Advertisement generation method based on big data

Abstract

The invention relates to the field of big data, and discloses an advertisement generation method based on big data, comprising a user information database and a browsing advertisement information database, and according to the user information database and the browsing advertisement information database, obtaining a user age, a user gender, a user active time point and an average advertisement browsing duration of a user, constructing a feature vector for each platform user, setting a monitoring period and a monitoring time interval, obtaining the number of user active time points of the platform users, calculating the average advertisement browsing duration of the platform users, classifying the duration, and executing a first screening strategy; obtaining a target user set by calculating the similarity between the first user and each platform user in the second user set, executing a second screening strategy, obtaining a recommended advertisement set according to the target user set, and recommending advertisements to the first user.

Description

Technical Field

[0001] The present invention relates to the field of big data, and in particular to an advertisement generation method based on big data. Background Art

[0002] With the development of Internet technology and the advent of the big data era, online advertising has become an important means for companies to acquire customers and enhance brand awareness. However, traditional advertising methods often adopt a wide-ranging strategy, lacking targeting, resulting in unsatisfactory advertising results. In addition, user feedback on advertising has not been fully utilized, making it impossible for advertising content to accurately meet user needs.

[0003] Existing ad generation methods have obvious deficiencies in user information collection, ad content customization, and push strategies. These methods often ignore the personalized needs and behavior patterns of users and fail to effectively use user feedback information to optimize ad content and delivery strategies. In addition, the timing and frequency of ad push also lack scientific basis, which affects the user experience and ad conversion rate.

[0004] Therefore, based on the detailed information and behavior data of users, user feature vectors are constructed to achieve personalized customization and precise push of advertising content. Through this method, companies can reach target users more effectively, improve the conversion rate of advertising, and provide users with more relevant and valuable advertising content. Summary of the invention

[0005] The present invention provides an advertisement generation method based on big data, which is used to promote the solution of the problems mentioned in the above background technology.

[0006] The present invention provides the following technical solutions: a method for generating advertisements based on big data, optionally, setting a user information database and a browsing advertisement information database;

[0007] The user information database includes user age, user gender and user active time points;

[0008] The browsing advertisement information database is used to store the average duration of browsing advertisements for each platform user;

[0009] The platform users are users in the advertising push platform;

[0010] Based on the user information database and the browsing advertisement information database, if the platform user agrees that the platform will collect personal information data, the user's age, gender, active time points, and average duration of the user's browsing advertisements will be obtained, and a feature vector will be constructed for each platform user.

[0011] If the platform user does not agree to collect personal information data, the feature vector will not be constructed for the platform user, and the advertisement with the highest number of clicks in each monitoring time period will be recommended to the platform user;

[0012] Get all platform users and form a set, denoted as set Q, Q = {q1,q2...q w};

[0013] where q w It is represented as the w-th platform user, and w is the number of platform users;

[0014] Set user feature vector;

[0015] The user feature vector is represented by q w =[a w ,s w ,t w ,e w ];

[0016] Among them, a w Represented as the age of the w-th platform user;

[0017] s w Represented as the user gender of the w-th platform user;

[0018] If the gender of the wth platform user is female, then s w The number 1 is stored in

[0019] If the gender of the wth platform user is male, then s w The number 2 is stored in

[0020] t w It is represented as the user active time of the w-th platform user. It is allowed that a platform user has multiple user active time points.

[0021] According to the user information database, the number of active time points of platform users is obtained and stored in t w middle;

[0022] e w It is represented as the average duration of viewing advertisements by the w-th platform user;

[0023] According to the browsing advertisement information database, the average browsing advertisement duration of platform users is obtained and stored in e w middle;

[0024] Set the corresponding basic weight for each feature in the user feature vector, specifically: a w The corresponding basic weight is 0.15, s w The corresponding basic weight is 0.4, t wThe corresponding basic weight is 0.4, e w The corresponding basic weight is 0.05, denoted as basic weight U = [0.15, 0.4, 0.4, 0.05];

[0025] Divide platform users into first users and second users;

[0026] The second users are all platform users except the first user, recorded as second users, and constitute a second user set;

[0027] The second user set is screened by the first screening strategy, and the screened second user set is recorded as the target user set;

[0028] According to the weight U and the user feature vector, the first screening strategy is executed to obtain the target user set;

[0029] According to the target user set, a second screening strategy is executed to obtain a recommended advertisement set;

[0030] All advertisements in the recommended advertisement set are obtained and recommended to the first user.

[0031] Optionally, the user information database is used to obtain the number of active time points of the platform users and store them in t w Including:

[0032] Set monitoring period;

[0033] The monitoring period is set to 24 hours, starting from 00:00 every day and ending at 00:00 the next day;

[0034] Set the monitoring time interval to 1h;

[0035] According to the monitoring time interval, 24 hours a day are divided into 24 monitoring time periods, namely 0:00-1:00, 1:00-2:00...23:00-0:00, and recorded as the first time period, the second time period...the twenty-fourth time period respectively;

[0036] For each monitoring period, set the number of time points to be stored in t w middle;

[0037] The time point number is: the time point number of the first time period is 1, the time point number of the second time period is 2, ... the time point number of the twenty-fourth time period is 24;

[0038] In a monitoring cycle, obtain the number of ad clicks by platform users in each monitoring period;

[0039] Set the ad click count threshold R;

[0040] During the monitoring period, if the number of ad clicks by platform users is greater than the ad click threshold R, the monitoring period is marked as an active period, and the time point corresponding to the active period is recorded;

[0041] If the number of ad clicks by platform users is less than the ad click threshold R, the monitoring period is marked as an inactive period.

[0042] Optionally, the user information database is used to obtain the number of active time points of the platform users and store them in t w Also included:

[0043] During the monitoring period, the active time periods of platform users are obtained, and the time points corresponding to the active time periods are grouped into a set, which is recorded as set Y;

[0044] Store the elements in the set Y in the feature vector t of the platform user w middle;

[0045] Set up a monitoring cycle update mechanism;

[0046] The monitoring cycle update mechanism is set to update daily;

[0047] The update time is 00:00 every day;

[0048] The update content is to clear the data in the feature vector corresponding to the platform users in the previous monitoring cycle. After clearing the data, the feature vector data of the users in the current monitoring cycle will be collected and updated at the same time.

[0049] Optionally, the average duration of viewing advertisements by platform users is obtained based on the viewing advertisement information database and stored in the e w Including:

[0050] During the monitoring period, obtain the number of times platform users browse ads and the total duration of browsing ads;

[0051] Calculate the average duration of ad viewing for platform users based on the number of ad viewings and the total duration of ad viewing;

[0052] The average duration of viewing ads is classified into:

[0053] In the first category, the average viewing time of ads is 0s-15s;

[0054] In the second category, the average viewing time of ads is 16s-60s;

[0055] The third category: the average viewing time of ads is more than 60 seconds;

[0056] If the average duration of viewing advertisements by platform users belongs to the first category, then the e w The number 1 is stored in

[0057] If the average duration of viewing advertisements by platform users belongs to the second category, then the e w The number 2 is stored in

[0058] If the average duration of viewing advertisements by platform users belongs to the third category, then the e w The number 3 is stored in .

[0059] Furthermore, the first screening strategy is executed according to the basic weight U and the user feature vector to obtain the target user set, including:

[0060] Obtain the attention of each platform user to each advertisement type and generate an attention vector;

[0061] The attention of advertisement types was obtained through an online questionnaire survey. After collecting the questionnaire data, it was analyzed to determine the average attention of each advertisement type, and the average score of each advertisement type could be calculated.

[0062] Let Z w is the attention vector of the w-th platform user to m types of advertisements, denoted as Z w =[Z w1 , Z w2 , ..., Z wn ];

[0063] set up is the average vector of all platform users’ attention to m advertising types, denoted as

[0064] According to the attention vector and the basic weight U, the dynamic weight of each feature of the platform user is calculated using the following formula:

[0065] Among them U j is the basic weight corresponding to the jth feature, N wj is the jth dynamic weight of the wth platform user, Z wi is the attention vector of the i-th advertising type of the w-th platform user, is the mean value vector of the i-th advertising type.

[0066] Optionally, executing a first screening strategy according to the basic weight U and the user feature vector to obtain a target user set further includes:

[0067] The first screening strategy is specifically:

[0068] According to the dynamic weight of each feature of the platform user, the similarity between the first user and each platform user in the second user set is calculated, and the similarity calculation formula is:

[0069] Where P is the similarity between the first user and one of the platform users in the second user set, n is the number of features in the user feature vector, q c,i is the value of the first user c on the i-th feature, U i is the basic weight corresponding to the i-th feature, N ci is the ith dynamic weight of the first user c, q v,i is the value of the i-th feature of one of the platform users v in the second user set;

[0070] According to the similarity calculation formula, the similarity between each platform user in the second user set and the first user is obtained, and the similarities are sorted from large to small to generate a sorting sequence;

[0071] Select the first to third similarities in the sorting sequence, and obtain the corresponding second users to form a target user set.

[0072] Optionally, executing a second screening strategy according to the target user set to obtain a recommended advertisement set includes:

[0073] The second screening strategy is:

[0074] Obtain all advertisement types browsed by the first user during the monitoring period and form a set, which is recorded as set L;

[0075] The advertisement type is defined according to the classification of goods or services, for example, clothes: including men's clothing, women's clothing, children's clothing, etc., home appliances: including televisions, refrigerators, washing machines and other home appliances, daily necessities: including food, beverages, personal care products, etc.;

[0076] Obtain all advertisement types browsed by each platform user in the target user set during the monitoring period and form a set, recorded as set G;

[0077] If there is one or more advertisement types in set L that are the same as those in set G, then the advertisements with the same advertisement types are removed from set G, and the set G after removal is recorded as the recommended advertisement set;

[0078] If there is no element in set L whose advertisement type is the same as that in set G, then set K is recorded as the recommended advertisement set.

[0079] The present invention has the following beneficial effects:

[0080] 1. This big data-based advertising generation method builds a feature vector for each user by analyzing the user's age, gender, active time, and advertising browsing time to achieve personalized advertising recommendations. It uses user feature vectors and dynamic weights to calculate the user's attention to different types of advertisements, improve the relevance of advertising content to user interests, identify target user sets through screening strategies, and recommend advertisements that are more likely to interest them. It also improves user engagement by providing advertising content that matches user interests and behavior patterns.

[0081] 2. The advertisement generation method based on big data constructs a feature vector for each user, making the user information more structured and quantitative, which is convenient for data analysis and processing. By encoding the gender information into numbers, 1 for female and 2 for male, the user's active time points are recorded and the user's active time is quantified, providing a basis for the best time to deliver advertisements. By setting basic weights and executing screening strategies, more relevant advertisements are recommended to the target user according to the user feature vector. Through the first screening strategy and the second screening strategy, the group matching the target user is located, so that the advertisement content received by the target user is more in line with his interests and needs.

[0082] 3. This big data-based advertising generation method identifies the user's active time period by setting a 24-hour monitoring cycle and a 1-hour monitoring time interval. By recording the number of ad clicks in each monitoring time period and comparing it with a threshold R, identifying the active time period helps optimize the advertising delivery time. Placing ads during the user's active time period can improve the relevance of the advertising content to the user's interests.

[0083] 4. This big data-based advertising generation method analyzes users' interest and participation in different advertising content by calculating the average time users spend browsing ads. The classification of average time helps evaluate the effectiveness of ads and identify which ads are more likely to attract users' attention. According to the time users spend browsing ads, the average time of browsing ads is classified into the first category, the second category, and the third category. Each category is then quantified, and the quantified data is stored in the feature vector of platform users, which helps calculate the similarity between the target user and other users.

[0084] 5. This big data-based advertising generation method obtains the attention of each platform user to each advertising type and generates an attention vector. It uses the advertising type attention obtained from online questionnaires to analyze user preferences and interests, and combines the basic weight U and user feature vector to calculate the dynamic weight of each feature, so that advertising recommendations are more in line with users' real-time interest changes.

[0085] 6. This big data-based advertising generation method calculates the similarity between the first user and each platform user in the second user set according to the dynamic weight of each feature, identifies the target user with similar interests to the first user, and constructs a target user set by selecting the three second users corresponding to the highest similarities in the sorting sequence. The advertisements that the first user is interested in may be advertisements clicked by users in the target user set. By recommending advertisements clicked by users in the target user set to the first user, the relevance of the advertisement content to the user's interests can be improved.

[0086] 7. The advertisement generation method based on big data obtains the active time period of platform users in the monitoring period, dynamically tracks the users' advertisement browsing behavior, forms a set Y of the time points corresponding to the active time period, and stores the set in the user's feature vector. A monitoring period update mechanism is set to ensure daily update of user data and maintain the timeliness of the user's feature vector. By updating the number of user active time points daily, the method can predict the user's active time period, thereby placing advertisements at the best time. Placing advertisements during the user's most active time period can reduce interference with users and increase users' attention to advertisements.

[0087] 8. This big data-based advertising generation method obtains all advertising types browsed by the first user during the monitoring period, and compares them with the browsing advertising types of users in the target user set, and eliminates advertisements of the same type as those browsed by the first user, which helps to improve the relevance of recommended advertisements and avoid duplicate recommendations. By screening advertisements that do not match the interests of the first user, advertising resources can be effectively utilized to improve the efficiency of advertising delivery. BRIEF DESCRIPTION OF THE DRAWINGS

[0088] Figure 1 It is a flow chart of the steps of the present invention. DETAILED DESCRIPTION

[0089] 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.

[0090] Embodiment 1: Setting a user information database and browsing an advertisement information database;

[0091] The user information database includes user age, user gender and user active time points;

[0092] The browsing advertisement information database is used to store the average duration of browsing advertisements for each platform user;

[0093] The platform users are users in the advertising push platform;

[0094] According to the user information database and the browsing advertisement information database, the user age, user gender, user active time point and the average duration of the user's browsing advertisement are obtained, and a feature vector is constructed for each platform user.

[0095] Optionally, obtaining the user's age, gender, active time point, and average duration of the user's browsing advertisements based on the user information database and the browsing advertisement information database, and constructing a feature vector for each platform user includes:

[0096] Get all platform users and form a set, denoted as set Q, Q = {q1,q2...q w};

[0097] where q w It is represented as the w-th platform user, and w is the number of platform users;

[0098] Set user feature vector;

[0099] The user feature vector is represented by q w =[a w ,s w ,t w ,e w ];

[0100] Among them, a w Represented as the age of the w-th platform user;

[0101] s w Represented as the user gender of the w-th platform user;

[0102] If the gender of the wth platform user is female, then s w The number 1 is stored in

[0103] If the gender of the wth platform user is male, then s w The number 2 is stored in

[0104] t w It is represented as the user active time of the w-th platform user. It is allowed that a platform user has multiple user active time points.

[0105] According to the user information database, the number of active time points of platform users is obtained and stored in t w middle;

[0106] e w It is represented as the average duration of viewing advertisements by the w-th platform user;

[0107] According to the browsing advertisement information database, the average browsing advertisement duration of platform users is obtained and stored in e w middle;

[0108] Set the corresponding basic weight for each feature in the user feature vector, specifically: a w The corresponding basic weight is 0.15, s w The corresponding basic weight is 0.4, t w The corresponding basic weight is 0.4, e w The corresponding basic weight is 0.05, denoted as basic weight U = [0.15, 0.4, 0.4, 0.05];

[0109] Divide platform users into first users and second users;

[0110] recommending an advertisement to the first user;

[0111] The second users are all platform users except the first user, recorded as second users, and constitute a second user set;

[0112] The second user set is screened by the first screening strategy, and the screened second user set is recorded as the target user set;

[0113] According to the weight U and the user feature vector, the first screening strategy is executed to obtain the target user set;

[0114] According to the target user set, a second screening strategy is executed to obtain a recommended advertisement set;

[0115] All advertisements in the recommended advertisement set are obtained and recommended to the first user.

[0116] Optionally, the user information database is used to obtain the number of active time points of the platform users and store them in t w Including:

[0117] Set monitoring period;

[0118] The monitoring period is set to 24 hours, starting from 00:00 every day and ending at 00:00 the next day;

[0119] Set the monitoring time interval to 1h;

[0120] According to the monitoring time interval, 24 hours a day are divided into 24 monitoring time periods, namely 0:00-1:00, 1:00-2:00...23:00-0:00, and recorded as the first time period, the second time period...the twenty-fourth time period respectively;

[0121] For each monitoring period, set the number of time points to be stored in t w middle;

[0122] The time point number is: the time point number of the first time period is 1, the time point number of the second time period is 2, ... the time point number of the twenty-fourth time period is 24;

[0123] In a monitoring cycle, obtain the number of ad clicks by platform users in each monitoring period;

[0124] Set the ad click count threshold R;

[0125] During the monitoring period, if the number of ad clicks by platform users is greater than the ad click threshold R, the monitoring period is marked as an active period, and the time point corresponding to the active period is recorded;

[0126] If the number of ad clicks by platform users is less than the ad click threshold R, the monitoring period is marked as an inactive period.

[0127] Optionally, the user information database is used to obtain the number of active time points of the platform users and store them in t w Also included:

[0128] During the monitoring period, the active time periods of platform users are obtained, and the time points corresponding to the active time periods are grouped into a set, which is recorded as set Y;

[0129] Store the elements in the set Y in the feature vector t of the platform user w middle;

[0130] Set up a monitoring cycle update mechanism;

[0131] The monitoring cycle update mechanism is set to update daily;

[0132] The update time is 00:00 every day;

[0133] The update content is to clear the data in the feature vector corresponding to the platform users in the previous monitoring cycle. After clearing the data, the feature vector data of the users in the current monitoring cycle will be collected and updated at the same time.

[0134] Optionally, the average duration of viewing advertisements by platform users is obtained based on the viewing advertisement information database and stored in the e w Including:

[0135] During the monitoring period, obtain the number of times platform users browse ads and the total duration of browsing ads;

[0136] Calculate the average duration of ad viewing for platform users based on the number of ad viewings and the total duration of ad viewing;

[0137] The average duration of viewing ads is classified into:

[0138] In the first category, the average viewing time of ads is 0s-15s;

[0139] In the second category, the average viewing time of ads is 16s-60s;

[0140] The third category: the average viewing time of ads is more than 60 seconds;

[0141] If the average duration of viewing advertisements by platform users belongs to the first category, then the e w The number 1 is stored in

[0142] If the average duration of viewing advertisements by platform users belongs to the second category, then the e w The number 2 is stored in

[0143] If the average duration of viewing advertisements by platform users belongs to the third category, then the e w The number 3 is stored in .

[0144] Furthermore, the first screening strategy is executed according to the basic weight U and the user feature vector to obtain the target user set, including:

[0145] Obtain the attention of each platform user to each advertisement type and generate an attention vector;

[0146] The attention of advertisement types was obtained through an online questionnaire survey. After collecting the questionnaire data, it was analyzed to determine the average attention of each advertisement type, and the average score of each advertisement type could be calculated.

[0147] Let Z w is the attention vector of the w-th platform user to m advertising types, denoted as Z w =[Z w1 , Z w2 , ..., Z wm ];

[0148] set up is the average vector of all platform users’ attention to m advertising types, denoted as

[0149] According to the attention vector and the basic weight U, the dynamic weight of each feature of the platform user is calculated using the following formula:

[0150] Among them U j is the basic weight corresponding to the jth feature, Nwj is the jth dynamic weight of the wth platform user, Z wi is the attention vector of the i-th advertising type of the w-th platform user, is the mean value vector of the i-th advertising type.

[0151] Optionally, executing a first screening strategy according to the basic weight U and the user feature vector to obtain a target user set further includes:

[0152] The first screening strategy is specifically:

[0153] According to the dynamic weight of each feature of the platform user, the similarity between the first user and each platform user in the second user set is calculated, and the similarity calculation formula is:

[0154] Where P is the similarity between the first user and one of the platform users in the second user set, n is the number of features in the user feature vector, q c,i is the value of the first user c on the i-th feature, U i is the basic weight corresponding to the i-th feature, N ci is the ith dynamic weight of the first user c, q v,i is the value of the i-th feature of one of the platform users v in the second user set;

[0155] According to the similarity calculation formula, the similarity between each platform user in the second user set and the first user is obtained, and the similarities are sorted from large to small to generate a sorting sequence;

[0156] Select the first to third similarities in the sorting sequence, and obtain the corresponding second users to form a target user set.

[0157] Optionally, executing a second screening strategy according to the target user set to obtain a recommended advertisement set includes:

[0158] The second screening strategy is:

[0159] Obtain all advertisement types browsed by the first user during the monitoring period and form a set, which is recorded as set L;

[0160] The advertisement type is defined according to the classification of goods or services, for example, clothes: including men's clothing, women's clothing, children's clothing, etc., home appliances: including televisions, refrigerators, washing machines and other home appliances, daily necessities: including food, beverages, personal care products, etc.;

[0161] Obtain all advertisement types browsed by each platform user in the target user set during the monitoring period and form a set, recorded as set G;

[0162] If there is one or more advertisement types in set L that are the same as those in set G, then the advertisements with the same advertisement types are removed from set G, and the set G after removal is recorded as the recommended advertisement set;

[0163] If there is no element in set L with the same advertisement type as that in set G, then set K is recorded as the recommended advertisement set;

[0164] 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.

[0165] The above is only a preferred embodiment of the present invention. It should be pointed out that for ordinary technicians in this technical field, several improvements and modifications can be made without departing from the technical principles of the present invention. These improvements and modifications should also be regarded as the scope of protection of the present invention.

Claims

1. A method for generating advertisements based on big data, characterized in that: include: Set up user information database and browse advertisement information database; The user information database includes user age, user gender and user active time points; The browsing advertisement information database is used to store the average duration of browsing advertisements for each platform user; The platform users are users in the advertising push platform; Get all platform users and form a set, denoted as set Q, Q = {q1,q2...q w }; where q w It is represented as the w-th platform user, and w is the number of platform users; Set user feature vector; The user feature vector is represented by q w =[a w ,s w ,t w ,e w ]; Among them, a w Represented as the age of the w-th platform user; s w Represented as the user gender of the w-th platform user; If the gender of the wth platform user is female, then s w The number 1 is stored in If the gender of the wth platform user is male, then s w The number 2 is stored in t w It is represented as the user active time of the w-th platform user. It is allowed that a platform user has multiple user active time points. According to the user information database, the number of active time points of platform users is obtained and stored in t w middle; e w It is represented as the average duration of viewing advertisements by the w-th platform user; According to the browsing advertisement information database, the average browsing advertisement duration of platform users is obtained and stored in e w middle; Set the corresponding basic weight for each feature in the user feature vector, specifically: a w The corresponding basic weight is 0.15, s w The corresponding basic weight is 0.4, t w The corresponding basic weight is 0.4, e w The corresponding basic weight is 0.05, denoted as basic weight U = [0.15, 0.4, 0.4, 0.05]; Divide platform users into first users and second users; recommending an advertisement to the first user; The second users are all platform users except the first user, recorded as second users, and constitute a second user set; The second user set is screened by the first screening strategy, and the screened second user set is recorded as the target user set; According to the weight U and the user feature vector, the first screening strategy is executed to obtain the target user set; According to the target user set, a second screening strategy is executed to obtain a recommended advertisement set; All advertisements in the recommended advertisement set are obtained and recommended to the first user.

2. The method for generating advertisements based on big data according to claim 1, characterized in that: According to the user information database, the number of active time points of platform users is obtained and stored in t w Including: Set monitoring period; The monitoring period is set to 24 hours, starting from 00:00 every day and ending at 00:00 the next day; Set the monitoring time interval to 1h; According to the monitoring time interval, 24 hours a day are divided into 24 monitoring time periods, namely 0:00-1:00, 1:00-2:00...23:00-0:00, and recorded as the first time period, the second time period...the twenty-fourth time period respectively; For each monitoring period, set the number of time points to be stored in t w middle; The time point numbers are: the time point number of the first time period is 1, the time point number of the second time period is 2, ... the time point number of the twenty-fourth time period is 24.

3. The method for generating advertisements based on big data according to claim 1, characterized in that: According to the user information database, the number of active time points of platform users is obtained and stored in t w Also included: In a monitoring cycle, obtain the number of ad clicks by platform users in each monitoring period; Set the ad click count threshold R; During the monitoring period, if the number of ad clicks by platform users is greater than the ad click threshold R, the monitoring period is marked as an active period, and the time point corresponding to the active period is recorded; If the number of ad clicks by platform users is less than the ad click threshold R, the monitoring period is marked as an inactive period.

4. The method for generating advertisements based on big data according to claim 1, characterized in that: According to the user information database, the number of active time points of platform users is obtained and stored in t w Also included: During the monitoring period, the active time periods of platform users are obtained, and the time points corresponding to the active time periods are grouped into a set, which is recorded as set Y; Store the elements in the set Y in the feature vector t of the platform user w middle; Set up a monitoring cycle update mechanism; The monitoring cycle update mechanism is set to update daily; The update time is 00:00 every day; The update content is to clear the data in the feature vector corresponding to the platform users in the previous monitoring cycle. After clearing the data, the feature vector data of the users in the current monitoring cycle will be collected and updated at the same time.

5. The method for generating advertisements based on big data according to claim 1, characterized in that: The average duration of viewing advertisements by platform users is obtained based on the viewing advertisement information database and stored in the e w Including: During the monitoring period, obtain the number of times platform users browse ads and the total duration of browsing ads; Calculate the average duration of ad viewing for platform users based on the number of ad viewings and the total duration of ad viewing; The average duration of viewing ads is classified into: In the first category, the average viewing time of ads is 0s-15s; In the second category, the average viewing time of ads is 16s-60s; The third category: the average viewing time of ads is more than 60 seconds; If the average duration of viewing advertisements by platform users belongs to the first category, then the e w The number 1 is stored in If the average duration of viewing advertisements by platform users belongs to the second category, then the e w The number 2 is stored in If the average duration of viewing advertisements by platform users belongs to the third category, then the e w The number 3 is stored in .

6. The method for generating advertisements based on big data according to claim 1, characterized in that: The first screening strategy is executed according to the basic weight U and the user feature vector to obtain the target user set, including: Obtain the attention of each platform user to each advertisement type and generate an attention vector; Let Z w is the attention vector of the w-th platform user to m advertising types, denoted as Z w =[Z w1 , Z w2 , ..., Z wm ]; set up is the average vector of all platform users’ attention to m advertising types, denoted as 7. The method for generating advertisements based on big data according to claim 1, characterized in that: The step of executing the first screening strategy according to the basic weight U and the user feature vector to obtain the target user set also includes: According to the attention vector and the basic weight U, the dynamic weight of each feature of the platform user is calculated. The calculation formula is: Among them U j is the basic weight corresponding to the jth feature, N wj is the jth dynamic weight of the wth platform user, Z wi is the attention vector of the i-th advertising type of the w-th platform user, is the mean value vector of the i-th advertising type.

8. The method for generating advertisements based on big data according to claim 1, characterized in that: The step of executing the first screening strategy according to the basic weight U and the user feature vector to obtain the target user set also includes: The first screening strategy is specifically: According to the dynamic weight of each feature of the platform user, the similarity between the first user and each platform user in the second user set is calculated. The similarity calculation formula is: Where P is the similarity between the first user and one of the platform users in the second user set, n is the number of features in the user feature vector, q c,i is the value of the first user c on the i-th feature, U i is the basic weight corresponding to the i-th feature, N ci is the ith dynamic weight of the first user c, q v,i is the value of the i-th feature of one of the platform users v in the second user set; According to the similarity calculation formula, the similarity between each platform user in the second user set and the first user is obtained, and the similarities are sorted from large to small to generate a sorting sequence; Select the first to third similarities in the sorting sequence, and obtain the corresponding second users to form a target user set.

9. The method for generating advertisements based on big data according to claim 1, characterized in that: The step of executing the second screening strategy according to the target user set to obtain the recommended advertisement set includes: The second screening strategy is: Obtain all advertisement types browsed by the first user during the monitoring period and form a set, which is recorded as set L; Obtain all advertisement types browsed by each platform user in the target user set during the monitoring period and form a set, recorded as set G; If there is one or more advertisement types in set L that are the same as those in set G, then the advertisements with the same advertisement types are removed from set G, and the set G after removal is recorded as the recommended advertisement set; If there is no element in set L whose advertisement type is the same as that in set G, then set K is recorded as the recommended advertisement set.

Images