Intelligent advertisement putting management system based on big data analysis

Abstract

The invention discloses an intelligent advertisement putting management system based on big data analysis, and relates to the technical field of big data, the intelligent advertisement putting management system comprises a management center, and the management center is connected with a multilateral acquisition module, a feature processing module, a preset putting module and an intelligent management module; performing text extraction on the advertising information to obtain advertising text information, and converting the advertising text information by setting an extraction instruction to obtain an advertising representation vector; setting a screening instruction to intercept and replace the user operation data to obtain an operation identity vector, and capturing and concluding the operation identity vector and the advertisement representation vector to obtain an advertisement putting stickiness sequence; matching and pre-selecting the putting group users according to the advertisement putting stickiness sequence to obtain an adaptive putting user set, and setting a waiting monitoring window to carry out putting monitoring replacement on the adaptive putting user set to obtain a replacement optimization advertisement sequence; the advertisement putting process is monitored in real time, and the advertisement putting accuracy and putting efficiency are improved.

Description

Technical Field

[0001] The present invention relates to the field of big data technology, and specifically to an intelligent advertising placement management system based on big data analysis. Background Art

[0002] With the rapid development of the Internet, advertising placement has become an important means of enterprise marketing. However, traditional advertising placement methods have problems such as low accuracy, difficulty in monitoring and optimizing the effects, etc. Therefore, intelligent advertising placement has emerged and gradually increased. Although it can reduce the need for manual operations through automated decision-making and optimization processes, and intelligent bidding and optimization strategies also help to reduce ineffective advertising placements and save advertising costs, intelligent advertising placement relies on a large amount of user data, which may raise concerns about privacy infringement. At the same time, there may be biases in the data or algorithms, which may lead to unfair advertising placement results. For this reason, an intelligent advertising placement management system based on big data analysis is provided herein. Through big data analysis, intelligent processing is performed on the collected comprehensive data, data privacy is protected, and intelligent management is carried out to improve the placement accuracy and alleviate unfair advertising placement results. Summary of the Invention

[0003] The object of the present invention can be achieved by the following technical solutions:

[0004] An intelligent advertising placement management system based on big data analysis includes a management center, and the management center is connected with a multi-sided collection module, a feature processing module, a preset placement module, and an intelligent management module;

[0005] The process of the multi-sided collection module collecting advertising placement information and user operation data includes:

[0006] Set a website platform collection end, and perform multi-dimensional collection on the placed advertisements through the website platform collection end to obtain advertising placement information;

[0007] Set a user collection plan for the website platform collection end, and perform planned collection on the user group through the website platform collection end based on the user collection plan to obtain the placed group of users;

[0008] Perform forward capture on the obtained placed group of users to obtain user operation data.

[0009] The process of performing characteristic conversion on the content information segment to obtain a segment representation value includes:

[0010] Set an extraction instruction for the advertisement text information, and perform segment interception on the advertisement text information according to the extraction instruction to obtain a content information segment;

[0011] Perform message conversion on the obtained content information segment to obtain a content segment signal;

[0012] Generate the content segment spectrum according to the content segment signal, extract the peaks from the content segment spectrum to obtain segment peaks, and perform frequency statistics on the content segment spectrum to obtain the segment bandwidth;

[0013] Perform characteristic merging based on the obtained segment peaks and segment bandwidth to obtain the segment characterization value.

[0014] The process of performing statistical substitution on the segment characterization value to obtain the advertisement characterization vector includes:

[0015] Obtain the segment characterization values included in the advertisement text information, perform instruction combination on the obtained segment characterization values based on the order of extraction instructions to obtain the permutation characterization value sequence;

[0016] Perform quantity statistics on the segment characterization value based on the extraction instruction to obtain the segment element quantity;

[0017] Set the original attribute vector according to the segment element quantity, and perform element exchange on the original attribute vector through the permutation characterization value sequence to obtain the advertisement characterization vector.

[0018] The process of performing replacement of the same type of attributes on the operation information segment to obtain the operation identity vector includes:

[0019] Set the screening instruction, obtain the user operation data, and perform matching and interception on the user operation data through the screening instruction to obtain the operation information segment;

[0020] Perform message conversion on the operation information segment to obtain the operation segment signal, and perform synchronous conversion on the operation segment signal to obtain the operation segment characteristic value;

[0021] Set the original identity vector, and perform identity replacement on the original identity vector according to the obtained operation segment characteristic value to obtain the operation identity vector.

[0022] The process of performing capture and induction on the operation identity vector and the advertisement characterization vector to obtain the advertisement placement stickiness sequence includes:

[0023] Perform state preprocessing on the operation identity vector and the advertisement characterization vector to obtain the standard operation vector and the standard advertisement vector;

[0024] Perform internal measurement on the standard advertisement vector through the standard operation vector to obtain the internal vector dot product;

[0025] Perform relevant capture on the standard operation vector and the standard advertisement vector according to the obtained internal vector dot product to obtain the user advertisement correlation degree;

[0026] Perform similar induction on the obtained user advertisement correlation degree based on the target audience users to obtain the advertisement placement stickiness sequence.

[0027] The process of periodically recording the set of users suitable for delivery and obtaining periodic activity data includes:

[0028] Match and preselect the users in the delivery group according to the advertising delivery stickiness sequence to obtain the set of users suitable for delivery, and the set of users suitable for delivery includes users suitable for delivery;

[0029] Conduct advertising delivery on the set of users suitable for delivery, set a waiting monitoring window, upload the obtained waiting monitoring window to the users suitable for delivery, and periodically record the users suitable for delivery according to the waiting monitoring window to obtain periodic activity data.

[0030] The process of obtaining the replacement-optimized advertisement sequence by performing replacement delivery through the improved delivery object set includes:

[0031] Conduct weight convergence on the periodic activity data to obtain the adapted activity index, and sort the adapted activity index to obtain the adapted object sequence;

[0032] Set an allocation threshold according to the obtained adapted object sequence, and perform threshold truncation on the adapted object sequence through the allocation threshold to obtain the improved delivery object set;

[0033] Perform replacement delivery on the obtained improved delivery object set to obtain the replacement-optimized advertisement sequence.

[0034] Compared with the prior art, the beneficial effects of the present invention are:

[0035] 1. Collect the advertising delivery information and user operation data, perform segment truncation on both the advertising delivery information and user operation data to obtain the content information segment and operation information segment, convert them into signals for feature extraction, and replace and update the extracted features with the constructed original vectors to obtain the advertisement representation vector and operation identity vector. Converting the advertising delivery information and user operation data into vectors representing identities through signal extraction is beneficial for protecting data privacy, preventing the leakage of user privacy, facilitating data management, and enabling data to be processed under the same standard;

[0036] 2. Capture and summarize the operation identity vector and advertisement representation vector to obtain the advertising delivery stickiness sequence, and combine and match the two vectors to facilitate the selection of advertisements and delivery objects, obtain the most suitable users for delivery, and improve the advertising revenue effect;

[0037] 3. Set a monitoring window in the placed advertisement to monitor the interactive behavior between the user and the advertisement, which can reflect the user's response behavior to the advertisement in real time. Based on the obtained response behavior, the degree of interest of the user in the pushed advertisement can be determined. When the degree of interest is low, generate a sequence of the degree of interest in the placed advertisement for the user, reduce the need for manual operation, enable equalized placement, reduce unfair advertisement placement, and enable personalized advertisement push for the user. BRIEF DESCRIPTION OF THE DRAWINGS

[0038] In order to more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the following will briefly introduce the drawings required in the description of the embodiments or the prior art. Obviously, the drawings in the following description are only some embodiments of the present invention. For those of ordinary skill in the art, other drawings can be obtained based on these drawings without creative efforts.

[0039] Figure 1 It is the schematic diagram of the present invention. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0040] The following will clearly and completely describe the technical solutions of the present invention in combination with the embodiments. Obviously, the described embodiments are only some embodiments of the present invention, rather than all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those of ordinary skill in the art without creative efforts belong to the scope of protection of the present invention.

[0041] As Figure 1 shown, an intelligent advertisement placement management system based on big data analysis includes a management center, and the management center is connected with a multi-sided collection module, a feature processing module, a preset placement module, and an intelligent management module;

[0042] The multi-sided collection module is used to collect advertisement placement information and user operation data. The specific process includes:

[0043] Set a website platform collection end, and perform multi-dimensional collection on the placed advertisement through the website platform collection end to obtain advertisement placement information;

[0044] The placed advertisement refers to the advertisement that needs to be placed to the user, which is an advertisement with independent content and meets the requirements of the placement platform. The advertisement placement information includes advertisement brand, advertisement theme information, advertisement material information, and placement platform information. Among them, the advertisement theme information indicates that the theme of this placed advertisement is promotion, new product release, or purchase link, etc., and the advertisement material information indicates the form of expression of the advertisement, such as pictures, videos, copywriting, etc.;

[0045] Perform span statistics on the obtained advertisement placement information to obtain the advertisement duration;

[0046] The span statistics represent the statistics of the advertising time of the obtained advertisements, and the advertising duration is obtained.

[0047] A user collection plan is set for the obtained website platform collection end, and based on the user collection plan, the user group is planned and collected through the website platform collection end to obtain the target users for advertising.

[0048] Furthermore, the user collection plan represents the collection plan of the advertising target corresponding to the platform where the advertisement is placed, including the collection object, collection time, collection requirements, and privacy protection measures. Among them, the privacy protection measures mean that when collecting user information through the website platform collection end, it is carried out under the condition of security and without disclosing user information, and the information of the collected users is protected to prevent the leakage of user privacy.

[0049] The target users for advertising represent the users in the advertising applicable platform, that is, the recipients who can receive advertisements. When conducting planned collection, the network accounts of the target users for advertising on the advertising platform are collected, and the obtained network accounts are associated with the corresponding target users for advertising.

[0050] Forward capture is performed on the obtained target users for advertising to obtain user operation data. The user operation data represents the historical data, browsing records, search records, interest preferences, and purchase behaviors of the users on the network platform where the advertisement is placed.

[0051] The obtained user operation data is associated with the corresponding target users for advertising.

[0052] In particular, according to the privacy protection measures in the user collection plan, the user operation data of the obtained target users for advertising is also kept confidential and the privacy information of the users will not be leaked.

[0053] The feature processing module is used to extract text from the advertising information, obtain the advertising text information, and convert the advertising text information by setting extraction instructions to obtain the advertising representation vector. The specific process includes:

[0054] Extract text from the obtained advertising information to obtain the advertising text information.

[0055] The above-mentioned text extraction refers to the text conversion of the advertised information, that is, according to the advertisement brand, advertisement theme information, advertisement material information, and advertising platform information contained in the advertised information, the text information therein is extracted to obtain the advertisement text information, that is, all the information owned by the advertisement is described in text content. For example, if the advertisement form is a video form, there are corresponding advertisement brand, advertisement theme information, advertisement material information, and advertising platform information in the obtained advertised information, and the text information corresponding to the video advertisement can be extracted according to the corresponding information, that is, the advertisement text information;

[0056] An extraction instruction is set for the obtained advertisement text information, and the advertisement text information is segmentally intercepted according to the obtained extraction instruction to obtain a content information segment;

[0057] Furthermore, the extraction instruction refers to the keyword entries set for the advertisement text information for content extraction. For example, the extraction instructions are advertisement brand, product features, and effect. Then, the advertisement brand in each advertisement text information and the features of the product corresponding to the advertisement are extracted, and the effects of this advertisement are marked, such as price reduction, promotion activities, or new product launch, etc. The segmental interception means intercepting the content corresponding to the extraction instruction in the advertisement text information according to the obtained extraction instruction, which is recorded as the content information segment, and associating the obtained content information segment with the corresponding advertisement text information;

[0058] In particular, for each extraction instruction, there is a corresponding content information segment, that is, the number of extraction instructions is equal to the number of content information segments included in each advertisement text information;

[0059] The obtained content information segment is subjected to message conversion to obtain a content segment signal;

[0060] The message conversion means converting the obtained text-form content information segment into a signal-form content segment signal;

[0061] A content segment spectrum is generated according to the obtained content segment signal, and peak extraction is performed on the obtained content segment spectrum to obtain a segment peak. Among them, the content segment spectrum represents the Fourier transform of the content segment signal, which is converted into a frequency domain form, the maximum value of the amplitude spectrum is obtained by analyzing the obtained Fourier transform result, and the frequency corresponding to the maximum value is recorded, which is the segment peak;

[0062] Frequency statistics are performed on the obtained content segment spectrum to obtain a segment bandwidth;

[0063] The frequency statistics represent calculating the square of the amplitude of the spectrum of the obtained content segment signal to obtain the segment power spectrum, obtaining the maximum value of the segment power spectrum and the frequency corresponding to the maximum value, and denoting the frequency corresponding to the maximum value as the ultimate frequency spectrum. Monitor the content segment spectrum. In the segment power spectrum, start searching downward from the ultimate frequency spectrum to obtain the first frequency point where the segment power spectrum drops to half of the ultimate frequency spectrum, denoted as the lower limit frequency point. Start searching upward from the ultimate frequency spectrum to obtain the first frequency point where the segment power spectrum drops to half of the ultimate frequency spectrum, denoted as the upper limit frequency point. Obtain the segment bandwidth based on the obtained upper limit frequency point and lower limit frequency point, where the segment bandwidth = upper limit frequency point - lower limit frequency point;

[0064] Perform characteristic merging based on the obtained segment peak value and segment bandwidth to obtain the segment characterization value, and mark the obtained segment characterization value as PD, where, , FZ represents the segment peak value, and DB represents the segment bandwidth;

[0065] Perform quantity statistics on the obtained segment characterization values based on the extraction instruction to obtain the segment element quantity;

[0066] The quantity statistics represent the number of content information segments corresponding to the extraction quality, and each content information segment corresponds to a segment characterization value. Then each extraction instruction can obtain the corresponding segment characterization value. Therefore, counting the number of segment characterization values included in the segment advertisement text information is to perform quantity statistics on the extraction instruction, which is the segment element quantity;

[0067] Set the original attribute vector according to the obtained segment element quantity. The original attribute vector is a vector with i columns in one row. The number of elements in the original attribute vector is equal to the number of segment element quantities, and all elements in the original attribute vector are 1. Mark the elements in the original attribute vector as i, where i = 1, 2, 3,..., v1, and v1 is a positive integer;

[0068] Obtain the segment characterization values included in the advertisement text information, and perform instruction combination on the obtained segment characterization values based on the order of the extraction instructions to obtain the permutation characterization value sequence;

[0069] Furthermore, the instruction combination represents arranging and combining the segment characterization values included in the advertisement text information according to the order of the content information segments corresponding to the extraction instructions to obtain the permutation characterization value sequence;

[0070] Exchange the elements of the original attribute vector according to the obtained permutation characterization value sequence to obtain the advertisement characterization vector;

[0071] The element exchange means corresponding the segment representation values to the elements in the original attribute vector one by one according to the order of the permutation representation value sequence, and replacing the elements at the corresponding positions with the segment representation values until all the elements in the original attribute vector are replaced with the segment representation values, obtaining an advertisement representation vector, and associating the obtained advertisement representation vector with the advertisement text information corresponding to the advertisement placement information.

[0072] The preset placement module is used to set a screening instruction to intercept and replace the user operation data, obtain an operation identity vector, and perform capture and induction with the advertisement representation vector to obtain an advertisement placement stickiness sequence. The specific process includes:

[0073] Set a screening instruction according to the extraction instruction. The screening instruction is a keyword extraction type set by extracting the content set by the extraction instruction from the user operation data, which is used to facilitate screening users through the advertisement placement information to obtain target users suitable for placement. Among them, according to the historical data, browsing records, search records, interest preferences, and purchase behaviors included in the user operation data, the screening instruction includes but is not limited to extracting the browsing website, extracting the purchased commodity information, and the online browsing time; in particular, since the screening instruction is set based on the extraction instruction, the number of screening instructions is equal to the number of extraction instructions;

[0074] Obtain the user operation data, and perform matching interception on the user operation data through the screening instruction to obtain an operation information segment;

[0075] The matching interception means intercepting the content corresponding to the instruction in the user operation data according to the screening instruction, which is the operation information segment, and each screening instruction has a corresponding operation information segment, that is, the number of screening instructions is equal to the number of operation information segments included in the user operation data corresponding to a group of placement users;

[0076] Perform message conversion on the obtained operation information segment to obtain an operation segment signal;

[0077] Perform synchronous conversion according to the obtained operation segment signal to obtain an operation segment eigenvalue;

[0078] It should be further noted that in the specific implementation process, the process of the synchronous conversion includes:

[0079] Generate an operation segment spectrum according to the obtained operation segment signal, perform peak extraction on the obtained operation segment spectrum to obtain an operation segment peak;

[0080] Perform frequency statistics on the obtained operation segment spectrum to obtain an operation segment bandwidth;

[0081] Perform characteristic merging based on the obtained peak value of the operation segment and the bandwidth of the operation segment to obtain the eigenvalue of the operation segment, and mark the obtained eigenvalue of the operation segment as YT, where, , CF represents the peak value of the operation segment, and CB represents the bandwidth of the operation segment;

[0082] Set the original identity vector according to the obtained amount of fragment elements. The original identity vector is a vector with one row and j columns, the number of elements in the vector is equal to j, and all elements in the original identity vector are 1. Mark the elements in the original identity vector as j, where j = 1, 2, 3, ……, v2, and v2 is a positive integer;

[0083] Perform identity replacement on the original identity vector according to the obtained eigenvalue of the operation segment to obtain the operation identity vector;

[0084] The identity replacement means performing instruction combination on the eigenvalue of the operation segment to obtain a sequence of permutation identity values, performing element exchange on the original identity vector according to the sequence of permutation identity values to obtain the operation identity vector, and associating the obtained operation identity vector with the user operation data;

[0085] Perform state preprocessing on the obtained operation identity vector and the advertisement representation vector to obtain a standard operation vector and a standard advertisement vector;

[0086] The process of the state preprocessing includes:

[0087] Perform vector modulus calculation on the obtained operation identity vector to obtain the operation modulus length, perform vector modulus calculation on the obtained advertisement representation vector to obtain the advertisement modulus length. The vector modulus calculation is to calculate the modulus length of the vector;

[0088] Perform scale scaling on the operation identity vector according to the obtained operation modulus length to obtain the standard operation vector. The scale scaling means dividing each element in the operation identity vector by the operation modulus length to obtain the standard operation vector. Similarly, perform scale scaling on the advertisement representation vector according to the advertisement modulus length to obtain the standard advertisement vector;

[0089] Perform internal measurement on the standard advertisement vector according to the obtained standard operation vector to obtain the internal vector dot product;

[0090] Furthermore, the process of the internal measurement includes:

[0091] Mark the obtained standard operation vector as , "→" represents a vector, m represents the number of the user operation data corresponding to the standard operation vector, m = 1, 2, 3, ……, v3, and v3 is a positive integer. Then the elements in the standard operation vector are marked as ;

[0092] Mark the obtained standard advertisement vectors, denoted as , where n represents the number of the advertisement information corresponding to the standard advertisement vector, n = 1, 2, 3, ……, v4, and v4 is a positive integer. Then the elements in the standard advertisement vector are denoted as ;

[0093] Obtain the inner vector dot product based on the obtained standard operation vector and standard advertisement vector, and mark the obtained inner vector dot product as , where , that is , and "m - n" represents the inner vector dot product of the nth standard advertisement vector matched by the mth standard operation vector;

[0094] Perform relevant capture on the standard operation vector and standard advertisement vector according to the obtained inner vector dot product to obtain the user - advertisement correlation degree;

[0095] Mark the obtained user - advertisement correlation degree as , where , represents the operation norm, represents the advertisement norm;

[0096] Based on the user group of the delivery, perform similar induction on the obtained user - advertisement correlation degree to obtain the advertisement delivery stickiness sequence;

[0097] The so - called similar induction means randomly select one in the advertisement delivery information as the pre - delivery advertisement, obtain the advertisement characterization vector corresponding to the pre - delivery advertisement, denoted as the pre - delivery characterization vector, obtain the user - advertisement correlation degree containing the pre - delivery characterization vector in the obtained user - advertisement correlation degree, and perform threshold screening on the user - advertisement correlation degree, that is, remove the user - advertisement correlation degree less than or equal to zero, and sort the remaining user - advertisement correlation degrees in descending order to obtain the advertisement delivery stickiness sequence, which represents the fitness ranking between the user group of the delivery and the pre - delivery advertisement. That is, the higher the ranking, the greater the use stickiness after the pre - delivery advertisement is delivered to the user, that is, the greater the user's interest in the pre - delivery advertisement. On the contrary, the smaller the user's interest in the pre - delivery advertisement, the less suitable it is to deliver the pre - delivery advertisement to this user.

[0098] The intelligent management module is used to perform matching pre - selection on the user group of the delivery according to the advertisement delivery stickiness sequence to obtain the set of adapted delivery users, set a waiting monitoring window to perform delivery monitoring and replacement on the set of adapted delivery users to obtain the replacement - optimized advertisement sequence. The specific process includes:

[0099] Perform matching pre - selection on the user group of the delivery according to the obtained advertisement delivery stickiness sequence to obtain the set of adapted delivery users, and the set of adapted delivery users includes suitable users;

[0100] The above-mentioned matching pre-selection means pre-screening the objects suitable for the pre-invested advertisement according to the advertisement placement stickiness sequence corresponding to the pre-invested advertisement, denoted as suitable users for investment. The users expected to be invested in by the pre-invested advertisement are counted as the set of users suitable for investment, indicating that the interest degree of the users in the set of users suitable for investment in the pre-invested advertisement is qualified and they are the objects that can be invested;

[0101] Perform advertisement investment on the obtained set of users suitable for investment, and set a waiting monitoring window. Upload the obtained waiting monitoring window to the suitable users, and perform periodic recording on the suitable users according to the obtained waiting monitoring window to obtain periodic activity data. The periodic activity data includes click-through rate, user stay time, user bounce rate, and user interaction rate;

[0102] It should be further noted that in the specific implementation process, the process of the periodic recording includes:

[0103] The waiting monitoring window is a monitoring window set for the pre-invested advertisement to monitor the response behavior of users on the interface of the suitable users during the advertisement duration of the pre-invested advertisement. The response behavior refers to the behavioral operations performed by the suitable users after receiving the pre-invested advertisement. For example, after clicking on the advertisement link, downloading, purchasing, or registering, the time point of closing the advertisement interface, and interactive behaviors such as liking and collecting the advertisement. Then, the periodic activity data is obtained according to the obtained response behavior data. The periodic activity data includes click-through rate, user stay time, user bounce rate, and user interaction rate. Among them, the click-through rate represents the proportion of completing the expected advertisement goals (such as purchasing, registering, downloading) after clicking on the advertisement. The user stay time represents the staying time of the pre-invested advertisement on the user interface of the pre-user obtained according to the time point of closing the advertisement interface. The user bounce rate represents the proportion of users leaving the page immediately after clicking on the advertisement. The user interaction rate represents the proportion of liking, commenting, and collecting. In particular, the higher the click-through rate, the longer the user stay time, the lower the user bounce rate, and the higher the user interaction rate, it means that this pre-invested advertisement conforms to the interest direction of the suitable users, that is, the advertisement investment is effective;

[0104] In particular, the duration of the waiting monitoring window is the same as the advertisement duration of the pre-invested advertisement;

[0105] Perform weight aggregation on the obtained periodic activity data to obtain an adaptation activity index;

[0106] The process of the weight classification includes:

[0107] Mark the obtained periodic activity data according to the response behavior of the user to the advertisement during the advertisement duration of the pre-invested advertisement, and obtain the decentralized activity coefficient, that is, mark the click-through conversion rate, user stay time, user bounce rate, and user interaction rate included in the periodic activity data. Then mark the click-through conversion rate as k1, the user stay time as k2, the user bounce rate as k3, and the user interaction rate as k4;

[0108] Obtain the adaptation activity index according to the obtained decentralized activity coefficient, and mark the obtained adaptation activity index as , where , β1, β2, β3, and β4 are weight factors, and β1, β2, β3, and β4 are all greater than 0 and β1 + β2 + β3 + β4 = 1;

[0109] Sort the obtained adaptation activity indexes in descending order to obtain the adaptation object sequence, indicating that for the same pre-invested advertisement, there are different response behaviors when it is put to different suitable users. According to the pooled statistics of the response behaviors, obtain the ranking of the interest degree of the suitable users in the pre-invested advertisement, which is the adaptation object sequence;

[0110] Set the allocation threshold according to the obtained adaptation object sequence, and perform threshold truncation on the adaptation object sequence through the allocation threshold to obtain the improved delivery object set, and the improved delivery object set includes improved objects;

[0111] The allocation threshold means that the suitable users corresponding to the first w adaptation activity indexes in the adaptation object sequence are recorded as preferred objects, and the optimal delivery object set is constructed according to the preferred objects, and the pre-invested advertisement is continued to be delivered to the users. Then the suitable users corresponding to the adaptation activity indexes after the wth are recorded as delivery improvement objects, and the improved delivery object set is constructed according to the delivery improvement objects;

[0112] Perform replacement delivery on the obtained improved delivery object set to obtain the replacement optimization advertisement sequence, and perform optimized delivery on the improved objects according to the obtained replacement optimization advertisement sequence;

[0113] It should be further noted that in the specific implementation process, the process of the replacement delivery includes:

[0114] Arbitrarily select an improved object in the improved delivery object set and record it as the target improved user, screen in the obtained user-advertisement relevance, and obtain the user-advertisement relevance including the target improved user, which is recorded as the target advertisement relevance. That is, the delivery group users of the obtained target advertisement relevance are the same target improved user. Sort the obtained target advertisement relevance in descending order, and the obtained sequence is the ranking of the matching degree between the target improved user and different delivery advertisements, which is the replacement optimization advertisement sequence;

[0115] Optimize the targeted improved users according to the obtained replacement-optimized advertisement sequence, that is, deliver the top s ranked advertisement to the targeted improved users, so that the matching degree of the advertisements obtained by the targeted improved users is higher, which is more convenient to attract the attention and interest of users.

[0116] The preferred embodiments of the present invention disclosed above are only used to help illustrate the present invention. The preferred embodiments do not describe all the details in detail, nor do they limit the present invention to the specific embodiments. Obviously, many modifications and variations can be made according to the content of this specification. These embodiments are selected and specifically described in this specification to better explain the principles and practical applications of the present invention, so that those skilled in the art can understand and utilize the present invention well. The present invention is only limited by the claims and their full scope and equivalents.

Claims

1. An intelligent advertising delivery management system based on big data analysis, including a management center, characterized in that: The management center is connected to a multilateral collection module, a feature processing module, a preset delivery module and an intelligent management module; The multilateral collection module is used to collect advertisement information and user operation data; The feature processing module is used to extract text from the placed advertisement information to obtain advertisement text information, set extraction instructions to intercept fragments, obtain content information fragments, perform feature conversion on the content information fragments, obtain fragment representation values, set original attribute vectors, perform statistical replacement on the fragment representation values, and obtain advertisement representation vectors; The preset delivery module is used to set a screening instruction to match and intercept user operation data, obtain operation information fragments, replace the same type of attributes of the operation information fragments, obtain operation identity vectors, capture and summarize the operation identity vectors and advertisement representation vectors, and obtain an advertisement delivery sticky sequence; The intelligent management module is used to match and pre-select users of the delivery group according to the sticky sequence of the advertisement delivery, obtain a set of users suitable for delivery, set a waiting monitoring window, periodically record the set of users suitable for delivery, obtain periodic activity data, and perform allocation and interception to obtain an improved delivery object set, perform replacement delivery through the improved delivery object set, and obtain a replacement optimized advertisement sequence.

2. According to claim 1, the intelligent advertising delivery management system based on big data analysis is characterized in that: The process of the multi-party collection module collecting advertisement information and user operation data includes: Set up a website platform collection terminal, and use the website platform collection terminal to collect multi-dimensional advertisements to obtain advertisement information; Set a user collection plan for the website platform collection terminal, and plan and collect user groups through the website platform collection terminal based on the user collection plan to obtain users of the delivery group; Carry out forward capture of the obtained delivery group users to obtain user operation data.

3. The intelligent advertising delivery management system based on big data analysis according to claim 1 is characterized in that: The process of converting the characteristics of the content information fragment to obtain the fragment representation value includes: Setting an extraction instruction for the advertisement text information, and performing segment cutting on the advertisement text information according to the extraction instruction to obtain content information segments; Performing letter-to-text conversion on the obtained content information fragment to obtain a content fragment signal; Generate a content segment spectrum according to the content segment signal, extract the peak value of the content segment spectrum to obtain the segment peak value, perform frequency statistics on the content segment spectrum to obtain the segment bandwidth; The characteristics are merged according to the obtained segment peak value and segment bandwidth to obtain the segment characterization value.

4. The intelligent advertising delivery management system based on big data analysis according to claim 3 is characterized in that: The process of performing statistical replacement on the segment representation value to obtain the advertisement representation vector includes: Acquire the segment representation values ​​included in the advertisement text information, and perform instruction combination on the acquired segment representation values ​​based on the order of extraction instructions to obtain a replacement representation value sequence; Performing quantitative statistics on the fragment representation values ​​based on the extraction instruction to obtain the fragment element quantity; The original attribute vector is set according to the number of fragment elements, and the elements of the original attribute vector are exchanged by replacing the representation value sequence to obtain the advertisement representation vector.

5. The intelligent advertising delivery management system based on big data analysis according to claim 1 is characterized in that: The process of replacing the same attributes of the operation information fragments and obtaining the operation identity vector includes: Set filtering instructions to obtain user operation data, match and intercept user operation data through filtering instructions to obtain operation information fragments; Performing message-to-text conversion on the operation information fragment to obtain an operation fragment signal, and synchronously converting the operation fragment signal to obtain an operation segment characteristic value; The original identity vector is set, and the identity of the original identity vector is replaced according to the obtained operation segment characteristic value to obtain the operation identity vector.

6. The intelligent advertising delivery management system based on big data analysis according to claim 5 is characterized in that: The process of capturing and summarizing the operation identity vector and the advertisement representation vector to obtain the advertisement delivery stickiness sequence includes: Performing state preprocessing on the operation identity vector and the advertisement representation vector to obtain a standard operation vector and a standard advertisement vector; Perform internal calculations on the standard advertisement vector through the standard operation vector to obtain the internal vector dot product; The standard operation vector and the standard advertisement vector are correlated and captured according to the obtained internal vector dot product to obtain the user advertisement correlation; Based on the users in the delivery group, the obtained user-advertising relevance is classified into the same category to obtain the advertising delivery stickiness sequence.

7. The intelligent advertising delivery management system based on big data analysis according to claim 6 is characterized in that: The process of periodically recording the set of adapted delivery users and obtaining periodic activity data includes: Pre-selecting users in the delivery group according to the stickiness sequence of the advertisement delivery to obtain a set of users suitable for delivery, wherein the set of users suitable for delivery includes users suitable for delivery; Advertisements are delivered to a set of suitable delivery users, and a waiting monitoring window is set. The obtained waiting monitoring window is uploaded to the suitable delivery users, and periodic records are made for the suitable delivery users according to the waiting monitoring window to obtain periodic activity data.

8. The intelligent advertising delivery management system based on big data analysis according to claim 7 is characterized in that: The process of replacing delivery by improving the delivery object set and obtaining the replacement optimized ad sequence includes: Perform weighted aggregation on the periodic activity data to obtain an adaptation activity index, and sort the adaptation activity indexes to obtain an adaptation object sequence; An allocation threshold is set according to the obtained adaptation object sequence, and a threshold is intercepted on the adaptation object sequence by the allocation threshold to obtain an improved delivery object set; The obtained improved delivery object set is replaced and delivered to obtain a replacement optimized advertisement sequence.

Images