CN114764727A - Target crowd mining method, advertisement pushing method and device - Google Patents
Target crowd mining method, advertisement pushing method and device Download PDFInfo
- Publication number
- CN114764727A CN114764727A CN202110049739.1A CN202110049739A CN114764727A CN 114764727 A CN114764727 A CN 114764727A CN 202110049739 A CN202110049739 A CN 202110049739A CN 114764727 A CN114764727 A CN 114764727A
- Authority
- CN
- China
- Prior art keywords
- advertisement
- crowd
- target
- recommended
- internet behavior
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06Q30/00—Commerce
- G06Q30/02—Marketing; Price estimation or determination; Fundraising
- G06Q30/0201—Market modelling; Market analysis; Collecting market data
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06Q30/00—Commerce
- G06Q30/02—Marketing; Price estimation or determination; Fundraising
- G06Q30/0241—Advertisements
- G06Q30/0251—Targeted advertisements
- G06Q30/0269—Targeted advertisements based on user profile or attribute
- G06Q30/0271—Personalized advertisement
Abstract
The target population mining method obtains internet behavior data corresponding to historical conversion users of a current website, and counts the internet behavior data to obtain a significant feature matched with the internet behavior, wherein the significant feature represents the preference of the historical conversion users. And obtaining a characteristic candidate set according to the order of the importance degree of the significant characteristics from high to low, and determining the crowd covered by the characteristic candidate set. And finally, obtaining the target population according to the population covered by the feature candidate set. Compared with a population mining scheme based on a model, the scheme does not need to train the model in advance, is simple to realize, is high in timeliness based on the statistical mode, can be copied quickly, and is high in universality.
Description
Technical Field
The application relates to the technical field of artificial intelligence, in particular to a target crowd mining method, an advertisement pushing method and an advertisement pushing device.
Background
With the rapid development of internet technology and the popularization of intelligent terminals, people increasingly rely on the intelligent terminals to conduct relevant affairs, such as online shopping, online financing and the like, and in order to provide more accurate services for users, various APPs or websites need to conduct specific target crowd mining, so that the recommendation of relevant businesses is achieved.
However, in the target population mining scheme in the related art, the target population is usually obtained by adopting a model training and inference mode, for example, a model for mining the target population in the e-commerce industry can be trained in advance, and then the target population in the e-commerce industry is obtained by predicting the total population. On one hand, the magnitude of the whole population is large and can reach 10 hundred million +, the model inference process is usually very slow and cannot guarantee timeliness, and timeliness is very important in some industries, for example, in the e-commerce industry, if a user wants to buy a commodity, the user can buy the commodity within 1-2 days, and the model is updated for 2-3 days or even longer, the situation that the user just buys a certain commodity occurs, and the system recommends an advertisement of the commodity for the user.
Moreover, when people groups in different industries are mined, model parameter adjustment needs to be carried out respectively. If the target people groups of different advertisement categories are mined, a more detailed multi-task learning model and the like need to be designed, and when the advertisement categories are increased, information such as a model architecture and the like needs to be modified, so that the model-based scheme has poor universality. In addition, the model also needs to be referred after the data distribution changes.
Disclosure of Invention
In view of this, the present application provides a target population mining method, an advertisement push method and an advertisement push device, so as to solve the technical problem that the timeliness of a target population mining result is poor due to the adoption of a model inference method in the related art, and further the service recommendation accuracy is low, and the disclosed technical scheme is as follows:
in one aspect, the present application provides a target population mining method, including:
acquiring internet behavior data corresponding to a history conversion user;
the internet behavior data are counted to obtain a significant feature matched with the internet behavior, and the significant feature represents the preference of the history conversion user;
according to the sequence of the importance degrees of the significant features from high to low, obtaining a feature candidate set and obtaining a crowd covered by the feature candidate set;
and obtaining target crowds matched with the mining target according to the crowds covered by the feature candidate set.
In a possible implementation manner, the obtaining, according to the population covered by the feature candidate set, a target population matching the mining target includes:
acquiring the category of the advertisement to be recommended in the mining target;
filtering the internet behavior data of a total number of users in a website to obtain an intentional population with an intention on the category of the advertisement to be recommended, wherein the website is a website for generating the internet behavior data by the users;
And determining the crowd in the intersection of the crowd covered by the characteristic candidate set and the intention crowd as the target crowd.
In another possible implementation manner, the filtering internet behavior data of a total number of users in a website to obtain an intended crowd with an intention on the category of the advertisement to be recommended includes:
analyzing whether the internet behavior data of the user contains information matched with the category of the advertisement to be recommended;
if the Internet behavior data of the user contains information matched with the category of the advertisement to be recommended, determining that the user has an intention on the advertisement to be recommended;
if the Internet behavior data of the user does not contain information matched with the category of the advertisement to be recommended, determining that the user does not intend to the advertisement to be recommended;
and screening out users with intention to the advertisement to be recommended from the total users of the website to obtain the intention groups.
In another possible implementation manner, the obtaining a feature candidate set according to the order of the importance of the salient features from high to low includes:
calculating the preference degree of the historical conversion user to each significant feature according to the internet behavior data of the historical conversion user;
And determining the front preset number of the significant features as the feature candidate set according to the sequence of the preference degrees from high to low.
In another possible implementation manner, the calculating, according to the data of the interconnectivity behavior of the historical conversion users, a preference degree of each significant feature of the historical conversion users includes:
calculating the ratio of the user proportion with the preset internet behavior in the historical conversion users to the user proportion with the preset internet behavior in the total users of the website to obtain the preference degree of the historical conversion users for the significant features corresponding to the preset internet behavior;
and the website is the website for the history conversion user to generate the preset internet behavior.
In a second aspect, the present application further provides an advertisement recommendation method, including:
acquiring the category of the advertisement to be recommended;
obtaining a target population matched with the category, the target population being obtained according to the method of any one of the possible implementations of the first aspect;
and pushing the advertisement to be recommended to the target crowd.
In a third aspect, the present application provides a target crowd digging implement, comprising:
The first acquisition module is used for acquiring internet behavior data corresponding to a history conversion user;
the statistical module is used for counting the internet behavior data to obtain a significant feature matched with the internet behavior, and the significant feature represents the preference of the history conversion user;
the first crowd acquisition module is used for acquiring a feature candidate set according to the sequence of the importance degrees of the significant features from high to low and acquiring crowds covered by the feature candidate set;
and the target crowd determining module is used for obtaining a target crowd matched with the advertisement to be recommended according to the crowd covered by the feature candidate set.
In a fourth aspect, the present application further provides an advertisement recommendation apparatus, including:
the advertisement category acquisition module is used for acquiring the category of the advertisement to be recommended;
a target population obtaining module, configured to obtain a target population matching the category, where the target population is obtained according to the method of the third aspect;
and the advertisement pushing module is used for pushing the advertisement to be recommended to the target crowd.
In a fifth aspect, the present application further provides a server, comprising
A processor and a memory;
wherein the processor is configured to execute a program stored in the memory;
The memory is for storing a program for at least:
acquiring internet behavior data corresponding to a history conversion user;
the internet behavior data are counted to obtain a significant feature matched with the internet behavior, and the significant feature represents the preference of the history conversion user;
according to the sequence of the importance degrees of the significant features from high to low, obtaining a feature candidate set and obtaining a crowd covered by the feature candidate set;
obtaining the target population according to the population covered by the feature candidate set;
in a sixth aspect, the present application further provides a storage medium, where the storage medium stores therein computer-executable instructions, and when the computer-executable instructions are loaded and executed by a processor, the target crowd mining method according to any one of the possible implementation manners of the first aspect is implemented, or the advertisement recommendation method according to the second aspect is implemented.
According to the target population mining method, the internet behavior data corresponding to the historical conversion users in the website are obtained, the internet behavior data are counted to obtain the significant features matched with the internet behaviors, and the significant features represent the preference of the historical conversion users. And obtaining a characteristic candidate set according to the order of the importance degree of the significant characteristics from high to low, and determining the crowd covered by the characteristic candidate set. And finally, obtaining target crowds matched with the mined target according to the crowds covered by the characteristic candidate set. Compared with a population mining scheme based on a model, the scheme does not need to train the model in advance, is simple to realize, is high in timeliness based on the statistical mode, can be copied quickly, and is high in universality.
Drawings
In order to more clearly illustrate the technical solutions in the embodiments of the present application, the drawings needed to be used in the description of the embodiments are briefly introduced below, and it is obvious that the drawings in the following description are only embodiments of the present application, and it is obvious for those skilled in the art to obtain other drawings based on the provided drawings without creative efforts.
FIG. 1 is a schematic structural diagram illustrating an advertisement recommendation system according to an embodiment of the present application;
FIG. 2 is a flow chart illustrating a target crowd mining method according to an embodiment of the present application;
FIG. 3 is a flow chart illustrating a process for obtaining a population covered by a feature candidate set according to an embodiment of the present application;
FIG. 4 is a flow chart of another target crowd mining method provided by the embodiment of the application;
FIG. 5 is a flowchart illustrating a process of obtaining a category to which an advertisement to be recommended belongs according to an embodiment of the present application;
FIG. 6 is a flow chart illustrating a process for obtaining an intended population provided by an embodiment of the present application;
FIG. 7 is a flowchart illustrating an advertisement push method according to an embodiment of the present application;
fig. 8 is a schematic structural diagram illustrating a target crowd excavating device according to an embodiment of the present application;
FIG. 9 is a schematic structural diagram of another target crowd excavating device provided by an embodiment of the present application;
FIG. 10 is a schematic structural diagram illustrating an advertisement recommendation apparatus according to an embodiment of the present application;
fig. 11 shows a schematic structural diagram of a server provided in an embodiment of the present application.
Detailed Description
The method is realized based on a statistical method, specifically, internet behavior data corresponding to historical conversion users of current websites are obtained, the internet behavior data are counted to obtain a significant feature matched with the internet behaviors, and the significant feature represents the preference of the historical conversion users. And obtaining a characteristic candidate set according to the order of the importance degree of the significant characteristics from high to low, and determining the population covered by the characteristic candidate set. And finally, obtaining a target population according to the population covered by the feature candidate set. The scheme does not need to train a model in advance, is simple to realize, is high in timeliness based on a statistical mode, can be copied quickly, and is strong in universality.
Referring to fig. 1, a schematic structural diagram of an advertisement recommendation system provided in an embodiment of the present application is shown, where the system includes a client 1 and a server 2.
The client 1 refers to a web page client installed in a terminal device (e.g., a PC, a smart phone, a tablet computer, or the like), or an application program.
The server 2 converts the internet behavior data reported by the client 1 corresponding to the user according to the history to obtain the significant features matched with the internet behavior, further obtains feature candidate sets from high to low according to the importance degree of the significant features, obtains the crowd covered by the feature candidate sets, and finally determines the target crowd matched with the advertisement to be recommended according to the crowd.
And after the server 2 determines the target crowd corresponding to the advertisement to be recommended, recommending the advertisement to be recommended to the client of the target crowd.
The process of the server mining the target population will be described in detail below:
referring to fig. 2, a flowchart of a target crowd mining method provided by an embodiment of the present application is shown, where the method is applied in a server, and as shown in fig. 2, the method may include the following steps:
and S110, acquiring Internet behavior data corresponding to the history conversion user.
Taking e-commerce industry as an example, if a certain user purchases a good or service in an advertisement after seeing a certain advertisement of the e-commerce platform, the user is called a conversion user. The historical conversion user refers to a user having a conversion behavior in the historical data.
When population expansion (namely, acquisition of potential conversion users) needs to be performed on conversion users of a certain dimension, internet behavior data corresponding to historical conversion users of the dimension are acquired.
In addition, no dimension setting is needed, and in this case, all history conversion users in the website need to be acquired.
During the internet interaction process, some behavior data are necessarily generated by the user, for example, behavior data in the social APP, such as microblog attention behavior data and buffeting attention behavior data, or other behavior data capable of distinguishing different preferences of the user, such as article reading behavior data and e-commerce behavior data.
And S120, counting the internet behavior data to obtain the significant features matched with the internet behaviors.
Wherein the salient features characterize the history of the preference characteristics of the transformed user.
The behavior data of the user in the internet has obvious characteristics, for example, the user pays attention to the behavior data, the preference characteristics of the user can be obtained through the object concerned by the user, and some people like games and some people like shopping. In this case, the object of interest to the user is a salient feature.
For another example, the behavior data of the user reading the article may also distinguish the preference characteristics of the user according to the type of the article read by the user, where the type of the article may be determined according to the title of the article, the content of the article, or the author of the article.
And S130, obtaining a characteristic candidate set according to the sequence of the importance degrees of the significant characteristics from high to low, and obtaining the crowd covered by the characteristic candidate set.
In one embodiment of the present application, the importance of the salient feature may be measured by the user' S preference for the salient feature, as shown in fig. 3, S130 may include the following steps:
s131, calculating the preference degree of the history conversion user to each remarkable feature according to the Internet behavior data of the history conversion user.
In one embodiment of the present application, the process of calculating the preference of the user for a certain salient feature may include: and calculating the ratio of the user ratio with the preset internet behavior in the historical conversion users to the user ratio with the preset internet behavior in the total users of the website to obtain the preference degree of the historical conversion users for the significant features corresponding to the preset internet behavior.
The total users of the website refer to all users of the website generating internet behaviors, for example, regarding the jitter attention behavior data, and the total users refer to all users in the jitter APP.
The calculation formula of the preference degree is as follows:
preference 100 vs. the percentage of people who have a certain pre-set internet behavior among historical conversion users/the percentage of people who have the pre-set internet behavior among the total number of users of the website (equation 1)
In different application scenarios, the meaning of each parameter in formula 1 is different, for example, taking the attention behavior data of the user as an example, the calculation formula of the preference of the user for a certain attention object is as follows:
and the preference degree is 100, the ratio of the number of people who focus on the target object in the historical conversion user to the number of people who focus on the target object in the platform where the target object is located is calculated.
Of course, in other application scenarios, the calculation formula of the preference degree becomes a parameter in the application scenario, and is not described in detail here.
And S132, determining the previous preset number of significant features as a feature candidate set according to the sequence of the preference degrees from high to low.
After the preference degrees of the history transformation users to the various significant features are obtained through calculation, the first N significant features can be selected to form a feature candidate set according to the sequence of the preference degrees from high to low. Where N may be set according to actual requirements, for example, N ═ 10.
And S133, determining the crowd covered by the feature candidate set.
And if the internet behavior data of the historical conversion user is matched with each significant feature in the feature candidate set, determining that the historical conversion user is a user covered by the feature candidate set, and obtaining all users covered by the feature candidate set according to the method.
The population covered by the feature candidate set refers to a population with high consumption of advertisements matching the features included in the feature candidate set.
In the advertisement recommendation process, certain audience users naturally have impulse of consuming a specific advertisement, for example, teenagers have more preference for game advertisements and have stronger consumption capacity; for another example, women have a greater preference for apparel, skin care products, and other types of advertisements, and have greater consumer appeal. Such a crowd with a high consumption of a certain type of advertisement may be referred to as an impulsive crowd of the advertisement. In other words, the population covered by a candidate set of features may be referred to as an impulsive population of advertisements matching the candidate set of features.
In other embodiments of the present application, other parameters may also be used to characterize the importance of the salient features, for example, a kini index may be used to measure the importance of the salient features.
And S140, obtaining target crowds matched with the mined target according to the crowds covered by the characteristic candidate set.
The mining target can be set according to actual requirements, for example, in an application scene recommended by e-commerce advertisements, corresponding target crowds can be respectively mined aiming at different categories of conversion goods or services of historical conversion users in the mining process of the target crowds, for example, for game related categories, crowd mining is performed according to behavior data of the historical conversion users of the game related categories to obtain the target crowds matched with the game categories.
Taking an excavation target as an e-commerce advertisement as an example, in an application scene, if the category of the advertisement to be recommended is not specified, under the condition, the determined impulsive crowd is the target crowd.
In another application scenario, if a category of an advertisement to be recommended is specified, a historical conversion user corresponding to the category is obtained, and then a crowd covered by a feature candidate set is obtained from the historical conversion user. And then people with intention in the category of the advertisement to be recommended are screened out from the total number of users of the website, and then the people corresponding to the intersection of the two people, namely the target people, are determined.
According to the target population mining method, the internet behavior data corresponding to the historical conversion users of the current website are obtained, the internet behavior data are counted to obtain the significant features matched with the internet behaviors, and the significant features represent the preference of the historical conversion users. And obtaining a characteristic candidate set according to the order of the importance degree of the significant characteristics from high to low, and determining the crowd covered by the characteristic candidate set. And finally, obtaining target crowds matched with the mined target according to the crowds covered by the characteristic candidate set. Compared with a population mining scheme based on a model, the scheme does not need to train the model in advance, is simple to realize, is high in timeliness based on the statistical mode, can be copied quickly, and is high in universality.
Referring to fig. 4, a flowchart of another target crowd mining method provided in the embodiment of the present application is shown, where the embodiment describes an application scenario in which a mining target is an e-commerce advertisement, and as shown in fig. 4, the method includes the following steps:
s210, acquiring Internet behavior data corresponding to the historical conversion users matched with the types of the advertisements to be recommended.
The historical conversion user matched with the category of the advertisement to be recommended refers to a conversion user of which the category to which the conversion goods or services contained in the historical conversion data belong is matched with the category of the advertisement to be recommended.
And S220, counting the internet behavior data to obtain the significant features matched with the internet behaviors.
Wherein the salient features are used for characterizing the preference of the history conversion user.
And S230, obtaining a characteristic candidate set according to the sequence of the importance degrees of the significant characteristics from high to low, and obtaining the crowd covered by the characteristic candidate set.
S240, obtaining the category of the advertisement to be recommended.
In an embodiment of the present application, as shown in fig. 5, the process of obtaining the category to which the advertisement to be recommended belongs may include the following processes:
and S241, obtaining the name of the category to which the advertisement to be recommended belongs.
The category name of the advertisement to be recommended can be determined according to the advertisement to be recommended specified by the operation side, for example, the category of the advertisement to be recommended specified by the operation side is "frost" and the category name is "frost".
And S242, expanding the category names to obtain categories of the advertisements to be recommended.
Generalizing the obtained category names of the advertisements to be recommended to obtain similar categories, wherein the categories are used as categories of the advertisements to be recommended.
In one embodiment of the present application, the category words may be generalized and expanded by manual means, or by collaborative filtering based on the items, or by using a method such as Natural Language Processing (NLP) word segmentation tool.
NLP is an important direction in the fields of computer science and artificial intelligence. It studies various theories and methods that enable efficient communication between a person and a computer using natural language. Natural language processing is a science integrating linguistics, computer science and mathematics. Therefore, the research in this field will involve natural language, i.e. the language people use daily, so it has a close relation with the research of linguistics. Natural language processing techniques typically include text processing, semantic understanding, machine translation, robotic question and answer, knowledge mapping, and the like. The NLP technology according to the present application is a technology such as text processing and semantic understanding.
For example, if the category word is "cream," the generalized extended similar categories include: the makeup mask comprises a mask, a lotion, a night cream, a skin care product, makeup, a facial essence, a makeup base, a pre-makeup cream, an eye essence, a cleansing cream, a day cream, a sun screen, a lipstick, a blush, a toner, a face make-up cream, a foundation, a perfume, a makeup removing cream, a color makeup tray, a facial soap, makeup, a makeup removing towel, an eye mask, a male essence, a lip mask and a makeup fixing spray.
And S250, filtering the internet behavior data of all users in the website to obtain the intentional population with intention on the category of the advertisement to be recommended.
The website refers to a website where the user generates the internet behavior data, for example, for a microblog website or an APP, the internet behavior data of all users of the microblog website are obtained; for the jittering sound APP, behavior data of all users in the jittering sound APP is obtained.
And matching the Internet behavior data of all users in the website with the category of the advertisement to be recommended, and if the matching is successful, indicating that the user has consumption intention on the advertisement to be recommended.
In one embodiment of the present application, as shown in fig. 6, the process of obtaining the intended population may include the steps of:
S251, analyzing whether the Internet behavior data of the user contains information matched with the type of the advertisement to be recommended; if so, go to S252; if not, S253 is performed.
For example, for the attention behavior data of the user, whether information matched with the category of the advertisement to be recommended is contained in the attention behavior data of the user is analyzed. If the tremble note number concerned by the user is related to cosmetics and the category of the advertisement to be recommended is a category related to cosmetics, it is determined that the light beam behavior data of the user matches the category of the advertisement to be recommended.
And S252, determining that the user has the intention to treat the recommended advertisement.
And S253, determining that the user has no intention to treat the recommended advertisement.
And if the information matched with the category of the advertisement to be recommended does not exist in the internet behavior data of the user, determining that the number of internet behaviors of the user is not matched with the category of the advertisement to be recommended, and further determining that the user has no intention to the advertisement to be recommended.
And S254, screening out the people with intention on the advertisement to be recommended from all the users of the website to obtain the intended people.
The determination process of S251 is performed for all users in a website, so that users with an intention to recommend an advertisement are screened from all users in the website, and the users with an intention to recommend an advertisement constitute an intention group.
In the embodiment of the present application, S240 to S250 may be executed first, and then S210 to S230 may be executed; alternatively, S210 to S230 and S240 to S250 are executed by different threads in parallel, which is not limited in the present application.
And S260, determining the crowd in the intersection of the crowd covered by the characteristic candidate set and the intention crowd as a target crowd.
And finally, the determined target population is the user covered by the intersection of the impulsive population of the advertisement to be recommended and the intentional population with the intention of the advertisement to be recommended.
According to the target crowd mining method provided by the embodiment, for historical conversion users matched with the types of the advertisements to be recommended in a website, internet behavior data of the historical conversion users are counted to obtain significant features matched with internet behaviors, and then according to the sequence feature candidate sets of which the importance degrees are from high to low, the crowd covered by the feature candidate sets, namely the impulsive crowd of the advertisements to be recommended, is determined. Meanwhile, users with intentions to be recommended advertisements, namely, intended crowds, are screened out from the total users of the website based on the internet behavior data of the total users of the website. And finally, taking intersection of impulsive population and intention population of the advertisement to be recommended to obtain population, namely target population of the advertisement to be recommended. The scheme utilizes a statistical method, does not need to train a model in advance, is simple to realize, has high time efficiency of the statistical method, can be quickly copied, and has strong universality.
The target crowd mining method can be executed in an off-line mode, after the target crowd corresponding to certain types of advertisements is obtained, the advertisements of related types can be pushed to the target crowd, and the conversion rate of the advertisements can be improved.
Referring to fig. 7, a flowchart of an advertisement recommendation method provided by an embodiment of the present application is shown, where the method is applied in a server, and as shown in fig. 7, the method may include the following steps:
s310, the server obtains the category of the advertisement to be recommended.
The category of the advertisement to be recommended is a category to which goods or services in the advertisement belong, for example, if the category of the advertisement to be recommended is a certain brand of face cream, the category of the advertisement to be recommended is "face cream".
The advertisement to be recommended can be selected according to actual requirements.
And S320, the server acquires the target crowd matched with the category.
And the server utilizes the target crowd matched with the category of the advertisement to be recommended after the target crowd mining method.
S330, the server pushes the advertisements to be recommended to the client of the target crowd.
And binding the advertisement to be recommended with the target crowd so as to push the recommended advertisement to the target crowd.
And S340, displaying the advertisement to be recommended by the client of the target crowd.
For example, it is required to optimize the advertisement of the detergent under the commercial advertisement, and the target population mined to obtain the detergent by the above-mentioned target population mining method is defined as L27. And configuring an advertiser account for the detergent advertisement to determine an advertisement to recommend. In the advertisement rearrangement phase, if the user belongs to L27 and the advertisement belongs to a preconfigured advertiser account (i.e., the advertisement is a preferred recommended advertisement), ECPM (effective cost per mile) support is performed on the advertisement, that is, the advertisement is weighted so that the advertisement is ranked more forward, and finally the advertisement is pushed to the matched crowd. The online use of the target crowd package is realized through the process.
In practical application, the advertisement recommendation method provided by the embodiment is already applied to leather clothing and cashmere sweater products, and the application shows that the advertisement conversion rate of the newly added user on the leather clothing and the cashmere sweater is improved from 1.2% to 2%, so that the advertisement conversion rate can be obviously improved by using the method.
The advertisement recommendation method provided by the embodiment obtains the category of the advertisement to be recommended, obtains the target crowd matched with the category by using the target crowd mining method, and finally pushes the recommended advertisement to the target crowd. On the basis of improving the accuracy of target population, the accuracy of advertisement pushing is realized, and the conversion rate of advertisements is further improved.
Corresponding to the embodiment of the target crowd mining method, the application also provides an embodiment of a target crowd mining device.
Referring to fig. 8, a schematic structural diagram of a target crowd mining apparatus provided in an embodiment of the present application is shown, where the apparatus is applied to a server, and as shown in fig. 8, the apparatus may include:
the first obtaining module 110 is configured to obtain internet behavior data corresponding to a history conversion user.
And the statistic module 120 is configured to count the internet behavior data to obtain a significant feature matched with the internet behavior, where the significant feature represents the preference of the history conversion user.
The first crowd obtaining module 130 is configured to obtain a feature candidate set according to a descending order of the importance degree of the significant feature, and obtain a crowd covered by the feature candidate set.
In an embodiment of the application, the first person group obtaining module is specifically configured to, when obtaining the feature candidate set according to a high-to-low order of importance degrees of the salient features:
calculating the preference degree of the historical conversion user to each significant feature according to the internet behavior data of the historical conversion user;
and determining the previous preset number of the significant features as the feature candidate set according to the sequence of the preference degrees from high to low.
In another embodiment of the present application, the calculating, according to the internet behavior data of the history conversion user, a preference degree of the history conversion user for each significant feature specifically includes:
calculating the ratio of the user proportion with the preset internet behavior in the historical conversion users to the user proportion with the preset internet behavior in the total amount of users of the website, and obtaining the preference of the historical conversion users for the significant features corresponding to the preset internet behavior;
and the website is the website for the history conversion user to generate the preset internet behavior.
And the target crowd determining module 140 is configured to obtain a target crowd matched with the mining target according to the crowd covered by the feature candidate set.
In one application scenario, if the mining target does not specify a specific category, the population covered by the feature candidate set is directly determined as the target population matching the mining target.
In another application scenario, if the mining target specifies a specific category, an intention group corresponding to the specified category needs to be further acquired, and a group covered by an intersection of a group covered by the feature candidate set and the intention group is selected as a final target group.
The target crowd mining device obtains internet behavior data corresponding to historical conversion users of a current website, and counts the internet behavior data to obtain a significant feature matched with the internet behavior, wherein the significant feature represents the preference of the historical conversion users. And obtaining a characteristic candidate set according to the order of the importance degree of the significant characteristics from high to low, and determining the population covered by the characteristic candidate set. And finally, obtaining a target crowd matched with the mining target according to the crowd covered by the characteristic candidate set. Compared with a population mining scheme based on a model, the scheme does not need to train the model in advance, is simple to realize, is high in timeliness based on the statistical mode, can be copied quickly, and is high in universality.
Referring to fig. 9, a schematic structural diagram of another target crowd excavating device provided in the embodiment of the present application is shown, where the device further includes, on the basis of the embodiment shown in fig. 8:
and the advertisement category obtaining module 210 is configured to obtain a category to which an advertisement to be recommended belongs in the mining target.
And acquiring the category name of the advertisement to be recommended, and expanding the category name to obtain the category of the advertisement to be recommended.
The intention group determining module 220 is configured to filter internet behavior data of a total number of users in a website to obtain an intention group having an intention on the category of the advertisement to be recommended, where the website is a website for the users to generate the internet behavior data.
In one embodiment of the present application, the intended population determination module 220 is specifically configured to:
analyzing whether the internet behavior data of the user contains information matched with the category of the advertisement to be recommended; if yes, determining that the user has intention on the advertisement to be recommended; if not, determining that the user has no intention on the advertisement to be recommended; and screening out users with intention to the advertisement to be recommended from the total users of the website to obtain the intention groups.
In an embodiment of the present application, the target population determining module 140 is specifically configured to: and determining the crowd in the intersection of the crowd covered by the characteristic candidate set and the intention crowd as a target crowd.
The target crowd mining device provided by this embodiment, for history conversion users matched with the category of the advertisement to be recommended in a website, statistics is performed on internet behavior data of such history conversion users to obtain a significant feature matched with the internet behavior, and then according to the order feature candidate set from high to low of the significance degree of the significant feature, a crowd covered by the feature candidate set, that is, an impulsive crowd of the advertisement to be recommended, is determined. Meanwhile, users with intentions to be recommended advertisements, namely, intended crowds, are screened out from the total users of the website based on the internet behavior data of the total users of the website. And finally, taking intersection of impulsive population and intention population of the advertisement to be recommended to obtain population, namely target population of the advertisement to be recommended. The scheme utilizes a statistical method, does not need to train a model in advance, is simple to realize, has high time efficiency of the statistical method, can be quickly copied, and has strong universality.
Referring to fig. 10, a schematic structural diagram of an advertisement recommendation apparatus provided in an embodiment of the present application is shown, where the apparatus is applied to a server, and as shown in fig. 10, the apparatus includes:
an advertisement category obtaining module 310, configured to obtain a category of an advertisement to be recommended.
And a target crowd acquiring module 320, configured to acquire a target crowd matching the category.
Wherein, the target crowd is obtained according to the target crowd digging device.
And the advertisement pushing module 330 is configured to push the advertisement to be recommended to the target group.
The advertisement recommendation device provided by the embodiment obtains the category of the advertisement to be recommended, obtains the target crowd matched with the category by using the target crowd mining method, and finally pushes the recommended advertisement to the target crowd. On the basis of improving the accuracy of target population, the accuracy of advertisement pushing is realized, and the conversion rate of advertisements is further improved.
On the other hand, the present application further provides a server, referring to fig. 11, which shows a schematic structural diagram of the server of the present application, where the server of this embodiment may include: a processor 1101 and a memory 1102.
Optionally, the server may further comprise a communication interface 1103, an input unit 1104 and a display 1105 and a communication bus 1106.
The processor 1101, the memory 1102, the communication interface 1103, the input unit 1104, and the display 1105 all communicate with each other via a communication bus 1106.
In this embodiment, the processor 1101 may be a Central Processing Unit (CPU), an asic (application specific integrated circuit), a digital signal processor, an off-the-shelf programmable gate array or other programmable logic device, etc.
The processor may call a program stored in the memory 1102. In particular, the processor may perform the targeted crowd mining method or the advertisement recommendation method described above.
The memory 1102 is used for storing one or more programs, which may include program codes including computer operation instructions, and in one embodiment of the present application, the memory stores at least the programs for implementing the following functions:
acquiring internet behavior data corresponding to historical conversion users;
the internet behavior data are counted to obtain a significant feature matched with the internet behavior, and the significant feature represents the preference of the history conversion user;
according to the sequence of the importance degree of the significant features from high to low, obtaining a feature candidate set and obtaining a crowd covered by the feature candidate set;
And obtaining target crowds matched with the mining target according to the crowds covered by the feature candidate set.
In a possible implementation manner, the obtaining a target population matching a mining target according to a population covered by the feature candidate set includes:
acquiring the category of the advertisement to be recommended in the mining target;
filtering the internet behavior data of a total number of users in a website to obtain an intention group having intention on the category of the advertisement to be recommended, wherein the website is a website for generating the internet behavior data by the users;
and determining the crowd in the intersection of the crowd covered by the characteristic candidate set and the intention crowd as the target crowd.
In a possible implementation manner, the filtering internet behavior data of a total number of users in a website to obtain an intended crowd with an intention on a category of the advertisement to be recommended includes:
analyzing whether the internet behavior data of the user contains information matched with the category of the advertisement to be recommended;
if the internet behavior data of the user contains information matched with the category of the advertisement to be recommended, determining that the user has an intention on the advertisement to be recommended;
If the Internet behavior data of the user does not contain information matched with the category of the advertisement to be recommended, determining that the user does not intend to the advertisement to be recommended;
and screening out users with intention to the advertisement to be recommended from the total users of the website to obtain the intention groups.
In a possible implementation manner, the obtaining a feature candidate set according to the order of importance of the significant features from high to low includes:
calculating to obtain the preference of the historical conversion user to each significant feature according to the Internet behavior data of the historical conversion user;
and determining the front preset number of the significant features as the feature candidate set according to the sequence of the preference degrees from high to low.
In a possible implementation manner, the calculating, according to the data of the interconnectivity behavior of the historical conversion users, a preference degree of each significant feature of the historical conversion users includes:
calculating the ratio of the user proportion with the preset internet behavior in the historical conversion users to the user proportion with the preset internet behavior in the total users of the website to obtain the preference degree of the historical conversion users for the significant features corresponding to the preset internet behavior;
And the website is the website for the history conversion user to generate the preset internet behavior.
In another embodiment of the present application, the memory stores at least a program for implementing the following functions:
acquiring the category of the advertisement to be recommended;
obtaining a target population matching the category, the target population obtained according to the method of any one of claims 1-5;
and pushing the advertisement to be recommended to the target crowd.
In one possible implementation, the memory 1102 may include a program storage area and a data storage area, wherein the program storage area may store an operating system, application programs required for at least one function (such as an image playing function, etc.), and the like; the storage data area may store data created according to the use of the computer, such as user data and image data, etc.
Further, the memory 1102 may include high-speed random access memory, and may also include non-volatile memory, such as at least one magnetic disk storage device or other volatile solid-state storage device.
The communication interface 1103 may be an interface of a communication module, such as an interface of a GSM module.
The present application may also include a display 1104 and an input unit 1105, and the like.
Of course, the structure of the server shown in fig. 11 does not constitute a limitation to the server in the embodiment of the present application, and in practical applications, the server may include more or less components than those shown in fig. 11, or some components may be combined.
On the other hand, an embodiment of the present application further provides a storage medium, where computer-executable instructions are stored, and when the computer-executable instructions are loaded and executed by a processor, the embodiment of the target crowd mining method as described above is implemented, or the embodiment of the advertisement recommendation method as described above is implemented.
It should be noted that, in the present specification, the embodiments are all described in a progressive manner, each embodiment focuses on differences from other embodiments, and the same and similar parts among the embodiments may be referred to each other. For the device-like embodiment, since it is basically similar to the method embodiment, the description is simple, and for the relevant points, reference may be made to the partial description of the method embodiment.
Finally, it should also be noted that, herein, relational terms such as first and second, and the like may be used solely to distinguish one entity or action from another entity or action without necessarily requiring or implying any actual such relationship or order between such entities or actions. Also, the terms "comprises," "comprising," or any other variation thereof, are intended to cover a non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements does not include only those elements but may include other elements not expressly listed or inherent to such process, method, article, or apparatus. Without further limitation, an element defined by the phrase "comprising an … …" does not exclude the presence of other identical elements in a process, method, article, or apparatus that comprises the element.
The previous description of the disclosed embodiments is provided to enable any person skilled in the art to make or use the present invention. Various modifications to these embodiments will be readily apparent to those skilled in the art, and the generic principles defined herein may be applied to other embodiments without departing from the spirit or scope of the invention. Thus, the present invention is not intended to be limited to the embodiments shown herein but is to be accorded the widest scope consistent with the principles and novel features disclosed herein.
The above is only a preferred embodiment of the present invention, and it should be noted that, for those skilled in the art, a plurality of modifications and embellishments can be made without departing from the principle of the present invention, and these modifications and embellishments should also be regarded as the protection scope of the present invention.
Claims (10)
1. A target crowd mining method is characterized by comprising the following steps:
acquiring internet behavior data corresponding to historical conversion users;
the internet behavior data are counted to obtain a significant feature matched with the internet behavior, and the significant feature represents the preference of the history conversion user;
according to the sequence of the importance degrees of the significant features from high to low, obtaining a feature candidate set and obtaining a crowd covered by the feature candidate set;
And obtaining target crowds matched with the mining target according to the crowds covered by the feature candidate set.
2. The method of claim 1, wherein obtaining a target population matching a mined target based on the population covered by the candidate set of features comprises:
acquiring the category of the advertisement to be recommended in the mining target;
filtering the internet behavior data of a total number of users in a website to obtain an intentional population with an intention on the category of the advertisement to be recommended, wherein the website is a website for generating the internet behavior data by the users;
and determining the crowd in the intersection of the crowd covered by the characteristic candidate set and the intention crowd as the target crowd.
3. The method of claim 2, wherein the filtering internet behavior data of a total number of users in a website to obtain an intended crowd with an intention to the category of the advertisement to be recommended comprises:
analyzing whether the internet behavior data of the user contains information matched with the category of the advertisement to be recommended;
if the Internet behavior data of the user contains information matched with the category of the advertisement to be recommended, determining that the user has an intention on the advertisement to be recommended;
If the internet behavior data of the user does not contain information matched with the category of the advertisement to be recommended, determining that the user has no intention on the advertisement to be recommended;
and screening out users with intention to the advertisement to be recommended from the total users of the website, and obtaining the intention group.
4. The method according to claim 1, wherein the obtaining a feature candidate set according to the order of importance of the salient features from high to low comprises:
calculating the preference degree of the historical conversion user to each significant feature according to the internet behavior data of the historical conversion user;
and determining the previous preset number of the significant features as the feature candidate set according to the sequence of the preference degrees from high to low.
5. The method according to claim 4, wherein the calculating the preference of the historical conversion user for each significant feature according to the interconnectivity behavior data of the historical conversion user comprises:
calculating the ratio of the user proportion with the preset internet behavior in the historical conversion users to the user proportion with the preset internet behavior in the total amount of users of the website, and obtaining the preference of the historical conversion users for the significant features corresponding to the preset internet behavior;
And the website is a website for the history conversion user to generate the preset internet behavior.
6. An advertisement recommendation method, comprising:
acquiring the category of an advertisement to be recommended;
obtaining a target population matching the category, the target population obtained according to the method of any one of claims 1-5;
and pushing the advertisement to be recommended to the target crowd.
7. A target crowd excavation apparatus, comprising:
the first acquisition module is used for acquiring internet behavior data corresponding to a history conversion user;
the statistical module is used for counting the internet behavior data to obtain a significant feature matched with the internet behavior, and the significant feature represents the preference of the history conversion user;
the first crowd acquisition module is used for acquiring a characteristic candidate set according to the sequence of the importance degrees of the significant characteristics from high to low and acquiring the crowd covered by the characteristic candidate set;
and the target population determining module is used for obtaining the target population matched with the advertisement to be recommended according to the population covered by the characteristic candidate set.
8. An advertisement recommendation apparatus, comprising:
The advertisement category acquisition module is used for acquiring the category of the advertisement to be recommended;
a target crowd acquisition module for acquiring a target crowd matching the category, the target crowd being acquired according to the target crowd mining device of claim 7;
and the advertisement pushing module is used for pushing the advertisement to be recommended to the target crowd.
9. A server, comprising
A processor and a memory;
wherein the processor is configured to execute a program stored in the memory;
the memory is to store a program to at least:
acquiring internet behavior data corresponding to a history conversion user;
the internet behavior data are counted to obtain a significant feature matched with the internet behavior, and the significant feature represents the preference of the history conversion user;
according to the sequence of the importance degrees of the significant features from high to low, obtaining a feature candidate set and obtaining a crowd covered by the feature candidate set;
and obtaining the target population according to the population covered by the feature candidate set.
10. A storage medium having stored thereon computer-executable instructions that, when loaded and executed by a processor, implement the targeted crowd mining method of any one of claims 1 to 5 above, or the advertisement recommendation method of claim 6.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202110049739.1A CN114764727A (en) | 2021-01-14 | 2021-01-14 | Target crowd mining method, advertisement pushing method and device |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202110049739.1A CN114764727A (en) | 2021-01-14 | 2021-01-14 | Target crowd mining method, advertisement pushing method and device |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN114764727A true CN114764727A (en) | 2022-07-19 |
Family
ID=82363406
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202110049739.1A Pending CN114764727A (en) | 2021-01-14 | 2021-01-14 | Target crowd mining method, advertisement pushing method and device |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN114764727A (en) |
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN114936885A (en) * | 2022-07-21 | 2022-08-23 | 成都薯片科技有限公司 | Advertisement information matching pushing method, device, system, equipment and storage medium |
-
2021
- 2021-01-14 CN CN202110049739.1A patent/CN114764727A/en active Pending
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN114936885A (en) * | 2022-07-21 | 2022-08-23 | 成都薯片科技有限公司 | Advertisement information matching pushing method, device, system, equipment and storage medium |
| CN114936885B (en) * | 2022-07-21 | 2022-11-04 | 成都薯片科技有限公司 | Advertisement information matching pushing method, device, system, equipment and storage medium |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN110996110B (en) | Commodity adjusting method and device in live broadcast process | |
| Kannan | Digital marketing: A framework, review and research agenda | |
| Atlason et al. | Product design in the circular economy: Users' perception of end-of-life scenarios for electrical and electronic appliances | |
| Drott | Music as a Technology of Surveillance | |
| Joo et al. | Effects of TV advertising on keyword search | |
| Woo et al. | Media channels and consumer purchasing decisions | |
| CN111784455A (en) | Article recommendation method and recommendation equipment | |
| González-Padilla et al. | Managing entrepreneurs’ behavior personalities in digital environments: A review | |
| CN110929138A (en) | Recommendation information generation method, device, equipment and storage medium | |
| Esmeli et al. | An analyses of the effect of using contextual and loyalty features on early purchase prediction of shoppers in e-commerce domain | |
| CN114862516A (en) | Document recommendation method, storage medium, and program product | |
| Bhattacharjee et al. | An integrated machine learning and DEMATEL approach for feature preference and purchase intention modelling | |
| US20070266048A1 (en) | System and Method for Determining Affinity Profiles for Research, Marketing, and Recommendation Systems | |
| Dörtyol et al. | A review of factors affecting turkish consumer behaviour | |
| Degutis et al. | Consumers’ willingness to disclose their personal data in e-commerce: A reciprocity-based social exchange perspective | |
| CN114764727A (en) | Target crowd mining method, advertisement pushing method and device | |
| Schultz et al. | Media synergy comes of age—Part I | |
| Roy et al. | Role of artificial intelligence in gamification for the emerging markets | |
| Lertwilairatanapong | How social network influencer influences on buying process of skincare VIA online platform in Thailand. | |
| Huang et al. | Improving social media engagements on paid and non-paid advertisements: a data mining approach | |
| Lee et al. | Kakao deep reading index: Consumption time as a key factor in news curation algorithm | |
| Casabayó et al. | Using AI techniques in the grocery industry: Identifying the customers most likely to defect | |
| de Vries et al. | Consumer behaviour and marketing fundamentals for business data analytics | |
| Amberg et al. | Mobile Application Marketplaces-an Investigation from Customers' Perspective. | |
| Yoo et al. | Online Review Mining: Identification of Important Features and Their Satisfaction Levels of Mobile Phones: Identification of Important Features and Their Satisfaction Levels of Mobile Phones |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination |