WO2020043001A1 - 广告投放方法、确定推广人群的方法、服务器和客户端 - Google Patents

广告投放方法、确定推广人群的方法、服务器和客户端 Download PDF

Info

Publication number
WO2020043001A1
WO2020043001A1 PCT/CN2019/101998 CN2019101998W WO2020043001A1 WO 2020043001 A1 WO2020043001 A1 WO 2020043001A1 CN 2019101998 W CN2019101998 W CN 2019101998W WO 2020043001 A1 WO2020043001 A1 WO 2020043001A1
Authority
WO
WIPO (PCT)
Prior art keywords
feature group
feature
crowd
user
salient
Prior art date
Application number
PCT/CN2019/101998
Other languages
English (en)
French (fr)
Inventor
杨子旭
王月颖
李莉
王志勇
张小洵
Original Assignee
阿里巴巴集团控股有限公司
Priority date (The priority date 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 date listed.)
Filing date
Publication date
Application filed by 阿里巴巴集团控股有限公司 filed Critical 阿里巴巴集团控股有限公司
Publication of WO2020043001A1 publication Critical patent/WO2020043001A1/zh

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
    • G06Q30/0241—Advertisements
    • G06Q30/0277—Online advertisement
    • 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

Definitions

  • the present application belongs to the field of Internet technology, and particularly relates to a method for advertising, a method for determining a promotion crowd, a server, and a client.
  • the demographic-based dimension mainly defines the population based on the user's gender, age, and purchasing power
  • the user's behavior-based dimension mainly refers to the user's browsing and clicks in the advertiser's store.
  • the purpose of this application is to provide an advertisement placing method, a method for determining a promotion crowd, a server, and a client, which can realize advertisement placement that meets the needs of advertisers, so that the wishes of advertisers can be effectively expressed.
  • This application provides an advertising method, a method for determining a promotion crowd, a server, and a client that are implemented as follows:
  • An advertising method includes:
  • a salient feature group for placing an advertisement based on a user-set seed crowd parameter, an extension parameter, and an extension target parameter, wherein the salient feature group includes a plurality of salient features and weight values of each salient feature;
  • a similar population is determined according to the adjusted significant feature group, and a promotion object for placing the advertisement is obtained.
  • a method for determining a promotion crowd includes:
  • Receive a user's selection operation for the salient feature group where the selection operation is used to adjust the salient feature group, determine a similar population based on the adjusted salient feature group, and obtain a promotion object.
  • a server includes a processor and a memory for storing processor-executable instructions.
  • processor executes the instructions, the steps of the following method are implemented:
  • a salient feature group for placing an advertisement based on a user-set seed crowd parameter, an extension parameter, and an extension target parameter, wherein the salient feature group includes a plurality of salient features and weight values of each salient feature;
  • a similar population is determined according to the adjusted significant feature group, and a promotion object for placing the advertisement is obtained.
  • a client includes a processor and a memory for storing processor-executable instructions.
  • the processor executes the instructions, the steps of the following method are implemented:
  • Receive a user's selection operation for the salient feature group where the selection operation is used to adjust the salient feature group, determine a similar population based on the adjusted salient feature group, and obtain a promotion object.
  • a computer-readable storage medium stores computer instructions thereon, the steps of the above method being implemented when the instructions are executed.
  • the advertisement placing method, the method for determining the promotion crowd, the server and the client provided in this application set an entry for users with promotion needs who can set seed crowd parameters, extension parameters, extension target parameters, and distinctive feature groups, so that The user's willingness to expand can be effectively expressed, and it solves the technical problem that the user cannot control the expansion direction caused by the existing automated method of crowd expansion and screening that is transparent to the user. In order to make the user's expansion willingness can be effectively expressed in technical effects, greatly improving the user experience.
  • FIG. 1 is a structural diagram of an advertisement placing system provided by the present application.
  • FIG. 2 is a structural block diagram of an advertisement placing system provided by the present application.
  • FIG. 3 is a method flowchart of an advertisement placing method provided by the present application.
  • FIG. 5 is a schematic diagram of setting salient features provided by the present application.
  • FIG. 6 is a method flowchart of an advertisement placing method provided by the present application.
  • FIG. 7 is a schematic structural diagram of an advertisement delivery server provided by the present application.
  • the current commonly used method is to determine a part of the target population by the advertiser, and then use the similar population expansion function to use the existing target personnel as the seed user, and find people similar to these seed users as the intentional crowd, and Ad serving.
  • the crowd is generally defined based on the dimensions of the user's gender, age, and purchasing power, while the dimension based on the user's behavior is mainly based on the behavior of the user's browsing and clicking in the advertiser's store.
  • circle method no matter which kind of circle method is adopted, there will be a problem that the number of circled users is limited and cannot meet the needs of the marketing plan.
  • the system automatically generates an expansion crowd that meets the requirements.
  • the advertiser cannot control the expansion direction of the intended crowd, and the expansion intention of the advertiser cannot be obtained. Satisfaction and reflection, the user experience is poor.
  • an advertisement placement system can be provided to provide users with a choice entrance for expressing the direction of expansion of the intended crowd, then the advertiser's expansion will be satisfied, and the final expansion result will be more in line with user needs.
  • an advertisement delivery system is provided.
  • the system may include a client 101 and a server 102.
  • the server 102 can push to the setting interface for setting the willingness to expand.
  • the user can set the desired expansion direction of the crowd through the setting interface displayed on the client.
  • the server 102 may perform crowd expansion according to the expansion intention set by the user, and perform advertisement placement on the expanded crowd.
  • the above-mentioned crowd expansion is based on the seed population to determine similar populations of the seed population to obtain a predetermined number of people.
  • the seed population is 10, and 90 people can be obtained by expanding the population, so that there are 100 people, that is, the population expansion is achieved.
  • the above server 102 may be a single server, a server cluster, or a server installed in the cloud.
  • the specific method of the server may be selected according to actual needs, which is not limited in this application. .
  • the client may be a terminal device or software used by a user for operation.
  • the client may be a terminal device such as a smart phone, tablet computer, notebook computer, desktop computer, smart watch, or other wearable device.
  • the user terminal may also be software that can run in the terminal device. For example: APP, browser and other application software.
  • the entrance for advertiser participation can be set at, but not limited to, one or more of the following nodes:
  • Advertisers can set some parameter information of the crowd they want to push on the interface for selecting the seed crowd, such as: target crowd, crowd characteristics, and so on.
  • the target group may be, for example, an untriggered user, a potential user, a mobile user, a purchase user, and the like.
  • the user may select or set which type of user the desired target group is according to needs.
  • the characteristics of the crowd may be, for example, gender, age, city, etc.
  • a reminder interface can be set to remind advertisers that they can complete the needs of advertising promotion without crowd expansion.
  • the number of crowds that advertisers want to promote is 300,000. After setting the target crowd and crowd characteristics, the number of crowds is 350,000, so there is no need to expand the crowd. If the number of people the advertiser wants to promote is 1 million, and after setting the target population and crowd characteristics, the number of people obtained is 350,000, then the population expansion is needed.
  • the accuracy and effectiveness of the users reached by the advertisement promotion can be effectively improved without causing unnecessary system Calculation can save costs and improve efficiency.
  • the formed crowd set can be used as a seed crowd for subsequent crowd promotion.
  • the advertiser can be provided with the right to select the number of expansions.
  • advertisers can use the method of entering the expansion number of the crowd, or the method of setting the expansion multiplier, which can be the method of entering numbers, the method of dragging the progress bar by the progress bar, or the setting of an expansion multiplier.
  • the number of ways for advertisers to choose, and which method is specifically used can be selected according to actual needs, which is not limited in this application.
  • advertisers can be provided with multiple expansion targets for advertisers to choose from, such as: brand promotion, performance promotion, click priority, conversion priority, and so on. That is, one or more of these extension targets may be selected by the advertiser.
  • advertisers can choose the desired expansion target according to their advertising promotion needs. For example, if the advertiser's promotion of this advertisement is mainly to achieve brand promotion, then brand promotion can be selected as the expansion target. If the advertiser hopes that the promotion of this advertisement can make more users willing to click, then click priority can be selected As an extension target.
  • expansion targets are only exemplary descriptions. In actual implementation, other expansion targets may be used, such as: purchase priority, etc., which is not limited in this application.
  • the server determines that the advertiser has selected brand promotion as the extension target, the server selects a feature group corresponding to "brand promotion", for example, the brand preference degree characteristics of each major brand. If you select “Effect promotion” and “Conversion priority” in the extended target option, you can select the feature group corresponding to “Effect promotion” and “Conversion priority”, for example, the characteristics of the purchase intention crowd in each store.
  • the expansion modeling is performed from the feature groups provided by the server, and the significant feature groups are filtered, that is, the features that have a greater impact on the expansion result are identified and displayed to Advertisers, for advertisers to choose whether to adopt these features. For features that advertisers do not want, advertisers have the right to delete.
  • a weight value can be set for each feature in the significant feature group.
  • a high weight value indicates a large impact on the expansion result, and a low weight value indicates the expansion result. Little impact.
  • the server can give a default salient feature group and the weight value of each feature in the salient feature group to the advertiser according to the seed population, the expanded target, and the expanded number of people. Advertisers can modify the weight value of each feature according to their own needs and wishes to control the expansion results of the crowd, that is, to achieve more fine-grained control.
  • a significant feature group When generating a significant feature group, it can be generated in one or more of the following ways, but is not limited to: point mutual information index, information gain, chi-square test value, and the like. You can choose one of them to generate a significant feature group and the recommended weight value of each feature, or you can choose two or more ways to generate a significant feature group and the recommended weight value of each feature, and then, for two or Cross-validation of results obtained in more than two ways, or weighted average, etc., to obtain more reasonable and accurate results.
  • the features in the generated salient feature group may include: buyer star rating, consumption quota for the most recent month, number of visits to the store in the last week, makeup preferences, women's preferences, and so on.
  • the importance score of various features are ranked, the top K features with the highest scores are selected as significant features, and the importance score can be used as the recommended weight corresponding to the significant features.
  • each feature corresponds to a group of people.
  • it does not take the intersection of each combination, but takes part of the union.
  • the cumulative total of the population covered by the first K features can be controlled to be ⁇ times the size of the expected expanded population.
  • ⁇ is a coincidence coefficient.
  • the advertiser chooses to reduce the feature's "age” weight to 0, that is, he chooses to delete the feature.
  • the weight value of the feature "Recent Purchase Amount” was adjusted from 0.3 to 1.0, and the weight value of the feature "Women's Preference” was adjusted from 0.4 to 0.1. After the weights are adjusted for each significant feature value, these weight values can be normalized as the final weight value.
  • the crowd expansion can be triggered.
  • each user in the full user group formed by the user groups corresponding to these salient features can be scored based on each salient feature, sorted by the score, and the preset extended number of users with the highest score is selected as the Users to be promoted.
  • the above-mentioned method for determining the expanded population based on the significant feature weight values is only an exemplary description. In actual implementation, other methods can be used, for example, after normalization processing.
  • the user of the weighting value is selected from the group of people corresponding to the weighting value of each significant feature as the final extended user. Which method is specifically adopted may be selected according to actual needs, which is not limited in this application.
  • the advertisement delivery system may include four modules: a parameter setting module, a salient feature extraction module, a salient feature screening module, and a crowd output module.
  • the advertiser Before the expansion of the crowd, the advertiser sets the expansion information through the advertisement placement setting interface. For example, the advertiser can set the appropriate seed population according to its own promotion needs, and then set the expansion scale and other parameters.
  • the target population may include, but is not limited to, unreachable users, potential users, shoppers, mobile users, and purchase users;
  • the crowd filtering parameters may include, but are not limited to, age, gender, and region.
  • the advertiser also needs to set the number of crowd expansion, or the crowd expansion multiple, which can be set as shown in Fig. 4.
  • the advertiser selects the expansion multiple, the corresponding estimated number of expanded users can be displayed. .
  • the advertiser sets the expansion target parameters (for example, the crowd expansion target). Specifically, the server can set different attribute characteristics according to different crowd expansion targets for expansion modeling.
  • the following crowd expansion goals can be set for advertisers to choose in advance: brand promotion, performance promotion, conversion priority, click priority, and so on.
  • the server can select the candidate group characteristic expression in the corresponding scene according to different crowd expansion targets. Therefore, the similarity between people can have multiple expression dimensions. According to the similar characteristics of different crowds, different feature expressions are abstracted, and different crowd expansion directions can be customized. Under specific feature expressions, the population expanded according to the expansion target is the population most similar to the seed population.
  • the characteristics that advertisers are interested in will also be different.
  • advertisers pay more attention to the degree of consumer preference for various brands. For example: advertisers want to promote a certain product, they need to circle a certain size of the brand intent crowd, the advertiser chooses the "brand promotion” option in the expansion target option, then for the server according to the selection, it can choose an adaptation for the candidate group " "Brand promotion” feature set, for example, the brand preference degree characteristics of major brands.
  • the advertiser promotes the sales of a certain product, and selects the "effect promotion” and "conversion priority" options in the extended target option. Then, based on the selection, the server personnel can select conversion-related feature expressions for the candidate group. For example, the characteristics of the purchase intention crowd at each store.
  • the server can perform expansion modeling based on these selections of the advertiser to filter out significant features.
  • the recommended salient features are recommended to advertisers. Among them, the salient features are multiple features matched with the expansion direction desired by the advertiser, and these features constitute a salient feature group.
  • the method for determining a salient feature may include, but is not limited to, one or more of a point mutual information index, an information gain, and a chi-square test value.
  • the server may display the weight corresponding to each salient feature. value.
  • the chi-square test value is used to use hypothesis testing methods to express the degree of correlation between each feature and the positive sample by calculating the degree of deviation of each feature under the assumption of "independence from the positive sample”.
  • the seed population can be defined as a positive sample, and the other candidate groups are used as unlabeled samples to construct a classification model. Then, the importance degree of each preset feature group in the classification model is calculated. Relevance score to measure the importance of each feature.
  • the top K features with the highest score can be selected as the salient features, and the importance score is used as the recommended weight corresponding to the salient feature.
  • the cumulative total of the candidate groups covered by the first K features can be set to be ⁇ times the expected expansion of the population size. Among them, ⁇ is a coincidence coefficient.
  • the corresponding coincidence coefficient can be obtained through offline experimental experience, so that the selected significant feature list can be covered. Enough to expand the crowd. For example, in the brand feature group, ⁇ is set to 4.5, in the click feature group, ⁇ is set to 5.0, and in the effect feature group, ⁇ is set to 3.5, etc. Therefore, different ⁇ can be set based on the selected expansion target. value.
  • the system selects the default crowd expression feature group, and uses the point mutual information index as the importance score calculation method. The calculation results are shown in the comparison table of the importance score of each feature and its inherent attributes (listed in descending order according to the importance score). :
  • the advertiser can filter the features of interest from the recommended feature list, and adjust it based on the feature weights recommended by the system to obtain the final feature list.
  • the server displays the salient features and corresponding weights for the advertiser.
  • the advertiser can interactively select the list of features of interest based on the system recommendation.
  • the advertiser can actively adjust the weight of the features of interest. Feature weights are weighted down and even set to zero (ie, the feature is deleted).
  • advertisers aim at a certain high-end beauty makeup baby, hoping to circle a group of active buyers with higher purchasing power for baby promotion. Based on the above promotion intention, the advertiser can choose to reduce the weight of the feature "age” to 0, that is, delete the feature; meanwhile, weight the feature "recent purchase amount” to 1.0 and adjust the feature "women's preference” to reduce the weight. To 0.1.
  • the advertiser's ability to control the intended crowd is enhanced, so that the advertiser can flexibly customize the expansion direction of the crowd according to their own promotion needs, which improves the user experience.
  • S5 The server sorts and truncates the candidate group according to the significant feature list selected by the advertiser and the adjusted feature weights to obtain the expanded crowd.
  • the candidate group can be scored one by one based on the salient features and corresponding weights finally selected by the advertiser.
  • the specific scoring method can be:
  • aid j represents the j-th user in the candidate group
  • f i represents the feature value of the corresponding user under the i-th feature
  • w i represents the weight of the i-th feature (normalized)
  • the feature weight is normalized. Normalize the adjusted weights for advertisers.
  • the users in the candidate group are sorted in descending order according to the similarity score with the seed group, and the group is truncated according to the expansion scale of the group, and finally the expanded group is generated.
  • the advertiser selects the buyer's star rating, beauty preference, recent purchase amount, and women's preference as features of interest, and adjusts the feature weights to 0.81, 0.76, 1.0, and 0.1, respectively.
  • the feature groups are normalized, and after normalization, they are 0.30, 0.28, 0.38, and 0.04, respectively.
  • the full candidate group is scored based on the feature weight of each feature in the normalized saliency feature group.
  • the advertiser sets the population expansion scale to 50,000, assuming that the user u2 score in the candidate group global sequence is just sorted 50,000, then the final expanded population is ⁇ ..., un,..., u3,..., u2 ⁇ .
  • advertisers are provided with a flexible entrance on the basis of the significant features of automatic recommendation, which can be used to precisely control the expansion direction of the promotion crowd and enhance the user experience.
  • the interactive setting of the optimization goal and the adjustment of the salient characteristics of the recommended population are performed.
  • the specific implementation method given in the above example is only an exemplary description. In actual implementation, other methods can also be adopted.
  • the system can automatically generate multiple sets of different distinctive feature lists, and advertisers can choose freely.
  • One group is used as the final distinctive feature group and other interactive forms, and the specific display form is not limited in this application.
  • the system can select the corresponding crowd feature expressions to achieve the purpose that the similar population to the seed crowd is the intended crowd with the promotion needs under the corresponding feature expressions, so that advertisers can set the expansion crowd Optimize the goal and guide the system to optimize the expanded population.
  • advertisers can interactively select the list of features of interest, adjust the weight of the features of interest, and adjust the weight of features that are not interested, until they are set to zero, that is, Delete the feature. Based on this, advertisers can finely control the expansion direction of the crowd at the feature level and according to the promotion intention, which greatly enhances the ability to control the expanded crowd. By controlling the expansion and optimization direction of the crowd, advertisers can accurately pinpoint the intended crowd that meets their promotion intentions, so that they can accurately target the crowd.
  • FIG. 6 is a method flowchart of an embodiment of an advertisement placing method described in this application.
  • this application provides method operation steps or device structures as shown in the following embodiments or drawings, based on conventional or no creative labor, the method or device may include more or fewer operation steps or module units. .
  • the execution order of these steps or the module structure of the device is not limited to the execution order or the module structure shown in the embodiments of the present application and shown in the accompanying drawings.
  • the method or the module structure shown in the embodiment or the drawings may be connected to execute sequentially or in parallel (for example, a parallel processor or multi-threaded processing). Environment, or even a distributed processing environment).
  • an advertising method provided by an embodiment of the present application may include:
  • Step 601 Obtain a seed crowd parameter (for example, a seed crowd), an expansion parameter (for example, an expansion number), and an expansion target parameter (for example, an expansion target, which can also be a promotion target) set by a user;
  • a seed crowd parameter for example, a seed crowd
  • an expansion parameter for example, an expansion number
  • an expansion target parameter for example, an expansion target, which can also be a promotion target
  • the interface for setting the seed crowd, the number of expansions, and the expansion target can be displayed on the terminal side for the advertiser to set according to the needs and wishes.
  • the expansion target may include but is not limited to at least one of the following: brand promotion, effect promotion, click-first, conversion-first.
  • Step 602 Determine a salient feature group according to the seed population, the number of extensions, and the expansion target, wherein the salient feature group includes a plurality of salient features;
  • Step 603 Acquire a setting operation of the distinctive feature group by the user
  • the obtained salient feature group can be displayed on the terminal side for the advertiser to modify, so that the final salient feature group can reflect the intention of the advertiser.
  • Step 604 Set the distinctive feature group according to the setting operation.
  • Step 605 Determine a similar population according to the set salient feature group, and obtain the expanded number of promotion objects
  • Step 606 Place an advertisement on the extended number of promotion objects.
  • the above-mentioned setting operation may include, but is not limited to, deletion of a feature and modification of a weight value of the feature, that is, an advertiser may delete a feature in a salient feature group, and may also perform operations on a salient feature group.
  • the weight value of the feature is modified so that the final expanded crowd can meet the advertiser's wishes.
  • S1 Use the set of people covered by each feature in the set salient feature group as the candidate group
  • the above-mentioned obtaining a seed crowd set by a user may include: obtaining a target crowd and a crowd feature set by the user; and using the target crowd and a crowd defined by the crowd feature as the seed crowd.
  • the above target population may include, but is not limited to, at least one of the following: untriggered users, potential users, mobile users, and purchase users; the population characteristics include at least one of the following: gender, age, and region.
  • determining a distinctive feature group for placing an advertisement may include:
  • S2 Use the seed population parameter as a positive sample to determine the importance of each feature in the adaptive feature group
  • the feature groups corresponding to each extension target can be set in advance, for example, the feature groups corresponding to brand promotion, the feature groups corresponding to effect promotion, and the feature groups corresponding to click priority. After obtaining the extension target parameters, they can be matched. The corresponding feature group is obtained, and the feature group is used as the adaptive feature group. Then, the importance degree of these features is determined by the seed population, and the final significant feature group is generated according to the determined importance degree.
  • the number can be set as the basic criterion, that is, the In the case where the candidate group corresponding to the feature in the salient feature group satisfies the number of extensions in the extension parameter, the feature whose importance degree satisfies a preset requirement is used as the feature in the determined salient feature group.
  • FIG. 7 is a block diagram of a server-side hardware structure of an advertisement placing method according to an embodiment of the present invention.
  • the server 10 may include one or more (only one shown in the figure) a processor 102 (the processor 102 may include but is not limited to a processing device such as a microprocessor MCU or a programmable logic device FPGA), A memory 104 for storing data, and a transmission module 106 for communication functions.
  • a processor 102 may include but is not limited to a processing device such as a microprocessor MCU or a programmable logic device FPGA
  • a memory 104 for storing data
  • a transmission module 106 for communication functions.
  • the structure shown in FIG. 7 is only schematic, and it does not limit the structure of the electronic device.
  • the server 10 may also include more or fewer components than those shown in FIG. 7, or have a different configuration from that shown in FIG.
  • the memory 104 may be used to store software programs and modules of application software, such as program instructions / modules corresponding to the advertising method in the embodiment of the present invention.
  • the processor 102 executes various software programs and modules stored in the memory 104 to execute various programs. Function application and data processing, that is, to implement the above-mentioned application advertising method.
  • the memory 104 may include a high-speed random access memory, and may further include a non-volatile memory, such as one or more magnetic storage devices, a flash memory, or other non-volatile solid-state memory.
  • the memory 104 may further include memory remotely disposed with respect to the processor 102, and these remote memories may be connected to a computer terminal through a network. Examples of the above network include, but are not limited to, the Internet, an intranet, a local area network, a mobile communication network, and combinations thereof.
  • the transmission module 106 is configured to receive or send data via a network.
  • a specific example of the network may include a wireless network provided by a communication provider of a computer terminal.
  • the transmission module 106 includes a network adapter (NIC), which can be connected to other network devices through a base station so as to communicate with the Internet.
  • the transmission module 106 may be a radio frequency (RF) module, which is used to communicate with the Internet in a wireless manner.
  • RF radio frequency
  • the above-mentioned advertisement placement device may include: a determination module, an adjustment module, and an expansion module, of which:
  • a determining module configured to determine a salient feature group for placing an advertisement based on a seed crowd parameter, an extension parameter, and an extension target parameter set by a user, wherein the salient feature group includes a plurality of salient features and each salient feature; Feature weight value;
  • An adjustment module configured to adjust the salient feature group based on a user's selection of the salient feature group
  • the expansion module is configured to determine a similar group according to the adjusted significant feature group, and obtain a promotion object for placing the advertisement.
  • the user's selection of the salient feature group includes: deletion of the salient feature, and modification of the weight value of the salient feature.
  • the determining module may include: a matching unit configured to match an adaptive feature group corresponding to the extended target parameter according to a correspondence relationship between a pre-established target parameter and an adaptive feature group; the determining unit, The seed population parameter is used as a positive sample to determine the importance degree of each feature in the adaptive feature group; and the generating unit is configured to use the feature whose importance degree meets a preset requirement as the determined significant feature group. feature.
  • the feature that satisfies a preset requirement as a feature in the determined salient feature group may include: determining that the candidate group corresponding to the feature in the salient feature group satisfies the extended parameter. In the case of expanding the number, the features whose importance degree meets the preset requirements are taken as the features in the determined significant feature group.
  • the determining module may specifically determine a distinctive feature group used for advertising in one of the following ways, but is not limited to the following: a point mutual information index, an information gain, and a chi-square test value.
  • determining a similar population according to the adjusted salient feature group and obtaining a promotion object for advertising may include: using a set of the population covered by each feature in the adjusted salient feature group as a candidate group, where The adjusted salient feature group includes: the number of extensions; ranking the candidate groups according to the feature values of each salient feature in the adjusted salient feature group; and using the most relevant extended number of objects as promotion objects .
  • the seed population parameters include: a target population and a population characteristic, and the combination of the target population and the population characteristic is a seed population.
  • the target population may include at least one of the following: untriggered users, potential users, mobile users, and purchase users; the population characteristics may include at least one of the following: gender, age, and region.
  • the extended target parameter may include but is not limited to at least one of the following: brand promotion, effect promotion, click-first, conversion-first.
  • a device for determining a promotion crowd may include: a display module, a first receiving module, a determining module, and a second receiving module, where:
  • a display module for displaying a setting interface for seed crowd parameters, extended parameters, and extended target parameters
  • a first receiving module configured to receive seed crowd parameters, extended parameters, and extended target parameters set by a user
  • a determining module configured to determine and display a significant feature group based on a seed crowd parameter, an extended parameter, and an extended target parameter set by a user, wherein the significant feature group includes a plurality of significant features
  • a second receiving module is configured to receive a user's selection operation on the salient feature group, wherein the selection operation is used to adjust the salient feature group, determine a similar population based on the adjusted salient feature group, and obtain a promotion object .
  • the second receiving module may specifically display the salient feature group and a weight value of each feature in the salient feature group; and receive a user's selection operation on the salient feature group, where the selection operation includes : A deletion operation of a feature in the salient feature group, and a modification operation of a weight value.
  • the above-mentioned advertisement placement device may include a first acquisition module, a determination module, a second acquisition module, a setting module, an extension module, and a delivery module, of which:
  • a first acquisition module configured to acquire a seed population, an expansion number, and an expansion target set by a user
  • a determining module configured to determine a salient feature group according to the seed population, the number of extensions, and an expansion target, wherein the salient feature group includes a plurality of salient features
  • a second obtaining module configured to obtain a setting operation of the distinctive feature group by the user
  • a setting module configured to set the distinctive feature group according to the setting operation
  • An extension module configured to determine a similar population according to the set salient feature group, and obtain the extended number of promotion objects
  • a delivery module is configured to deliver advertisements to the extended number of promotion objects.
  • the second obtaining module is specifically configured to obtain a setting operation of the user, wherein the setting operation includes: deleting a feature and modifying a weight value of the feature.
  • the extension module may be specifically configured to use the set of people covered by each feature in the set salient feature group as a candidate group; according to the feature values of each salient feature in the set salient feature group, The candidate group sorts the degree of relevance; the expanded number of objects with the highest degree of relevance are taken as promotion objects.
  • the first acquisition module may specifically acquire a target crowd and crowd characteristics set by the user; and set the target crowd and a crowd circled by the crowd characteristics as the seed crowd.
  • the target population may include, but is not limited to, at least one of the following: untriggered users, potential users, mobile users, and purchase users; the population characteristics include at least one of the following: gender, age, and region.
  • the expansion target may include, but is not limited to, at least one of the following: brand promotion, effect promotion, click-first, conversion-first.
  • the device for determining the promotion crowd is located in the terminal and may include: a first display module, a first receiving module, a second display module, and a second receiving module, where:
  • a first display module for displaying a seed population, an expansion number, and an expansion target setting interface
  • a first receiving module configured to receive a seed population, an expansion number, and an expansion target set by a user
  • a second display module configured to display a salient feature group determined through the seed population, the number of extensions, and the expansion target, wherein the salient feature group includes a plurality of salient features
  • a second receiving module is configured to receive a user's setting operation on the salient feature group, wherein the setting operation is used to set the salient feature group, determine a similar population based on the set salient feature group, and obtain The extended number of promotion objects.
  • the above-mentioned second display module may display the salient feature group and a weight value of each feature in the salient feature group; correspondingly, the second receiving module may receive the features of the salient feature group of the user. Delete operation, modify the weight value.
  • the application achieves the following effects: for users with promotion needs, an entry can be set that can set the seed population, the number of expansions, the expansion target, and the distinctive feature group, so that the user's expansion intention can be effectively expressed, and the existing Crowd expansion and screening is a transparent and automated way for users.
  • the devices or modules described in the foregoing embodiments may be specifically implemented by a computer chip or entity, or may be implemented by a product having a certain function.
  • the functions are divided into various modules and described separately.
  • the functions of each module may be implemented in the same or multiple software and / or hardware.
  • a module that implements a certain function may also be implemented by combining multiple submodules or subunits.
  • the method, device or module described in this application may be implemented in a computer-readable program code by the controller in any suitable manner.
  • the controller may adopt, for example, a microprocessor or processor and the storage may be processed by the (micro) Computer-readable program code (such as software or firmware), computer-readable media, logic gates, switches, Application Specific Integrated Circuits (ASICs), programmable logic controllers, and embedded microcontrollers
  • microcontrollers include, but are not limited to, the following microcontrollers: ARC 625D, Atmel AT91SAM, Microchip PIC18F26K20, and Silicone Labs C8051F320.
  • the memory controller can also be implemented as part of the control logic of the memory.
  • controller logic gates, switches, application-specific integrated circuits, programmable logic controllers, and embedded controllers by logically programming the method steps. Microcontrollers, etc. to achieve the same function. Therefore, the controller can be considered as a hardware component, and the device included in the controller for implementing various functions can also be considered as a structure within the hardware component. Or even, the means for implementing various functions can be regarded as a structure that can be both a software module implementing the method and a hardware component.
  • program modules include routines, programs, objects, components, data structures, classes, etc. that perform specific tasks or implement specific abstract data types.
  • program modules may be located in local and remote computer storage media, including storage devices.
  • the present application can be implemented by means of software plus necessary hardware. Based on such an understanding, the technical solution of the present application in essence or a part that contributes to the existing technology may be embodied in the form of a software product, or may be reflected in the implementation process of data migration.
  • the computer software product can be stored in a storage medium, such as ROM / RAM, magnetic disk, optical disc, etc., and includes a number of instructions to enable a computer device (which can be a personal computer, a mobile terminal, a server, or a network device, etc.) to execute this software. Apply for the method described in each embodiment or some parts of the embodiment.

Landscapes

  • Business, Economics & Management (AREA)
  • Strategic Management (AREA)
  • Engineering & Computer Science (AREA)
  • Accounting & Taxation (AREA)
  • Development Economics (AREA)
  • Finance (AREA)
  • Economics (AREA)
  • Game Theory and Decision Science (AREA)
  • Entrepreneurship & Innovation (AREA)
  • Marketing (AREA)
  • Physics & Mathematics (AREA)
  • General Business, Economics & Management (AREA)
  • General Physics & Mathematics (AREA)
  • Theoretical Computer Science (AREA)
  • Management, Administration, Business Operations System, And Electronic Commerce (AREA)
  • Information Transfer Between Computers (AREA)

Abstract

本申请提供了一种广告投放方法、确定推广人群的方法、服务器和客户端，其中，该方法包括：基于用户设定的种子人群参数、扩展参数和扩展目标参数，确定出用于投放广告的显著特征组，其中，所述显著特征组中包括有多个显著特征，以及各个显著特征的权重值；基于用户对所述显著特征组的选择，对所述显著特征组进行调整；根据调整后的显著特征组确定相似人群，得到用于投放所述广告的推广对象。通过上述方案解决了现有的人群扩展和筛选对用户是透明的自动化的方式所导致的用户无法把控扩展方向，而导致的扩展结果不满足用户需求的技术问题，达到了使得用户的扩展意愿可以得到有效表达的技术效果，大大提升了用户体验。

Description

广告投放方法、确定推广人群的方法、服务器和客户端
本申请要求2018年08月27日递交的申请号为201810980484.9、发明名称为“广告投放方法、确定推广人群的方法、服务器和客户端”的中国专利申请的优先权，其全部内容通过引用结合在本申请中。
技术领域
本申请属于互联网技术领域，尤其涉及一种广告投放方法、确定推广人群的方法、服务器和客户端。
背景技术
随着互联网和电商的不断发展，广告投放的需求也越来越大，目前，广告投放过程中广告主圈定人群的方式主要包括两种：一种是基于人口统计的方式，另一种是基于用户行为的方式。其中，基于人口统计的维度，主要是基于用户的性别、年龄和购买力等维度圈定人群，而基于用户行为的维度，主要是基于用户在广告主店铺的浏览点击等行为圈定人群。
然而，无论采用哪种圈人方式，都会存在圈定的用户数量有限，无法满足营销计划的需求的问题。
针对上述问题，目前尚未提出有效的解决方案。
发明内容
本申请目的在于提供一种广告投放方法、确定推广人群的方法、服务器和客户端，可以实现满足广告主需求的广告投放，使得广告主意愿可以得到有效表达。
本申请提供一种广告投放方法、确定推广人群的方法、服务器和客户端是这样实现的：
一种广告投放方法，所述方法包括：
基于用户设定的种子人群参数、扩展参数和扩展目标参数，确定出用于投放广告的显著特征组，其中，所述显著特征组中包括有多个显著特征，以及各个显著特征的权重值；
基于用户对所述显著特征组的选择，对所述显著特征组进行调整；
根据调整后的显著特征组确定相似人群，得到用于投放所述广告的推广对象。
一种确定推广人群的方法，所述方法包括：
显示种子人群参数、扩展参数和扩展目标参数的设定界面；
接收用户设定的种子人群参数、扩展参数和扩展目标参数；
基于用户设定的种子人群参数、扩展参数和扩展目标参数，确定并显示显著特征组，其中，所述显著特征组中包括有多个显著特征；
接收用户对所述显著特征组的选择操作，其中，所述选择操作用于对所述显著特征组进行调整，基于调整后的显著特征组确定相似人群，得到推广对象。
一种服务器，包括处理器以及用于存储处理器可执行指令的存储器，所述处理器执行所述指令时实现如下方法的步骤：
基于用户设定的种子人群参数、扩展参数和扩展目标参数，确定出用于投放广告的显著特征组，其中，所述显著特征组中包括有多个显著特征，以及各个显著特征的权重值；
基于用户对所述显著特征组的选择，对所述显著特征组进行调整；
根据调整后的显著特征组确定相似人群，得到用于投放所述广告的推广对象。
一种客户端，包括处理器以及用于存储处理器可执行指令的存储器，所述处理器执行所述指令时实现如下方法的步骤：
显示种子人群参数、扩展参数和扩展目标参数的设定界面；
接收用户设定的种子人群参数、扩展参数和扩展目标参数；
基于用户设定的种子人群参数、扩展参数和扩展目标参数，确定并显示显著特征组，其中，所述显著特征组中包括有多个显著特征；
接收用户对所述显著特征组的选择操作，其中，所述选择操作用于对所述显著特征组进行调整，基于调整后的显著特征组确定相似人群，得到推广对象。
一种计算机可读存储介质，其上存储有计算机指令，所述指令被执行时实现上述方法的步骤。
本申请提供的广告投放方法、确定推广人群的方法、服务器和客户端，为有推广需求的用户设置了可以设定种子人群参数、扩展参数、扩展目标参数，以及显著特征组的入口，从而使得用户的扩展意愿可以得到有效表达，解决了现有的人群扩展和筛选对用户是透明的自动化的方式所导致的用户无法把控扩展方向，而导致的扩展结果不满足用户需求的技术问题，达到了使得用户的扩展意愿可以得到有效表达的技术效果，大大提升了用户体验。
附图说明
为了更清楚地说明本申请实施例或现有技术中的技术方案，下面将对实施例或现有技术描述中所需要使用的附图作简单地介绍，显而易见地，下面描述中的附图仅仅是本申请中记载的一些实施例，对于本领域普通技术人员来讲，在不付出创造性劳动性的前提下，还可以根据这些附图获得其他的附图。
图1是本申请提供的广告投放系统的架构图；
图2是本申请提供的广告投放系统的结构框图；
图3是本申请提供的广告投放方法的方法流程图；
图4是本申请提供的扩展参数选择示意图；
图5是本申请提供的显著特征设置示意图；
图6是本申请提供的广告投放方法的方法流程图；
图7是本申请提供的广告投放服务器的架构示意图。
具体实施方式
为了使本技术领域的人员更好地理解本申请中的技术方案，下面将结合本申请实施例中的附图，对本申请实施例中的技术方案进行清楚、完整地描述，显然，所描述的实施例仅仅是本申请一部分实施例，而不是全部的实施例。基于本申请中的实施例，本领域普通技术人员在没有作出创造性劳动前提下所获得的所有其他实施例，都应当属于本申请保护的范围。
为了进行人群扩展，目前常用的方式是，由广告主确定出一部分目标人群，然后通过相似人群扩展功能，以已有的目标人员作为种子用户，找到与这些种子用户相似的人作为意向人群，以进行广告投放。在实现的时候，一般是基于用户的性别、年龄和购买力等维度圈定人群，而基于用户行为的维度，主要是基于用户在广告主店铺的浏览点击等行为圈定人群。然而，无论采用哪种圈人方式，都会存在圈定的用户数量有限，无法满足营销计划的需求的问题。进一步的，目前的广告投放系统中，一般是在广告主选定扩展规模之后，由系统自动生成符合要求的扩展人群，广告主是无法控制意向人群的扩展方向的，广告主的扩展意愿无法得到满足和体现，用户体验效果较差。
考虑到如果可以提供一种广告投放系统，提供给用户可以表达意向人群的扩展方向的选择入口，那么可以使得广告主的扩展意愿得到满足，使得最终的扩展结果更为符合 用户需求。为此，提供了一种广告投放系统，如图1所示，可以包括：客户端101和服务器102。在用户选择进行人群扩展的情况下，服务器102可以推送给用于设置扩展意愿的设置界面，用户可以通过客户端上显示的设置界面，设置希望的人群扩展方向，在用户选定完人群扩展意愿之后，服务器102可以根据用户设置的扩展意愿进行人群扩展，并对扩展的人群进行广告投放。
其中，上述的人群扩展就是基于种子人群，确定出种子人群的相似人群，以得到预定数量规模的人群。例如，种子人群为10，可以进行人群扩展得到90个人，这样就有100个人，即，实现了人群扩展。
在一个实施方式中，上述服务器102可以是单一的服务器，也可以是服务器集群，也可以是设置在云端的服务器，具体采用哪种方式的服务器，可以根据实际需要选择，本申请对此不作限定。
上述客户端可以是用户操作使用的终端设备或者软件。具体的，客户端可以是智能手机、平板电脑、笔记本电脑、台式计算机、智能手表或者其它可穿戴设备等终端设备。当然，用户端也可以是能运行于上述终端设备中的软件。例如：APP、浏览器等应用软件。
对于服务端而言，为了使得用户(即，广告主)的扩展意愿可以得到有效的表达，从而使得最终广告所推广至的群体更为符合广告主的需求，可以为广告主提供可以表达扩展意愿的入口，结合广告主希望的推广条件和信息，确定扩展人群。
在实现的时候，可以但不限于在以下的一个或多个节点设置广告主参与的入口：
1)选择种子人群；
2)设定人群扩展目标；
3)设置特征权重。
下面分别对上述几个节点的内容进行说明如下：
1)选择种子人群：
广告主可以在选择种子人群的界面设置希望推送人群的一些参数信息，例如：目标人群、人群特征等。其中，目标人群例如可以是：未触发用户、潜在用户、行动用户和购买用户等等，用户可以根据需要选择或者设定希望的目标人群是哪种类型的用户。人群特征例如可以是：性别、年龄、所在城市等等。
在实际实现的过程中，如果广告主所希望的推广人数较少，在设置两个条件：目标人群和人群特征之后，所形成的人群集合中人群数量已经达到广告主所希望的广告推广 量，那么就不需要再进行人群扩展。为此，可以设置一个提醒界面，提醒广告主不需要进行人群扩展，就可以完成广告推广需求。
只有在所希望推广的人数是设置上述两个条件后所得到的人群集合中的人群数量不能满足的情况下，才进行人群扩展。
例如，广告主希望推广的人群数量为30万，在设置目标人群和人群特征之后，所得到的人群数量为35万，那么不需要进行人群扩展。如果广告主希望推广的人群数量为100万，在设置目标人群和人群特征之后，所得到的人群数量为35万，那么就需要进行人群扩展。
通过上述这种根据所希望推广的人群数量的大小来确定是否需要进行人群扩展，并提醒用户的方式，可以有效提升广告推广所触达用户的准确性和有效性，且不会造成系统不必要的运算，可以节省成本提升效率。
上述在设置两个条件：目标人群和人群特征之后，所形成的人群集合可以作为种子人群，用于后续人群推广的时候使用。
然而，值得注意的是，上述所设置的目标人群和人群特征中所列举的选项信息仅是一种示意性描述，在实际使用的时候，可以根据需要进行选择，本申请对此不作限定。
在确定需要进行人群扩展的情况下，可以提供给广告主选择扩展人数的权利。例如，广告主可以采用输入人群扩展数量的方式，也可以采用设置扩展倍数的方式，可以是采用输入数字的方式，也可以是采用进度条拖拽进度条的方式，也可以是设置一个扩展倍数或者扩展数量，供广告主点选的方式，具体采用哪种方式，可以根据实际需要选择，本申请对此不作限定。
2)设定人群扩展目标：
在确定需要进行人群扩展的情况下，可以为广告主提供多个扩展目标供广告主选择，例如：品牌推广、效果推广、点击优先、转化优先等等。即，可以但不限于这些扩展目标中的一个或多个供广告主选择。
基于可供选择的扩展目标，广告主可以根据自身的广告推广需求选择希望的扩展目标。例如，如果广告主本次广告的推广主要是希望实现品牌的推广，那么可以选择品牌推广作为扩展目标，如果广告主希望本次广告的推广可以让更多的用户愿意点击，那么可以选择点击优先作为扩展目标。
然而值得注意的是，上述所列举的扩展目标仅是一种示例性描述，在实际实现的时候，可以采用其他的扩展目标，例如：加购优先等等，本申请对此不作限定。
3)设置特征权重：
考虑到仅提供给用户选择种子人群和人群扩展目标的权限，并不能实现精确反应广告主意愿的目的。
为此，考虑到可以为广告主提供设置特征权重的权限，在广告主选择种子人群，并设置了扩展人数、设定了人群扩展目标之后，可以为不同的扩展目标关联到不同的特征组，并结合。
例如，广告主希望对某款商品进行推广，需要圈定一定规模的品牌意向人群，因此，在选择扩展目标的时候，选择“品牌推广”。服务器在确定广告主选择的是品牌推广作为扩展目标的情况下，选择与“品牌推广”对应的特征组，例如，各大品牌下的品牌偏好程度特征等。如果在扩展目标选项选择“效果推广”和“转化优先”，那么可以选择与“效果推广”和“转化优先”对应的特征组，例如：各个店铺下的购买意向人群特征。
基于广告主选定的种子人群、扩展目标和扩展人数，从服务器提供的特征组进行扩展建模，筛选出显著特征组，即，确定出对扩展结果影响较大的特征，将这些特征展示给广告主，供广告主选择是否采用这些特征，对于广告主不希望的特征，广告主有权利删除。
进一步的，为了进一步提升广告主对扩展结果的控制粒度，可以为显著特征组中的每个特征设置权重值，权重值高的表明对扩展结果的影响大，权重值低的表明对扩展结果的影响小。服务器可以根据种子人群、扩展目标和扩展人数，给出一个默认的显著特征组，以及显著特征组中每个特征的权重值展示给广告主。广告主可以根据自己的需求和意愿，对各个特征的权重值进行修改，以实现对人群扩展结果的控制，即，实现更细粒度的控制。
在生成显著特征组的时候，可以按照但不限于以下方式中的一种或多种生成：点互信息指数、信息增益、卡方检验值等。可以是选择其中的一种方式生成显著特征组以及各个特征的建议权重值，也可以是选择两种或两种以上的方式生成显著特征组以及各个特征的建议权重值，然后，对两个或两个以上的方式得到的结果进行交叉验证，或者进行加权平均等等，得到更为合理准确的结果。
例如，生成的显著特征组中的特征可以包括：买家星级、最近一月消费额度、最近一周到店次数、彩妆偏好、女装偏好等等。
为了确定出显著特征组，就需要对得到的多个特征进行重要程度排序，选出一定数量的特征，即，重要程度靠前的特征作为显著特征。例如：对各个特征的重要性得分排 序，选出得分最大的前K个特征作为显著特征，重要性得分可以作为对应显著特征的建议权重。
对于每个特征本身而言，是对应着一个人群组合的，在进行扩展的时候，并非取各个组合的交集，而是取并集中的一部分。考虑到特征所覆盖的人群存在一定的重合度，为了保证最终扩展结果的数量可以满足设定的扩展人群数量，可以控制前K个特征所覆盖的人群累加总数为期望扩展人群规模的α倍，其中，α为重合系数。
例如：广告主选择将特征“年龄”权重下调至0，即选择删除该特征。将特征“最近购买金额”的权重值由0.3调整至1.0，将特征“女装偏好”的权重值由0.4调整至0.1。在对每个显著特征值都调整完权重之后，可以这这些权重值进行归一化处理，作为最终的权重值。
在一个实施方式中，在广告主对各个显著特征的权重值设置完成之后，就可以触发进行人群扩展。为了实现人群扩展，可以对这些显著性特征对应的用户组所形成的全量用户组中的各个用户基于各显著特征进行打分，并按照得分进行排序，选择得分靠前的预设扩展数量个用户作为待推广的用户。
然而，值得注意的是，上述所提供的基于显著特征的权重值确定扩展人群的方式仅是一种示例性描述，在实际实现的时候，可以采用其它的方式，例如，按照归一化处理后的权重值，从各个显著特征的权重值对应的人群组合中选择扩展总数与对应权重值的乘积的数量的用户作为最终的扩展用户。具体采用哪种方式，可以根据实际需要选择，本申请对此不作限定。
下面结合一个具体实施例对上述广告投放系统和方法进行说明，然而，值得注意的是，该具体实施例仅是为了更好地说明本申请，并不构成对本申请的不当限定。
如图2所示，该广告投放系统可以包括：参数设置模块、显著特征提取模块、显著特征筛选模块和人群输出模块四个模块。
基于该系统可以按照如图3所示的方法进行广告投放，可以包括如下步骤：
S1：广告主在进行人群扩展前，通过广告投放设置界面设置好扩展信息，例如，广告主可以根据自身的推广需求，设定合适的种子人群，然后，设置扩展规模等参数。
具体的，在选定种子人群的时候，可以如图4所示，通过设定目标人群和人群过滤参数的方式选定。例如，目标人群可以包括但不限于：未触达用户、潜在用户、进店用户、行动用户和购买用户等；人群过滤参数可以包括但不限于：年龄、性别、地域等。
进一步的，广告主还需要设定人群扩展数量，或者是人群扩展倍数，可以按照如图 4所示的方式进行设置，可以通过在广告主选择扩展倍数的同时，显示对应的预计的扩展用户数。
S2：广告主设定扩展目标参数(例如：人群扩展目标)，具体的，服务器可以根据不同的人群扩展目标，设置不同的属性特征，用于扩展建模。
例如，可以预先设定如下的人群扩展目标供广告主选择：品牌推广、效果推广、转化优先、点击优先等。服务器可以根据不同的人群扩展目标，选取对应场景下的候选人群特征表达。因此，人与人之间的相似可以有多种表达维度，根据不同的人群相似特点，抽象出不同的特征表达，可以定制不同的人群扩展方向。特定的特征表达下，按照该扩展目标扩展出来的人群即为与种子人群最为相似的人群。
在不同的推广目标下，广告主所感兴趣的特征也会有所不同，例如：品牌推广场景下，广告主更关注消费者对各品牌的偏好程度。例如：广告主希望对某款商品进行推广，需要圈定一定规模的品牌意向人群，广告主在扩展目标选项选择“品牌推广”选项，那么对于服务器而言根据该选择可以为候选人群选择适配“品牌推广”目标的特征组，例如，各大品牌下的品牌偏好程度特征。广告主在大促期间，为某款商品进行销货推广，在扩展目标选项选择“效果推广”和“转化优先”选项，那么对于服务器人员根据该选择可以为候选人群选择转化相关的特征表达，例如，各个店铺下的购买意向人群特征。
S3：在广告主选定种子人员参数(例如：种子人群)、扩展参数(例如：人群扩展规模)扩展目标等之后，服务器可以基于广告主的这些选择进行扩展建模，以筛选出显著特征，并将筛选出的显著特征推荐给广告主。其中，显著特征就是匹配出的与广告主所希望的扩展方向匹配的多个特征，这些特征组成了显著特征组。
其中，确定显著特征的方法可以包括但不限于：点互信息指数、信息增益、卡方检验值等中的一种或多种，在输出显著特征的同时，服务器可以显示各显著特征对应的权重值。
1)点互信息指数，用于描述各个特征与正例样本的相关性；
2)信息增益，用于通过计算模型在有此特征和无此特征时带来的信息变化量来衡量单个特征的重要性程度；
3)卡方检验值，用于使用假设检验的方法，通过计算各个特征在“和正例样本相互独立”假设下的偏差程度来表达其与正例样本的关联度。
具体的，可以将种子人群定义为正例样本，其他候选人群作为无标记样本，构建分类模型，然后，计算各个预设特征组在该分类模型中的重要性程度，通过单个特征与正 例样本的关联度得分来衡量各个特征的重要性程度。
在对各个特征的重要程度的得分进行排序后，可以选出得分最大的前K个特征作为显著特征，重要性得分作为对应显著特征的建议权重。考虑到对于不同的特征而言，覆盖的人群存在一定的重合度，为了保证可以找回足够的人群数量，可以设定前K个特征覆盖的候选人群累加总数为期望扩展人群规模的α倍，其中，α为重合系数。
进一步的，因为在不同的特征表达下，人群的覆盖度是不同的，针对不同的特征组，可以通过线下实验经验得出对应的重合系数，从而可以保证选定的显著特征列表能够覆盖到足够多的扩展人群。例如，品牌特征组下，α设为4.5，点击特征组下，α设为5.0，效果特征组下，α设为3.5等，因此，可以基于选择的扩展目标的不同，设定不同的α取值。
结合一个具体实施对显著特征进行说明如下：
假设广告主设定种子人群S，人群规模为seed_cnt＝10000，并圈定候选人群；广告主设定期望扩展人群规模crowd_cnt＝50000，其他参数均为默认参数。系统选取默认人群表达特征组，并使用点互信息指数作为重要性得分计算方法，计算结果如表1各个特征的重要性得分及其固有属性(已按照重要性得分做降序排列)对照表所示：
表1
特征ID 特征名称 重要性得分 覆盖人群数目
1 买家星级 0.81 12425
2 美妆偏好 0.76 76230
3 最近购买金额 0.73 9003
4 女装偏好 0.57 142353
5 年龄 0.23 92463
6 性别 0.19 819042
7 职业 0.06 34764
遍历上述表1并依次累加计算前个K特征覆盖的人群规模cnt，直至cnt>＝crowd_cnt×α(默认选择“点击优先”优化目标，对应重合系数α＝5.0)，在遍历终止时，选定K＝5，此时，cnt＝332474>＝round(50000××5.0)＝250000(round表示四舍五入)。因此，可以选取前5个特征作为显著特征推荐给广告主，同时展示各个特征对应的建议权重(即，重要程度得分)如表2和图5所示：
表2
特征ID 特征名称 重要性得分
1 买家星级 0.81
2 美妆偏好 0.76
3 最近购买金额 0.73
4 女装偏好 0.57
5 年龄 0.23
S4：广告主可以从推荐的显著特征列表中筛选出感兴趣的特征，并在系统推荐的特征权重基础上进行调整，得到最终的显著特征列表。
服务器为广告主展示生成的显著特征及对应权重，广告主可在系统推荐的基础上，交互选择感兴趣的特征列表，且广告主可以主动对感兴趣的特征的权重进行加权调整，不感兴趣的特征权重进行降权调整，甚至置为零(即，删除该特征)。
例如，广告主针对某款高端美妆宝贝，希望圈定一群具备较高购买力的活跃买家进行宝贝推广。基于上述推广意愿，广告主可以选择将特征“年龄”权重下调至0，即，选择删除该特征；同时对特征“最近购买金额”进行加权调整至1.0，将特征“女装偏好”进行降权调整至0.1。
通过这种方式，增强了广告主对意向人群的把控能力，使得广告主可以根据自身的推广需求来灵活订制人群的扩展方向，提升了用户体验。
S5：服务器根据广告主选定的显著特征列表和调整后的特征权重对候选人群进行排序和截断，得到扩展人群。
具体的，可以根据广告主最终选定的显著特征及其对应权重，为候选人群逐一打分，具体打分方法可以是：
Figure PCTCN2019101998-appb-000001
Figure PCTCN2019101998-appb-000002
其中，aid j表示候选人群中第j个用户，f i表示对应用户在第i维特征下的特征值，w i表示第i维特征(归一化后的)权重，特征权重归一化表示对广告主调整之后的权重做归一化。
对候选人群中的用户按照与种子人群的相似度得分做降序排列，并根据人群扩展规模对人群进行截断，最终生成扩展人群。
例如，广告主选定买家星级、美妆偏好、最近购买金额、女装偏好这几个特征作为感兴趣特征，并分别调整特征权重为0.81、0.76、1.0、0.1，进而服务器可以对该显著特征组做归一化，归一化后分别为0.30、0.28、0.38、0.04。
然后，基于归一化后的显著特征组中各个特征的特征权重对全量候选人群进行打分，例如，候选人权集合为U＝{u1,u2,u3,…,un}及其特征表达如表3所示：
表3
用户 买家星级 美妆偏好 最近购买金额 女装偏好
u1 0.29 0.72 0.23 0.77
u2 0.32 0.83 0.34 0.32
u3 0.21 0.29 0.87 0.63
u4 0.15 0.64 0.12 0.52
…        
un 0.30 0.41 0.92 0.12
按照如下方式计算各个用户与种子人群的相似度得分：
Score(u1)＝0.29×0.30+0.72×0.28+0.23×0.38+0.77×0.04＝0.4068
Score(u2)＝0.32×0.30+0.83×0.28+0.34×0.38+0.32×0.04＝0.4704
Score(u3)＝0.21×0.30+0.29×0.28+0.87×0.38+0.63×0.04＝0.5
Score(u4)＝0.15×0.30+0.64×0.28+0.12×0.38+0.52×0.04＝0.2906
…
Score(un)＝0.30×0.30+0.41×0.28+0.92×0.38+0.12×0.04＝0.5592
对候选人群U按照相似度得分排序，得到…,un,…,u3,…,u2,…,u1,…,u4,…
因广告主设定人群扩展规模为50000，假定候选人群全局序列中，用户u2得分刚好排序50000，那么最终的扩展人群为{…,un,…,u3,…,u2}。
在上例中，在自动推荐显著特征的基础上，提供给广告主一个灵活的入口，用于精准控制推广人群的扩展方向，增强了用户体验，具体的，在本例中，主要从扩展人群优化目标的交互设定和推荐人群显著特征的调整两方面进行。上例所给出的具体的实现方式仅是一种示例性描述，在实际实现的时候，还可以采用其它的方式，例如：系统可以自动生成多组不同的显著特征列表，广告主可以自由选择其中一组作为最终显著特征组 等交互形式，具体的展现形式，本申请对此不作限定。
具体的，在本例中，可以全自动为广告主扩展出符合与种子人群相似度预期的意向人群，同时，提供了广告主推广意愿表达的入口，极大程度上增强了广告主对意向人群扩展方向的把控能力，增强了用户体验。基于此，广告主可以选择性设定不同的人群扩展优化目标，从候选人群中筛选出具备较高转化/点击效果的人群，或者定位到具备品牌偏好的意向人群。针对不同的扩展目标，系统可以选取对应的人群特征表达，以达到在对应特征表达下，与种子人群相似的人群即为具备该推广需求的意向人群的目的，从而使得广告主可以设定扩展人群优化目标，有针对性地指导系统进行扩展人群优化。
另一方面，在系统推荐的显著特征列表中，广告主可以交互选择感兴趣的特征列表，对感兴趣的特征权重进行加权调整，不感兴趣的特征权重进行降权调整，直至置为零，即删除该特征。基于此，广告主可以在特征级别上，根据推广意愿，细粒度控制人群的扩展方向，极大程度地增强了对扩展人群的把控能力。通过对人群扩展优化方向的控制，广告主可以精准圈定出符合自身推广意愿的意向人群，从而可以有针对性地对该人群进行精准投放。
图6是本申请所述一种广告投放方法一个实施例的方法流程图。虽然本申请提供了如下述实施例或附图所示的方法操作步骤或装置结构，但基于常规或者无需创造性的劳动在所述方法或装置中可以包括更多或者更少的操作步骤或模块单元。在逻辑性上不存在必要因果关系的步骤或结构中，这些步骤的执行顺序或装置的模块结构不限于本申请实施例描述及附图所示的执行顺序或模块结构。所述的方法或模块结构的在实际中的装置或终端产品应用时，可以按照实施例或者附图所示的方法或模块结构连接进行顺序执行或者并行执行(例如并行处理器或者多线程处理的环境，甚至分布式处理环境)。
具体的如图6所述，本申请一种实施例提供的一种广告投放方法可以包括：
步骤601：获取用户设定的种子人群参数(例如：种子人群)、扩展参数(例如：扩展数量)和扩展目标参数(例如：扩展目标，也可以成为推广目标)；
即，可以在终端侧显示用于设置的种子人群、扩展数量和扩展目标的界面，供广告主按照需求和希望进行设定。其中，扩展目标可以包括但不限于以下至少之一：品牌推广、效果推广、点击优先、转化优先。
步骤602：根据所述种子人群、扩展数量和扩展目标，确定出显著特征组，其中，所述显著特征组中包括有多个显著特征；
步骤603：获取所述用户对所述显著特征组的设置操作；
即，可以在终端侧显示得到的显著特征组，供广告主进行修改，以便使得最终的显著特征组是可以反映广告主意愿的。
步骤604：根据所述设置操作，对所述显著特征组进行设定；
步骤605：根据设定后的显著特征组确定相似人群，得到所述扩展数量个推广对象；
步骤606：向所述扩展数量个推广对象投放广告。
具体的，上述的设置操作可以包括但不限于：对特征的删除、对特征的权重值的修改，即，广告主可以对显著特征组中的特征进行删除操作，还可以对显著特征组中的特征的权重值进行修改，以便使得最终的扩展人群可以满足广告主的意愿。
在进行人群扩展的时候，可以按照如下步骤进行扩展：
S1：将设定后的显著特征组中各个特征所覆盖的人群的集合作为候选人群；
S2：根据设定后的显著特征组中各个显著特征的特征值，对所述候选人群进行相关度排序；
S3：将相关度最高的所述扩展数量个对象作为推广对象。
上述的获取用户设定的种子人群，可以包括：获取用户设定的目标人群和人群特征；将所述目标人群和所述人群特征所圈定的人群集合，作为所述种子人群。
上述的目标人群可以包括但不限于以下至少之一：未触发用户、潜在用户、行动用户和购买用户；所述人群特征包括以下至少之一：性别、年龄、地域。
在上述步骤602中，基于用户设定的种子人群参数、扩展参数和扩展目标参数，确定出用于投放广告的显著特征组，可以包括：
S1：根据预先建立的目标参数与适配特征组之间的对应关系，匹配出所述扩展目标参数对应的适配特征组；
S2：将种子人群参数作为正样本，确定出所述适配特征组中各个特征的重要性程度；
S3：将重要性程度满足预设要求的特征作为确定出的显著特征组中的特征。
即，可以预先设置每个扩展目标所对应的特征组，例如：品牌推广对应的特征组、效果推广对应的特征组、点击优先对应的特征组等，在获取到扩展目标参数之后，就可以匹配出对应的特征组，并将该特征组作为适配特征组，然后，以种子人群对这些特征进行重要性程度确定，按照确定出的重要性程度生成最终的显著特征组。
进一步的，考虑到在实际确定特征组的情况下，需要保证特征组所能匹配到的人群数量可以满足扩展数量要求，因此，可以设置数量达到扩展数量要求作为基本准则，即，可以在确定出显著特征组中的特征对应的候选人群满足所述扩展参数中的扩展数量的情 况下，将重要性程度满足预设要求的特征作为确定出的显著特征组中的特征。
本申请上述实施例所提供的方法实施例可以在移动终端、计算机终端或者类似的运算装置中执行。以运行在服务器端上为例，图7是本发明实施例的一种广告投放方法的服务器端的硬件结构框图。如图7所示，服务器10可以包括一个或多个(图中仅示出一个)处理器102(处理器102可以包括但不限于微处理器MCU或可编程逻辑器件FPGA等的处理装置)、用于存储数据的存储器104、以及用于通信功能的传输模块106。本领域普通技术人员可以理解，图7所示的结构仅为示意，其并不对上述电子装置的结构造成限定。例如，服务器10还可包括比图7中所示更多或者更少的组件，或者具有与图7所示不同的配置。
存储器104可用于存储应用软件的软件程序以及模块，如本发明实施例中的广告投放方法对应的程序指令/模块，处理器102通过运行存储在存储器104内的软件程序以及模块，从而执行各种功能应用以及数据处理，即实现上述的应用程序的广告投放方法。存储器104可包括高速随机存储器，还可包括非易失性存储器，如一个或者多个磁性存储装置、闪存、或者其他非易失性固态存储器。在一些实例中，存储器104可进一步包括相对于处理器102远程设置的存储器，这些远程存储器可以通过网络连接至计算机终端。上述网络的实例包括但不限于互联网、企业内部网、局域网、移动通信网及其组合。
传输模块106用于经由一个网络接收或者发送数据。上述的网络具体实例可包括计算机终端的通信供应商提供的无线网络。在一个实例中，传输模块106包括一个网络适配器(Network Interface Controller，NIC)，其可通过基站与其他网络设备相连从而可与互联网进行通讯。在一个实例中，传输模块106可以为射频(Radio Frequency，RF)模块，其用于通过无线方式与互联网进行通讯。
在软件层面，上述广告投放装置可以包括：确定模块、调整模块和扩展模块，其中：
确定模块，用于基于用户设定的种子人群参数、扩展参数和扩展目标参数，确定出用于投放广告的显著特征组，其中，所述显著特征组中包括有多个显著特征，以及各个显著特征的权重值；
调整模块，用于基于用户对所述显著特征组的选择，对所述显著特征组进行调整；
扩展模块，用于根据调整后的显著特征组确定相似人群，得到用于投放所述广告的推广对象。
在一个实施方式中，用户对所述显著特征组的可以选择，包括：对显著特征的删除、对显著特征的权重值的修改。
在一个实施方式中，确定模块可以包括：匹配单元，用于根据预先建立的目标参数与适配特征组之间的对应关系，匹配出所述扩展目标参数对应的适配特征组；确定单元，用于将种子人群参数作为正样本，确定出所述适配特征组中各个特征的重要性程度；生成单元，用于将重要性程度满足预设要求的特征作为确定出的显著特征组中的特征。
在一个实施方式中，将重要性程度满足预设要求的特征作为确定出的显著特征组中的特征，可以包括：在确定出显著特征组中的特征对应的候选人群满足所述扩展参数中的扩展数量的情况下，将重要性程度满足预设要求的特征作为确定出的显著特征组中的特征。
在一个实施方式中，确定模块具体可以按照以下但不限于以下方式之一确定用于投放广告的显著特征组：点互信息指数、信息增益、卡方检验值。
在一个实施方式中，根据调整后的显著特征组确定相似人群，得到用于投放广告的推广对象，可以包括：将调整后的显著特征组中各个特征所覆盖的人群的集合作为候选人群，其中，调整后的显著特征组中包括：扩展数量；根据调整后的显著特征组中各个显著特征的特征值，对所述候选人群进行相关度排序；将相关度最高的扩展数量个对象作为推广对象。
在一个实施方式中，种子人群参数，包括：目标人群和人群特征，其中，所述目标人群和人群特征所圈定的人群组合为种子人群。
在一个实施方式中，目标人群可以包括以下至少之一：未触发用户、潜在用户、行动用户和购买用户；所述人群特征可以包括以下至少之一：性别、年龄、地域。
在一个实施方式中，扩展目标参数可以包括但不限于以下至少之一：品牌推广、效果推广、点击优先、转化优先。
在软件层面，提供了一种确定推广人群的装置，可以包括：显示模块、第一接收模块、确定模块和第二接收模块，其中：
显示模块，用于显示种子人群参数、扩展参数和扩展目标参数的设定界面；
第一接收模块，用于接收用户设定的种子人群参数、扩展参数和扩展目标参数；
确定模块，用于基于用户设定的种子人群参数、扩展参数和扩展目标参数，确定并显示显著特征组，其中，所述显著特征组中包括有多个显著特征；
第二接收模块，用于接收用户对所述显著特征组的选择操作，其中，所述选择操作用于对所述显著特征组进行调整，基于调整后的显著特征组确定相似人群，得到推广对象。
在一个实施方式中，第二接收模块具体可以显示所述显著特征组和所述显著特征组中各个特征的权重值；接收用户对所述显著特征组的选择操作，其中，所述选择操作包括：对所述显著特征组中的特征的删除操作、对权重值的修改操作。
在软件层面，上述广告投放装置可以包括：第一获取模块、确定模块、第二获取模块、设定模块、扩展模块和投放模块，其中：
第一获取模块，用于获取用户设定的种子人群、扩展数量和扩展目标；
确定模块，用于根据所述种子人群、扩展数量和扩展目标，确定出显著特征组，其中，所述显著特征组中包括有多个显著特征；
第二获取模块，用于获取所述用户对所述显著特征组的设置操作；
设定模块，用于根据所述设置操作，对所述显著特征组进行设定；
扩展模块，用于根据设定后的显著特征组确定相似人群，得到所述扩展数量个推广对象；
投放模块，用于向所述扩展数量个推广对象投放广告。
在一个实施方式中，第二获取模块具体用于获取用户的设置操作，其中，所述设置操作包括：对特征的删除、对特征的权重值的修改。
在一个实施方式中，扩展模块具体可以用于将设定后的显著特征组中各个特征所覆盖的人群的集合作为候选人群；根据设定后的显著特征组中各个显著特征的特征值，对所述候选人群进行相关度排序；将相关度最高的所述扩展数量个对象作为推广对象。
在一个实施方式中，第一获取模块具体可以获取用户设定的目标人群和人群特征；将所述目标人群和所述人群特征所圈定的人群集合，作为所述种子人群。
在一个实施方式中，目标人群可以包括但不限于以下至少之一：未触发用户、潜在用户、行动用户和购买用户；所述人群特征包括以下至少之一：性别、年龄、地域。
在一个实施方式中，扩展目标可以包括但不限于以下至少之一：品牌推广、效果推广、点击优先、转化优先。
在软件层面，确定推广人群的装置，位于终端中，可以包括：第一显示模块、第一接收模块、第二显示模块和第二接收模块，其中：
第一显示模块，用于显示种子人群、扩展数量和扩展目标设定界面；
第一接收模块，用于接收用户设定的种子人群、扩展数量和扩展目标；
第二显示模块，用于显示通过所述种子人群、扩展数量和扩展目标，确定的显著特征组，其中，所述显著特征组中包括有多个显著特征；
第二接收模块，用于接收用户对所述显著特征组的设置操作，其中，所述设置操作用于对所述显著特征组进行设定，基于设定后的显著特征组确定相似人群，得到所述扩展数量个推广对象。
在一个实施方式中，上述第二显示模块可以显示所述显著特征组和所述显著特征组中各个特征的权重值；相应的，第二接收模块可以接收用户所述显著特征组中的特征的删除操作、对权重值的修改操作。
本申请达到了如下效果：为有推广需求的用户设置了可以设定种子人群、扩展数量、扩展目标，以及显著特征组的入口，从而使得用户的扩展意愿可以得到有效表达，解决了现有的人群扩展和筛选对用户是透明的自动化的方式所导致的用户无法把控扩展方向，而导致的扩展结果不满足用户需求的技术问题，达到了使得用户的扩展意愿可以得到有效表达的技术效果，大大提升了用户体验。
虽然本申请提供了如实施例或流程图所述的方法操作步骤，但基于常规或者无创造性的劳动可以包括更多或者更少的操作步骤。实施例中列举的步骤顺序仅仅为众多步骤执行顺序中的一种方式，不代表唯一的执行顺序。在实际中的装置或客户端产品执行时，可以按照实施例或者附图所示的方法顺序执行或者并行执行(例如并行处理器或者多线程处理的环境)。
上述实施例阐明的装置或模块，具体可以由计算机芯片或实体实现，或者由具有某种功能的产品来实现。为了描述的方便，描述以上装置时以功能分为各种模块分别描述。在实施本申请时可以把各模块的功能在同一个或多个软件和/或硬件中实现。当然，也可以将实现某功能的模块由多个子模块或子单元组合实现。
本申请中所述的方法、装置或模块可以以计算机可读程序代码方式实现控制器按任何适当的方式实现，例如，控制器可以采取例如微处理器或处理器以及存储可由该(微)处理器执行的计算机可读程序代码(例如软件或固件)的计算机可读介质、逻辑门、开关、专用集成电路(Application Specific Integrated Circuit，ASIC)、可编程逻辑控制器和嵌入微控制器的形式，控制器的例子包括但不限于以下微控制器：ARC 625D、Atmel AT91SAM、Microchip PIC18F26K20以及Silicone Labs C8051F320，存储器控制器还可以被实现为存储器的控制逻辑的一部分。本领域技术人员也知道，除了以纯计算机可读程序代码方式实现控制器以外，完全可以通过将方法步骤进行逻辑编程来使得控制器以逻辑门、开关、专用集成电路、可编程逻辑控制器和嵌入微控制器等的形式来实现相同功能。因此这种控制器可以被认为是一种硬件部件，而对其内部包括的用于实现各种功 能的装置也可以视为硬件部件内的结构。或者甚至，可以将用于实现各种功能的装置视为既可以是实现方法的软件模块又可以是硬件部件内的结构。
本申请所述装置中的部分模块可以在由计算机执行的计算机可执行指令的一般上下文中描述，例如程序模块。一般地，程序模块包括执行特定任务或实现特定抽象数据类型的例程、程序、对象、组件、数据结构、类等等。也可以在分布式计算环境中实践本申请，在这些分布式计算环境中，由通过通信网络而被连接的远程处理设备来执行任务。在分布式计算环境中，程序模块可以位于包括存储设备在内的本地和远程计算机存储介质中。
通过以上的实施方式的描述可知，本领域的技术人员可以清楚地了解到本申请可借助软件加必需的硬件的方式来实现。基于这样的理解，本申请的技术方案本质上或者说对现有技术做出贡献的部分可以以软件产品的形式体现出来，也可以通过数据迁移的实施过程中体现出来。该计算机软件产品可以存储在存储介质中，如ROM/RAM、磁碟、光盘等，包括若干指令用以使得一台计算机设备(可以是个人计算机，移动终端，服务器，或者网络设备等)执行本申请各个实施例或者实施例的某些部分所述的方法。
本说明书中的各个实施例采用递进的方式描述，各个实施例之间相同或相似的部分互相参见即可，每个实施例重点说明的都是与其他实施例的不同之处。本申请的全部或者部分可用于众多通用或专用的计算机系统环境或配置中。例如：个人计算机、服务器计算机、手持设备或便携式设备、平板型设备、移动通信终端、多处理器系统、基于微处理器的系统、可编程的电子设备、网络PC、小型计算机、大型计算机、包括以上任何系统或设备的分布式计算环境等等。
虽然通过实施例描绘了本申请，本领域普通技术人员知道，本申请有许多变形和变化而不脱离本申请的精神，希望所附的权利要求包括这些变形和变化而不脱离本申请的精神。

Claims (14)

  1. 一种广告投放方法，其特征在于，所述方法包括：
    基于用户设定的种子人群参数、扩展参数和扩展目标参数，确定出用于投放广告的显著特征组，其中，所述显著特征组中包括有多个显著特征，以及各个显著特征的权重值；
    基于用户对所述显著特征组的选择，对所述显著特征组进行调整；
    根据调整后的显著特征组确定相似人群，得到用于投放所述广告的推广对象。
  2. 根据权利要求1所述的方法，其特征在于，用户对所述显著特征组的选择，包括：对显著特征的删除、对显著特征的权重值的修改。
  3. 根据权利要求1所述的方法，其特征在于，基于用户设定的种子人群参数、扩展参数和扩展目标参数，确定出用于投放广告的显著特征组，包括：
    根据预先建立的目标参数与适配特征组之间的对应关系，匹配出所述扩展目标参数对应的适配特征组；
    将种子人群参数作为正样本，确定出所述适配特征组中各个特征的重要性程度；
    将重要性程度满足预设要求的特征作为确定出的显著特征组中的特征。
  4. 根据权利要求3所述的方法，其特征在于，将重要性程度满足预设要求的特征作为确定出的显著特征组中的特征，包括：
    在确定出显著特征组中的特征对应的候选人群满足所述扩展参数中的扩展数量的情况下，将重要性程度满足预设要求的特征作为确定出的显著特征组中的特征。
  5. 根据权利要求1所述的方法，其特征在于，基于用户设定的种子人群参数、扩展参数和扩展目标参数，确定出用于投放广告的显著特征组，包括：
    按照以下方式至少之一确定用于投放广告的显著特征组：点互信息指数、信息增益、卡方检验值。
  6. 根据权利要求2所述的方法，其特征在于，根据调整后的显著特征组确定相似人群，得到用于投放广告的推广对象，包括：
    将调整后的显著特征组中各个特征所覆盖的人群的集合作为候选人群，其中，调整后的显著特征组中包括：扩展数量；
    根据调整后的显著特征组中各个显著特征的特征值，对所述候选人群进行相关度排序；
    将相关度最高的扩展数量个对象作为推广对象。
  7. 根据权利要求1所述的方法，其特征在于，种子人群参数，包括：目标人群和人群特征，其中，所述目标人群和人群特征所圈定的人群组合为种子人群。
  8. 根据权利要求7所述的方法，其特征在于，所述目标人群包括以下至少之一：未触发用户、潜在用户、行动用户和购买用户；所述人群特征包括以下至少之一：性别、年龄、地域。
  9. 根据权利要求1所述的方法，其特征在于，所述扩展目标参数包括以下至少之一：品牌推广、效果推广、点击优先、转化优先。
  10. 一种确定推广人群的方法，其特征在于，所述方法包括：
    显示种子人群参数、扩展参数和扩展目标参数的设定界面；
    接收用户设定的种子人群参数、扩展参数和扩展目标参数；
    基于用户设定的种子人群参数、扩展参数和扩展目标参数，确定并显示显著特征组，其中，所述显著特征组中包括有多个显著特征；
    接收用户对所述显著特征组的选择操作，其中，所述选择操作用于对所述显著特征组进行调整，基于调整后的显著特征组确定相似人群，得到推广对象。
  11. 根据权利要求10所述的方法，其特征在于，接收用户对所述显著特征组的选择操作，包括：
    显示所述显著特征组和所述显著特征组中各个特征的权重值；
    接收用户对所述显著特征组的选择操作，其中，所述选择操作包括：对所述显著特征组中的特征的删除操作、对权重值的修改操作。
  12. 一种服务器，包括处理器以及用于存储处理器可执行指令的存储器，所述处理器执行所述指令时实现权利要求1至9中任一项所述方法的步骤。
  13. 一种客户端，包括处理器以及用于存储处理器可执行指令的存储器，所述处理器执行所述指令时实现权利要求10至11中任一项所述方法的步骤。
  14. 一种计算机可读存储介质，其上存储有计算机指令，所述指令被执行时实现权利要求1至9中任一项所述方法的步骤。
PCT/CN2019/101998 2018-08-27 2019-08-22 广告投放方法、确定推广人群的方法、服务器和客户端 WO2020043001A1 (zh)

Applications Claiming Priority (2)

Application Number Priority Date Filing Date Title
CN201810980484.9 2018-08-27
CN201810980484.9A CN110866766A (zh) 2018-08-27 2018-08-27 广告投放方法、确定推广人群的方法、服务器和客户端

Publications (1)

Publication Number Publication Date
WO2020043001A1 true WO2020043001A1 (zh) 2020-03-05

Family

ID=69643892

Family Applications (1)

Application Number Title Priority Date Filing Date
PCT/CN2019/101998 WO2020043001A1 (zh) 2018-08-27 2019-08-22 广告投放方法、确定推广人群的方法、服务器和客户端

Country Status (2)

Country Link
CN (1) CN110866766A (zh)
WO (1) WO2020043001A1 (zh)

Cited By (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN112633912A (zh) * 2020-11-19 2021-04-09 杭州五色云文化传播有限公司 基于人物画像数据的广告投放系统及投放方法
CN113450130A (zh) * 2020-12-15 2021-09-28 北京新氧科技有限公司 一种页面广告投放方法、装置、设备及存储介质

Families Citing this family (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN111612516A (zh) * 2020-04-11 2020-09-01 上海淇玥信息技术有限公司 一种基于rta投放的dmp平台、方法及电子设备
CN111899049A (zh) * 2020-07-23 2020-11-06 广州视源电子科技股份有限公司 广告投放方法、装置及设备
CN113222651B (zh) * 2021-04-29 2024-05-07 西安点告网络科技有限公司 广告投放模型统计类特征离散化方法、系统、设备及介质
CN113191808B (zh) * 2021-04-30 2023-04-25 北京深演智能科技股份有限公司 发布人群包的方法及装置
CN113393269B (zh) * 2021-06-11 2024-03-15 上海明略人工智能(集团)有限公司 触点媒体转化率的确定方法、装置、电子设备和存储介质

Citations (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US20120197733A1 (en) * 2011-01-27 2012-08-02 Linkedln Corporation Skill customization system
CN105427129A (zh) * 2015-11-12 2016-03-23 腾讯科技（深圳）有限公司 一种信息的投放方法及系统
CN106570718A (zh) * 2015-10-13 2017-04-19 深圳市腾讯计算机系统有限公司 信息的投放方法及投放系统
CN108280670A (zh) * 2017-01-06 2018-07-13 腾讯科技（深圳）有限公司 种子人群扩散方法、装置以及信息投放系统
CN108427690A (zh) * 2017-02-15 2018-08-21 腾讯科技（深圳）有限公司 信息投放方法及装置

Family Cites Families (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN108241990A (zh) * 2016-12-26 2018-07-03 北京奇虎科技有限公司 移动广告投放配置、控制方法及其相应的装置

Patent Citations (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US20120197733A1 (en) * 2011-01-27 2012-08-02 Linkedln Corporation Skill customization system
CN106570718A (zh) * 2015-10-13 2017-04-19 深圳市腾讯计算机系统有限公司 信息的投放方法及投放系统
CN105427129A (zh) * 2015-11-12 2016-03-23 腾讯科技（深圳）有限公司 一种信息的投放方法及系统
CN108280670A (zh) * 2017-01-06 2018-07-13 腾讯科技（深圳）有限公司 种子人群扩散方法、装置以及信息投放系统
CN108427690A (zh) * 2017-02-15 2018-08-21 腾讯科技（深圳）有限公司 信息投放方法及装置

Cited By (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN112633912A (zh) * 2020-11-19 2021-04-09 杭州五色云文化传播有限公司 基于人物画像数据的广告投放系统及投放方法
CN112633912B (zh) * 2020-11-19 2023-10-27 杭州五色云文化传播有限公司 基于人物画像数据的广告投放系统及投放方法
CN113450130A (zh) * 2020-12-15 2021-09-28 北京新氧科技有限公司 一种页面广告投放方法、装置、设备及存储介质
CN113450130B (zh) * 2020-12-15 2024-03-19 北京新氧科技有限公司 一种页面广告投放方法、装置、设备及存储介质

Also Published As

Publication number Publication date
CN110866766A (zh) 2020-03-06

Similar Documents

Publication Publication Date Title
WO2020043001A1 (zh) 广告投放方法、确定推广人群的方法、服务器和客户端
US11288702B1 (en) Exploration in a real time messaging platform
US8543518B2 (en) Deducing shadow user profiles for ad campaigns
US20170098236A1 (en) Exploration of real-time advertising decisions
US10223742B2 (en) Systems and methods for selecting third party content based on feedback
US10134053B2 (en) User engagement-based contextually-dependent automated pricing for non-guaranteed delivery
US9460451B2 (en) Quality scoring system for advertisements and content in an online system
KR101644667B1 (ko) 확장가능한 클러스터링
TWI648642B (zh) Data search processing method and system
US20160191450A1 (en) Recommendations Engine in a Layered Social Media Webpage
US20160188734A1 (en) Method and apparatus for programmatically synthesizing multiple sources of data for providing a recommendation
US20160253325A1 (en) Method and apparatus for programmatically adjusting the relative importance of content data as behavioral data changes
US20160283481A1 (en) Method and apparatus for combining text search and recommendation engines
US20160189207A1 (en) Enhanced online content delivery system using action rate lift
US11288709B2 (en) Training and utilizing multi-phase learning models to provide digital content to client devices in a real-time digital bidding environment
EP2821952A1 (en) Unified marketplace for advertisements and content in an online system
US20150278877A1 (en) User Engagement-Based Contextually-Dependent Automated Reserve Price for Non-Guaranteed Delivery Advertising Auction
US20140006172A1 (en) Method of calculating a reserve price for an auction and apparatus conducting the same
CN110163703B (zh) 一种分类模型建立方法、文案推送方法和服务器
US20210357977A1 (en) Method for matching influencers to brands and pairing influencer videos to products for intra-video transactions
CN109903086B (zh) 一种相似人群扩展方法、装置及电子设备
US20120116875A1 (en) Providing advertisements based on user grouping
CN111767466A (zh) 基于人工智能的推荐信息推荐方法、装置及电子设备
WO2022095701A1 (zh) 一种对象推荐方法及装置、设备、存储介质
KR102453535B1 (ko) 온라인 쇼핑 플랫폼을 제공하는 방법 및 장치

Legal Events

Date Code Title Description
121 Ep: the epo has been informed by wipo that ep was designated in this application

Ref document number: 19855742

Country of ref document: EP

Kind code of ref document: A1

NENP Non-entry into the national phase

Ref country code: DE

122 Ep: pct application non-entry in european phase

Ref document number: 19855742

Country of ref document: EP

Kind code of ref document: A1