterized in that: The advertising analysis model building module includes a historical daily browsing data splitting unit, a historical operation data indicator analysis unit and a model building unit; The historical daily browsing data splitting unit is used to extract the timestamp of the page data before the historical jump. The system sets a time range for extracting the historical daily browsing data, and extracts the historical daily browsing data of the historical years according to the extraction time range; the historical daily browsing data includes the product details page data and the corresponding browsing time data of a single product in the historical browsing records of the shopping software; The historical operation data indicator analysis unit is used to analyze the similarity between the historical page data before the jump and the product details page data of a single product to form a page similarity value SF; generate a mobile terminal twist angle RE according to the mobile terminal twist data; generate a subjective true value STV of the jump operation according to the browsing data after the historical jump and the browsing time data corresponding to the product details page data of a single product, and form a historical operation data feature combination according to the page similarity value, the mobile terminal twist angle and the subjective true value; The model building unit is used to utilize the historical operation data feature combination, take the page similarity value and the mobile terminal twist angle as independent variables, take the subjective true value as the dependent variable, establish the application software advertising analysis model, and use the least squares method to calculate and solve the fitting coefficient in the application software advertising analysis model.

9. The artificial intelligence-based advertising optimization system according to claim 5, characterized in that: The application software control module includes a subjective truth value calculation unit and an instruction generation and transmission unit; The subjective truth value calculation unit is used to substitute the pre-jump page data and the mobile terminal twist data in the real-time operation data into the application software advertisement analysis model to generate a subjective truth value; The instruction generation and transmission unit generates a jump inhibition instruction according to the subjective true value; and sends the jump inhibition instruction to the mobile terminal system execution end to prevent the current jump operation of the mobile terminal.

Images