Directional advertisement pushing method based on video content

Abstract

The invention discloses an advertisement directional pushing method based on video content, relates to the technical field of advertisement pushing, and solves the technical problem that target audiences really having demands cannot be effectively reached by fully considering personalized characteristics and demand differences of each user. According to the method, information such as browsing history records, search records and use time of the push users is obtained in detail, and scientific analysis methods such as time period division, classification of advertisements of the same type and frequency calculation are applied, so that a user portrait which is more accurate than that of a traditional mode can be constructed; therefore, advertisements meeting user requirements are pushed in a targeted mode, the matching degree of the advertisements and the users is improved, invalid advertisement putting is reduced, and by analyzing software use behaviors and advertisement browsing conditions of the users in different time segments, for example, the browsing ratio of different advertisement types and the software use frequency in the target time period are calculated, so that the user experience is improved. And the advertisement pushing times and types are reasonably adjusted.

Description

Technical Field

[0001] The present invention relates to the technical field of advertisement push, and in particular to a method for targeted advertisement push based on video content. Background Art

[0002] In today's era of digital advertising, the accuracy of advertising has become a key goal pursued by the advertising industry. Traditional advertising push often adopts a more extensive approach, such as casting a wide net, without fully considering the personalized characteristics and demand differences of each user.

[0003] Publication number CN116757745A discloses an Internet-based targeted advertising data push system and method, which belongs to the field of advertising push. The targeted advertising data push system includes a driving monitoring module, a data management module, an advertising analysis module and a targeted push module. The driving monitoring module is used to intelligently monitor the advertising data of users during their travels. The data management module is used to perform distributed management of collected data and analyzed advertising data, and perform data cleaning. The advertising analysis module is used to intelligently analyze the advertising push data of users during their travels. The targeted push module is used to perform targeted advertising push to users based on the analysis results.

[0004] However, some existing advertising push methods do not fully consider the personalized characteristics and demand differences of each user when they are used, and are unable to effectively reach the target audience that really has needs, resulting in inaccurate advertising content push, causing trouble to users who do not need it. Summary of the invention

[0005] In view of the deficiencies in the prior art, the present invention provides a method for targeted advertising push based on video content, which solves the problem of failing to effectively reach the target audience with real needs while fully considering the personalized characteristics and demand differences of each user.

[0006] To achieve the above objectives, the present invention is implemented through the following technical solutions: a method for targeted advertising push based on video content, the method specifically comprising the following steps:

[0007] Step 1: Obtain the basic information of the push user, and the basic information includes the browsing history and search history of the push user;

[0008] Step 2: Divide the browsing history records into periods, and classify the types of advertisements browsed by the push user, and calculate the browsing frequency of the push user corresponding to different types of advertisements within the time period, and generate the browsing value of the same type;

[0009] Step 3: Analyze the search history of the push user, calculate the corresponding search value based on the user search history, and calculate the intention value of the push user for the same type of advertisements in combination with the same type of browsing value, and sort the intention values ​​from large to small to generate sorting information;

[0010] Step 4: Analyze the advertisement type according to the obtained intention value, and determine the relevant advertisement and the non-relevant advertisement based on the advertisement content, and obtain the usage time of the push user, and segment the usage time to obtain the time segment information;

[0011] Step 5: Perform push analysis based on the obtained time segment information and related ads and non-related ads, and perform comprehensive analysis based on the software usage frequency and push browsing status corresponding to different time segments of the push users to generate advertising push information.

[0012] As a further solution of the present invention, the specific method of generating the same type of browsing value in step 2 is:

[0013] Obtain the browsing history of the push user, divide the browsing history into multiple time periods based on time T, and select a group of time periods as the target period, and then analyze the browsing history corresponding to the target object;

[0014] Get the corresponding advertisement type in the browsing history, and classify different advertisements into the same type to obtain classified advertisement information, and record the advertisement type number as i, and i=1, 2, ..., j, where j represents the number of advertisement types. Then obtain the number of views of advertisement type i in the target period and record it as Ci, and calculate the total number of views corresponding to all advertisement types and record it as C1. At the same time, calculate the browsing frequency corresponding to the push user and record it as P, and record it as the same type browsing value. Then analyze the browsing situation of the natural push advertisements corresponding to the push user.

[0015] As a further solution of the present invention, the specific method of analyzing the browsing situation of the natural push advertisement corresponding to the push user in step 2 is:

[0016] Perform a same-type analysis on the naturally pushed ads to obtain natural classified ad information, and determine whether the natural classified ad information has an ad intersection with the classified ad information. If there is an intersection, mark the corresponding ad type, and record the same-type browsing value as P+1. If there is no intersection, analyze the browsing history records in the same way and generate a same-type browsing value.

[0017] As a further solution of the present invention, the specific method of calculating the search value of the user search record in step 3 is:

[0018] Obtain all search records within the target period of the push user, and classify the advertisement types corresponding to the search records into the same type to obtain classified advertisement information, and mark the corresponding search type in the classified advertisement information as n, and n=1, 2, ..., m, where m represents the number of search advertisement types, then obtain the number of searches corresponding to the search advertisement type n, record it as Ln, and obtain the number of operations corresponding to the search advertisement type n, record it as Kn, substitute the obtained parameters into the formula search value = number of operations ÷ number of searches, and calculate the search value corresponding to the search advertisement type n, record it as Sn.

[0019] As a further solution of the present invention, the specific method of generating the sorting information in step 3 is:

[0020] Determine whether there is an intersection between the search ad type and the ad type in the browsing history. If there is an intersection, calculate the sum of the search value Sn and the same type of browsing value as the intention value. If there is no intersection, record the search value as the intention value, and calculate the average of the intention values ​​corresponding to all time periods. The calculated average is used as the intention value of the same type of advertisement in the time period, and sort from large to small according to the intention value to generate sorting information.

[0021] As a further solution of the present invention, the specific method of analyzing the advertisement type according to the obtained intention value in step 4 is:

[0022] Obtain all advertisement types, then analyze the contents of all advertisement types, obtain advertisement types with content association and mark them as associated advertisements, and at the same time obtain the advertisement type with the largest intention value among the associated advertisements in the same group as the standard;

[0023] Then, the usage time of the push user is obtained, and the usage time is divided into time segment information, and the time segment information is sorted from front to back in chronological order.

[0024] As a further solution of the present invention, the specific method of generating the push information in step 5 is:

[0025] Get the time segment information of the push user, and at the same time get any group of time segment information as the target time period, then get the number of times the push user software is used during the target time period, and calculate the software usage frequency corresponding to the push user, and at the same time get the corresponding push records within the target time period, then get the number of views of different advertising types in the push records, and calculate the views percentage of different advertising types, and perform push analysis based on the obtained views percentage and software usage frequency.

[0026] As a further solution of the present invention, the specific method of performing push analysis based on the browsing percentage value and software usage frequency obtained in step 4 is:

[0027] Generate push information to obtain the maximum browsing ratio value corresponding to the target time period, and obtain the corresponding advertisement type as the target push type, then obtain whether there is any related advertisement of the target push type, if so, push it based on the advertisement type corresponding to the maximum intention value, and obtain the original push number corresponding to the target push type, then judge the browsing ratio value, if the browsing ratio value is the largest, do not process the original push number, generate push number information, if the browsing ratio value is not the largest, add one push on the basis of the original push number, and generate push number information;

[0028] Then the target time period is evenly divided according to the number of pushes, and a push of the target push type is performed within the evenly divided target time period. Meanwhile, the remaining advertisement types are pushed from large to small according to the intention value to generate push information.

[0029] The present invention provides a method for targeted advertising push based on video content. Compared with the prior art, it has the following beneficial effects:

[0030] The present invention can construct a more accurate user portrait than the traditional method by obtaining detailed information such as the browsing history, search history and usage time of the push user, and using scientific analysis methods such as time period division, classification of the same type of advertisements and frequency calculation, so that advertisers can deeply understand the user's interest tendencies and consumption intentions, so as to push advertisements that meet the user's needs in a targeted manner, improve the matching degree between advertisements and users, and reduce invalid advertising.

[0031] By determining whether there is an intersection between the search ad type and the ad type in the browsing history, and combining the corresponding browsing frequency and search value for calculation, it is possible to more accurately assess the user's potential interest in a specific ad type. Compared with the traditional ad delivery evaluation method based on only a single factor, this multi-factor comprehensive evaluation method can more accurately determine the priority of ad delivery, allocate ad resources to ad types with higher intent values, and improve the effectiveness and conversion rate of ad delivery;

[0032] In the process of analyzing the advertisement type, by calculating the cosine similarity between the advertisement type contents (or calculating the distance by the vector conversion and the Euclidean formula in the second embodiment), the related advertisements can be determined, so that the related advertisement contents related to the user's interests can be mined, and richer and more targeted advertisement recommendations can be provided to the user;

[0033] By analyzing users' software usage behavior and ad browsing in different time segments, such as calculating the browsing percentage of different ad types and software usage frequency in the target time period, the number and type of ad push can be reasonably adjusted. BRIEF DESCRIPTION OF THE DRAWINGS

[0034] Figure 1 It is a diagram of the steps of the present invention. DETAILED DESCRIPTION

[0035] 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.

[0036] For example, see Figure 1 The present application provides a method for targeted advertising push based on video content, which specifically includes the following steps:

[0037] Step 1: Obtain the basic information of the push user, and the basic information includes the push user's browsing history and search history, and record the user's browsing behavior by placing cookies (small text files) on the user's device or using device fingerprint recognition technology.

[0038] Step 2: Divide the browsing history records into periods, and classify the types of advertisements browsed by the push users. Meanwhile, calculate the browsing frequency of the push users corresponding to different types of advertisements within the time period, and generate the browsing value of the same type.

[0039] The browsing history records of the push user are obtained, and the browsing history records are divided into multiple time periods with time T as the period, and a group of time periods is selected as the target period, and then the browsing history records corresponding to the target object are analyzed; for example, if the time period T is set to one month, then the browsing history records of the push user in the past year are collected through corresponding technical means, and according to the one-month time period, these records can be divided into 12 time period segments (each month is a segment).

[0040] Obtain the corresponding advertisement type in the browsing history, and classify different advertisements into the same type to obtain classified advertisement information, such as automobile advertisements, beauty advertisements, electronic product advertisements, travel advertisements, etc., and record the advertisement type number as i, and i=1, 2, ..., j, where j represents the number of advertisement types. Then obtain the number of views of advertisement type i in the target period and record it as Ci, and calculate the total number of views corresponding to all advertisement types and record it as C1. At the same time, calculate the browsing frequency corresponding to the push user and record it as P, and record it as the same type browsing value. The browsing frequency P here represents the browsing frequency of all types of advertisements by the push user in the time period. The specific calculation method is browsing frequency = total number of views ÷ target period;

[0041] Then, the browsing situation of the natural push ads corresponding to the push users is analyzed, and the natural push ads here are represented as randomly pushed ads. At the same time, the natural push ads are analyzed of the same type to obtain the natural classified ad information, and it is determined whether the natural classified ad information has an ad intersection with the classified ad information, and the intersection means whether there are ads of the same type. If there is an intersection, the corresponding ad type is marked, and the same type browsing value is recorded as P+1, and P+1 here means adding 1 to the same type browsing value obtained by the above analysis. For example, the ad type with an intersection is the car type, and the originally calculated same type browsing value is 4.3, then further adding 1 to obtain the final same type browsing value of 5.3. If there is no intersection, the browsing history records are analyzed in the same way and the same type browsing value is generated, and the same type browsing value generated here is the same as the above analysis of the browsing history records.

[0042] Step 3: Analyze the search history of the pushed user, and calculate the corresponding search value based on the user's search history. At the same time, calculate the intention value of the pushed user's same type of advertisement in combination with the same type of browsing value, and sort the intention value from large to small to generate sorting information.

[0043] Obtain all search records within the push user target period, and classify the advertisement types corresponding to the search records into the same type to obtain classified advertisement information, and mark the corresponding search type in the classified advertisement information as n, and n=1, 2, ..., m, where m represents the number of search advertisement types, then obtain the number of searches corresponding to search advertisement type n, record it as Ln, and obtain the number of operations corresponding to search advertisement type n, record it as Kn, and the number of operations here includes a series of operations such as purchase, click and copy, substitute the obtained parameters into the formula search value=number of operations÷number of searches, and calculate the search value corresponding to search advertisement type n, record it as Sn;

[0044] Then determine whether there is an intersection between the search ad type and the ad type in the browsing history. If there is an intersection, calculate the sum of the search value Sn and the same type of browsing value as the intention value. If there is no intersection, record the search value as the intention value, and calculate the average of the intention values ​​corresponding to all time periods. The calculated average is used as the intention value of the same type of advertisement in the time period, and the intention values ​​are sorted from large to small to generate sorting information.

[0045] Suppose that on an e-commerce platform, user a's browsing history includes browsing records of advertisements for electronic products such as mobile phones, headphones, and mobile phone cases. The corresponding browsing values ​​are mobile phone browsing value L1=30 (calculated based on browsing time, number of browsing times, etc.), headphone browsing value L2=20, and mobile phone case browsing value L3=1. At a certain moment, user a conducted a keyword search, and the search advertisement type was "tablet computer", with a search value Sn=18. Since tablet computers, mobile phones, headphones, and mobile phone cases are all electronic product advertisements and have an intersection, the intention value Y=Sn+L1+L2+L3=18+30+20+15=83.

[0046] Assume another user b, whose browsing history is mainly clothing advertisements, such as shirts, jeans, etc., and user b conducts a keyword search for "fitness equipment", and the search value Sm=25. Since there is no intersection between clothing and fitness equipment advertisements, the intention value is directly the search value Sm=25.

[0047] Step 4: Analyze the advertisement type according to the obtained intention value, and determine the relevant advertisement and the non-relevant advertisement based on the advertisement content, and at the same time obtain the usage time of the push user, and segment the usage time to obtain the time segment information.

[0048] All advertisement types are obtained, and the advertisement types here include advertisement types in browsing history, organic push advertisement types, and search advertisement types. Then, the contents of all advertisement types are analyzed, and advertisement types with content association are obtained and marked as associated advertisements. Here, the cosine similarity between the contents of advertisement types is calculated to determine the number of advertisement types included in the associated advertisements. The number of advertisement types included in the associated advertisements is multiple groups, and the advertisement type corresponding to the largest intention value in the associated advertisements in the same group is obtained as the standard.

[0049] Then, the usage time of the pushed user is obtained, and the usage time here is represented by the time period when the pushed user uses the app, and the usage time is divided to obtain time segment information, and the time segment information is sorted from front to back in chronological order.

[0050] Step 5: Perform push analysis based on the obtained time segment information and related ads and non-related ads, and perform comprehensive analysis based on the software usage frequency and push browsing status corresponding to different time segments of the push users to generate advertising push information.

[0051] Obtain the time segment information of the push user, and at the same time obtain any group of time segment information as the target time period, then obtain the number of times the push user software is used in the target time period, and calculate the software usage frequency corresponding to the push user, and the usage frequency = total usage time ÷ number of times used, and at the same time obtain the corresponding push record in the target time period, and the push record here is represented by the corresponding pushed advertisement type and push number, then obtain the number of views of different advertisement types in the push record, and calculate the browsing proportion of different advertisement types;

[0052] According to the obtained browsing percentage value and software use frequency, push analysis is performed, the browsing percentage value with the largest value corresponding to the target time period is obtained, and the corresponding advertisement type is obtained and recorded as the target push type, and then whether there is any related advertisement of the target push type is obtained. If so, the advertisement type corresponding to the largest intention value is used as the standard for push, and the original push number corresponding to the target push type is obtained at the same time, and then the browsing percentage value is judged. If the browsing percentage value has the largest value, the original push number is not processed and the push number information is generated. If the browsing percentage value is not the largest value, one push is added based on the original push number, and the push number information is generated;

[0053] Then the target time period is evenly divided according to the number of pushes, and a push of the target push type is performed within the evenly divided target time period, and the content pushed here also includes the corresponding related content. At the same time, the remaining advertising types are pushed from large to small according to the intention value to generate push information.

[0054] Embodiment 2: This embodiment is implemented on the basis of Embodiment 1, and the difference from Embodiment 1 is as follows:

[0055] In step 4, when analyzing the related content, the advertisement type is transformed into a vector, and the distance between different advertisement types is calculated by the Euclidean formula, and then the related content is obtained by filtering according to the distance.

[0056] Embodiment 3, as the embodiment 3 of the present invention, focuses on combining the implementation processes of embodiment 3 of embodiment 1 and embodiment 2 for implementation.

[0057] Some of the data in the above formulas are dimensionless and numerically calculated. Meanwhile, the contents not described in detail in this specification belong to the prior art known to those skilled in the art.

[0058] The above embodiments are only used to illustrate the technical method of the present invention rather than to limit it. Although the present invention has been described in detail with reference to the preferred embodiments, those skilled in the art should understand that the technical method of the present invention may be modified or replaced by equivalents without departing from the spirit and scope of the technical method of the present invention.

Claims

1. A method for targeted advertising push based on video content, characterized in that: The method specifically comprises the following steps: Step 1: Obtain the basic information of the push user, and the basic information includes the browsing history and search history of the push user; Step 2: Divide the browsing history records into periods, and classify the types of advertisements browsed by the push user, and calculate the browsing frequency of the push user corresponding to different types of advertisements within the time period, and generate the browsing value of the same type; Step 3: Analyze the search history of the push user, calculate the corresponding search value based on the user search history, and calculate the intention value of the push user for the same type of advertisements in combination with the same type of browsing value, and sort the intention values ​​from large to small to generate sorting information; Step 4: Analyze the advertisement type according to the obtained intention value, and determine the relevant advertisement and the non-relevant advertisement based on the advertisement content, and obtain the usage time of the push user, and segment the usage time to obtain the time segment information; Step 5: Perform push analysis based on the obtained time segment information and related ads and non-related ads, and perform comprehensive analysis based on the software usage frequency and push browsing status corresponding to different time segments of the push users to generate advertising push information.

2. A method for targeted advertising push based on video content according to claim 1, characterized in that: The specific method of generating the same type of browse value in step 2 is: Obtain the browsing history of the push user, divide the browsing history into multiple time periods based on time T, and select a group of time periods as the target period, and then analyze the browsing history corresponding to the target object; Get the corresponding advertisement type in the browsing history, and classify different advertisements into the same type to obtain classified advertisement information, and record the advertisement type number as i, and i=1, 2, ..., j, where j represents the number of advertisement types. Then obtain the number of views of advertisement type i in the target period and record it as Ci, and calculate the total number of views corresponding to all advertisement types and record it as C1. At the same time, calculate the browsing frequency corresponding to the push user and record it as P, and record it as the same type browsing value. Then analyze the browsing situation of the natural push advertisements corresponding to the push user.

3. A method for targeted advertising push based on video content according to claim 2, characterized in that: The specific method for analyzing the browsing situation of organic push ads corresponding to push users in step 2 is as follows: Perform a same-type analysis on the naturally pushed ads to obtain natural classified ad information, and determine whether the natural classified ad information has an ad intersection with the classified ad information. If there is an intersection, mark the corresponding ad type, and record the same-type browsing value as P+1. If there is no intersection, analyze the browsing history records in the same way and generate a same-type browsing value.

4. The method for targeted advertising push based on video content according to claim 1, characterized in that: The specific method of calculating the search value of the user search record in step 3 is: Obtain all search records within the target period of the push user, and classify the advertisement types corresponding to the search records into the same type to obtain classified advertisement information, and mark the corresponding search type in the classified advertisement information as n, and n=1, 2, ..., m, where m represents the number of search advertisement types, then obtain the number of searches corresponding to the search advertisement type n, record it as Ln, and obtain the number of operations corresponding to the search advertisement type n, record it as Kn, substitute the obtained parameters into the formula search value = number of operations ÷ number of searches, and calculate the search value corresponding to the search advertisement type n, record it as Sn.

5. The method for targeted advertising push based on video content according to claim 1, characterized in that: The specific method of generating the sorting information in step 3 is: Determine whether there is an intersection between the search ad type and the ad type in the browsing history. If there is an intersection, calculate the sum of the search value Sn and the same type of browsing value as the intention value. If there is no intersection, record the search value as the intention value, and calculate the average of the intention values ​​corresponding to all time periods. The calculated average is used as the intention value of the same type of advertisement in the time period, and sort from large to small according to the intention value to generate sorting information.

6. The method for targeted advertising push based on video content according to claim 1, characterized in that: The specific method of analyzing the advertisement type according to the obtained intention value in step 4 is: Obtain all advertisement types, then analyze the contents of all advertisement types, obtain advertisement types with content association and mark them as associated advertisements, and at the same time obtain the advertisement type with the largest intention value among the associated advertisements in the same group as the standard; Then, the usage time of the push user is obtained, and the usage time is divided into time segment information, and the time segment information is sorted from front to back in chronological order.

7. The method for targeted advertising push based on video content according to claim 1, characterized in that: The specific method of generating the push information in step 5 is: Get the time segment information of the push user, and at the same time get any group of time segment information as the target time period, then get the number of times the push user software is used during the target time period, and calculate the software usage frequency corresponding to the push user, and at the same time get the corresponding push records within the target time period, then get the number of views of different advertising types in the push records, and calculate the views percentage of different advertising types, and perform push analysis based on the obtained views percentage and software usage frequency.

8. The method for targeted advertising push based on video content according to claim 7, characterized in that: The specific method of performing push analysis based on the browsing percentage and software usage frequency obtained in step 4 is as follows: Generate push information to obtain the maximum browsing ratio value corresponding to the target time period, and obtain the corresponding advertisement type as the target push type, then obtain whether there is any related advertisement of the target push type, if so, push it based on the advertisement type corresponding to the maximum intention value, and obtain the original push number corresponding to the target push type, then judge the browsing ratio value, if the browsing ratio value is the largest, do not process the original push number, generate push number information, if the browsing ratio value is not the largest, add one push on the basis of the original push number, and generate push number information; Then the target time period is evenly divided according to the number of pushes, and a push of the target push type is performed within the evenly divided target time period. Meanwhile, the remaining advertisement types are pushed from large to small according to the intention value to generate push information.

Images