WO2020147429A1 - 广告数据生成 - Google Patents

广告数据生成 Download PDF

Info

Publication number
WO2020147429A1
WO2020147429A1 PCT/CN2019/120677 CN2019120677W WO2020147429A1 WO 2020147429 A1 WO2020147429 A1 WO 2020147429A1 CN 2019120677 W CN2019120677 W CN 2019120677W WO 2020147429 A1 WO2020147429 A1 WO 2020147429A1
Authority
WO
WIPO (PCT)
Prior art keywords
information
screened
advertisement template
target
objects
Prior art date
Application number
PCT/CN2019/120677
Other languages
English (en)
French (fr)
Inventor
马武
杨肖
徐凤阳
张腾
顾惟祎
Original Assignee
北京三快在线科技有限公司
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 北京三快在线科技有限公司 filed Critical 北京三快在线科技有限公司
Publication of WO2020147429A1 publication Critical patent/WO2020147429A1/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

Definitions

  • the embodiments of the present disclosure relate to the field of Internet technology, and in particular, to a method, system, electronic device, and storage medium for generating advertisement data.
  • Advertising is an important means of monetizing Internet traffic. Appropriate advertising will promote a win-win situation for platforms, advertisers and users.
  • the general platform is pre-configured with one or several advertising templates. After the advertiser provides the advertising content, the platform will fixedly determine the advertising template displayed on the current page according to the business logic, and fill the advertising content into the advertising template.
  • the display of the advertisement is processed only in accordance with the business logic, and the display method and template are relatively monotonous, which makes users easy to cause aesthetic fatigue.
  • the embodiments of the present disclosure provide a method, system, electronic device, and storage medium for generating advertisement data to solve the above-mentioned problems of advertisement data generation in the prior art.
  • the embodiments of the present disclosure provide a method for generating advertisement data, which is applied to a trading platform that involves multiple user objects, multiple merchant objects, multiple advertisement template objects, and the advertisement template objects It has corresponding multiple material information, including:
  • the target material information is filled into the target advertisement template object to generate advertisement data.
  • the advertisement template object is provided with scene identification information; the step of filtering out the target advertisement template object from the plurality of advertisement template objects according to the first characteristic information and the second characteristic information includes :
  • the second characteristic information is used to determine the target advertisement template object in the set of advertisement template objects to be screened.
  • the first feature information includes one or more feature tags; the step of using the first feature information and the scene identification information to determine the set of advertisement template objects to be screened includes:
  • the first preset number of advertisement template objects are selected as the set of advertisement template objects to be filtered.
  • the step of using the second characteristic information to determine the target advertisement template object in the set of advertisement template objects to be screened includes:
  • the current advertising template object matches the second feature information, it is determined that the current advertising template object is the target advertising template object.
  • the advertisement template object is provided with scene identification information, and the scene identification information includes one or more scene tags; the first feature information includes one or more feature tags; Information and second characteristic information, the step of filtering out target advertisement template objects from the plurality of advertisement template objects includes:
  • the advertisement template object corresponding to the maximum value in the second matching value is the target advertisement template object.
  • the preset exposure threshold includes a first threshold and a second threshold less than the first threshold; the priority identification information includes the first priority identification, the second priority identification, and the third priority identification.
  • the step of generating priority identification information corresponding to the material information to be screened according to the historical exposure and the preset exposure threshold includes:
  • the first priority identification is generated
  • the third priority identification is generated.
  • the step of determining the target material information based on the priority identification information and the estimated click-through rate includes:
  • the preset probability is used to randomly select one of the material information to be screened as the target material information;
  • the set probability includes a first probability corresponding to the first identifier, a second probability corresponding to the second identifier, and a third probability corresponding to the third identifier, the first probability being greater than the second probability and the third probability .
  • the material information further includes material elements; the material information to be screened is generated by the following method:
  • a second preset number of candidate material information is selected as the material information to be screened.
  • the estimated click-through rate is generated by the following method:
  • the input data includes: the first feature information, the scene identification information, and the material element;
  • the model input data is sent to a preset logistic regression model to generate the predicted click rate.
  • the method further includes:
  • the step of filling the target material information into the target advertisement template object to generate advertisement data includes:
  • the embodiments of the present disclosure provide an advertisement data generation system, which is applied to a trading platform that involves multiple user objects, multiple business objects, multiple advertising template objects, and the advertising template objects It has corresponding multiple material information, including:
  • a collection module configured to collect the first characteristic information of the user object and the second characteristic information of the merchant object
  • a template sorting module configured to filter out target advertisement template objects from the plurality of advertisement template objects according to the first characteristic information and the second characteristic information;
  • the material decision module is used to determine the target material information corresponding to the target advertisement template object
  • the online configuration module is used to fill the target material information into the target advertisement template object to generate advertisement data.
  • the advertisement template object is set with scene identification information;
  • the template sorting module includes:
  • the first object to be screened submodule is configured to use the first feature information and the scene identification information to determine a set of advertisement template objects to be screened;
  • the first target object determination sub-module is configured to use the second characteristic information to determine the target advertisement template object in the set of advertisement template objects to be screened.
  • the first feature information includes one or more feature tags;
  • the first object to be screened submodule includes:
  • a scene identification unit configured to determine identification information of multiple scenes to be screened that match the first characteristic information
  • the first matching unit is configured to calculate the number of matches between the scene label and the feature label in the scene identification information to be screened, and generate a first matching value
  • the first sorting unit is configured to sort the first matching values in a descending manner to generate matching value sorting information
  • the to-be-screened object unit is configured to sort the information according to the matching value, and select a first preset number of advertisement template objects as the to-be-screened advertisement template object set.
  • the first target object determination submodule includes:
  • the second feature matching unit is configured to sequentially determine whether the advertisement template objects in the set of advertisement template objects to be screened match the second feature information
  • the target object determining unit is configured to determine that the current advertising template object is the target advertising template object if the current advertising template object matches the second characteristic information.
  • the advertisement template object is provided with scene identification information, the scene identification information includes one or more scene tags; the first characteristic information includes one or more characteristic tags; the template sorting module includes:
  • the second object-to-be-screened neutron module is used to determine a plurality of advertising template objects to-be-screened that match the first feature information and the second feature information;
  • the second matching sub-module is used to calculate the number of matches between the scene label corresponding to the advertisement template object to be screened and the feature label, and generate a second matching value
  • the second target object determining sub-module is configured to determine the advertisement template object corresponding to the maximum value in the second matching value as the target advertisement template object.
  • the material information includes historical exposure and estimated click rate;
  • the material decision-making module includes:
  • the material to be screened sub-module is used to determine multiple material information to be screened corresponding to the target advertisement template object;
  • An identification generating module configured to generate priority identification information corresponding to the material information to be screened according to the historical exposure and the preset exposure threshold;
  • the target material sub-module is used to determine the target material information according to the priority identification information and the estimated click rate.
  • the preset exposure threshold includes a first threshold and a second threshold less than the first threshold;
  • the priority identification information includes the first priority identification, the second priority identification, and the third priority identification.
  • the identification generating module includes:
  • the first priority identification unit is configured to generate the first priority identification when the historical exposure of the material information to be screened is greater than the first threshold;
  • a second priority identification unit configured to generate the second priority identification when the historical exposure of the material information to be screened is less than the first threshold and greater than the second threshold;
  • the third priority identification unit is configured to generate the third priority identification when the historical exposure of the material information to be screened is less than the second threshold.
  • the target material sub-module includes:
  • the first material selection unit is configured to determine that the material information corresponding to the maximum expected click-through rate is the target material information when there is only material information to be screened corresponding to the first priority identification;
  • the second material selection unit is configured to randomly select one of the material information to be screened as the target material information when there is no material information to be screened corresponding to the first priority identification;
  • the third material selection unit is used for when there are screening material information corresponding to the first priority identification, the second priority identification, and the third priority identification at the same time, using a preset probability to randomly select one of the material information to be screened as the Target material information; wherein the preset probability includes a first probability corresponding to the first identifier, a second probability corresponding to the second identifier, and a third probability corresponding to the third identifier, the first probability being greater than the first probability The second probability and the third probability.
  • the material information further includes material elements; the material information to be screened is generated by the following modules:
  • the candidate material module is used to determine candidate material information that matches the advertisement template object
  • the quality scoring module is used to use a preset convolutional neural network to score the material elements of the candidate material information to generate a quality scoring value
  • the quality ranking module is used to rank the quality score values in a descending manner to generate quality ranking information
  • the material selection module is configured to sort the information according to the quality, and select the second preset number of candidate material information as the material information to be screened.
  • the estimated click-through rate is generated by the following modules:
  • the model input module is used to obtain model input data;
  • the input data includes: the first feature information, the scene identification information, and the material element;
  • the click-through rate generating module is used to send the model input data to a preset logistic regression model to generate the predicted click-through rate.
  • system further includes:
  • the click-through rate sorting module is used to sort the predicted click-through rates in a descending manner to generate predicted click-sort information
  • the advertisement data sorting module is used to sort the advertisement data by using the predicted click sort information
  • the display module is used to display the sorted advertisement data.
  • the online configuration module includes:
  • the material content assembly configuration sub-module is used to fill the material elements in the target material information into the target advertisement template object;
  • the material style assembly configuration sub-module is used to configure a corresponding display format for each material element in the target advertisement template object.
  • an electronic device including:
  • an embodiment of the present disclosure provides a computer program, including computer-readable code, which when the computer-readable code runs on a computing processing device, causes the computing processing device to execute the aforementioned advertising data generation method.
  • embodiments of the present disclosure provide a computer-readable storage medium in which the aforementioned computer program is stored.
  • the embodiment of the present disclosure provides a method for generating advertisement data.
  • the target advertisement target objects are filtered out from the preset advertisement template objects, the target material information corresponding to the target advertisement template objects is extracted, and the target material information is filled into the target advertisement template objects, so as to realize the first characteristic information according to different user objects , To generate advertisement data matching the user object, and improve the adaptability of the advertisement data and the user object.
  • generating advertisement data suitable for different user objects can promote the success rate of transactions between user objects and merchant objects.
  • Fig. 1 shows a flow chart of the steps of the first embodiment of a method for generating advertisement data of the present disclosure
  • Fig. 2 shows a flowchart of the steps of the second embodiment of an advertisement data generating method of the present disclosure
  • Fig. 3 shows a structural block diagram of an embodiment of an advertisement data generation system of the present disclosure
  • FIG. 4 schematically shows a block diagram of an electronic device for executing the method according to the present disclosure.
  • Fig. 5 schematically shows a storage unit for holding or carrying program codes for implementing the method according to the present disclosure.
  • FIG. 1 shows a flow chart of the steps in the first embodiment of an advertisement data generation method of the present disclosure.
  • the method can be applied to a trading platform that involves multiple user objects, multiple merchant objects, and multiple An advertisement template object, and the advertisement template object has multiple corresponding material information, including:
  • the advertising template object can be a template that displays material information according to the set divided areas.
  • Step 101 Collect first characteristic information of the user object and second characteristic information of the merchant object;
  • the first characteristic information may be characteristics of one or more dimensions of the buyer user, for example: current location, gender, occupation, and so on.
  • the second feature information may be features of one or more dimensions of the seller user, for example: store location, business category, etc.
  • Step 102 Filter out target advertisement template objects from the plurality of advertisement template objects according to the first characteristic information and the second characteristic information;
  • the advertisement template object with the highest degree of matching with the first characteristic information and the second characteristic information is filtered out, and the advertisement template object is determined It is the target advertising template.
  • Step 103 Determine target material information corresponding to the target advertisement template object
  • Different advertisement template objects may be associated with different material information. After the target advertisement template object is determined, the material information corresponding to the target advertisement template object can be extracted as the target material information.
  • Step 104 Fill the target material information into the target advertisement template object to generate advertisement data.
  • the extracted target material information is filled into the target advertisement template object, so that the target material information is displayed according to a certain division rule, and advertisement data is generated.
  • the present disclosure by collecting the first characteristic information of the user object of the trading platform and the second characteristic information of the merchant object, according to the first characteristic information and the second characteristic information, among a plurality of preset advertising template objects Filter out the target advertising target object, extract the target material information corresponding to the target advertising template object, and fill the target material information into the target advertising template object, so as to realize that it can generate matching user objects according to the first characteristic information of different user objects Advertisement data to improve the adaptability of advertisement data and user objects.
  • generating advertisement data suitable for different user objects can promote the success rate of transactions between user objects and merchant objects.
  • the generation of advertisement data is generally realized through the access layer, the service layer, the retrieval layer, the index layer and the data layer.
  • the access layer is used to receive business requests from the front-end SDK (Software Development Kit), generate request rewrite messages, and request advertising content from the business layer; and after obtaining the advertising content returned by the business layer, complete the style assembly And send it to SDK.
  • SDK Software Development Kit
  • the business layer is used to specify a kind of advertisement template object according to the logic of the product (for example: application platform) after receiving the request to rewrite the information; and request the advertisement content from the retrieval layer; after the retrieval layer returns the result, obtain the unique according to the product logic After assembling the specified advertising template object and unique material information, it returns to the access layer.
  • the retrieval layer is used to retrieve business objects and corresponding material information that match the advertisement template objects.
  • the index layer is used to index data in the data layer for retrieval by the retrieval layer.
  • the data layer is used to access the underlying data, including advertising data and material information sets.
  • FIG. 2 there is shown a step flow chart of the second embodiment of an advertisement data generation method of the present disclosure.
  • the method can be applied to a trading platform that involves multiple user objects, multiple merchant objects, and multiple The advertisement template object, and the advertisement template object has multiple corresponding material information
  • the embodiment of the present disclosure may specifically include the following steps:
  • the trading platform can connect multiple user objects and multiple merchant objects through the Internet, and the user objects can conduct transactions with the merchant objects on the trading platform.
  • the trading platform may be a group buying website, a group buying APP (application, application), a takeaway website, a takeaway APP, etc.
  • the user object may be a buyer user of the trading platform, and the merchant object may be a seller user of the trading platform (for example, a virtual store).
  • the advertising template object can be a template that displays material information according to the set divided areas.
  • Advertisement template objects can have corresponding one or more material information.
  • advertising template object A corresponds to material a, material b, and material c.
  • Template B corresponds to material a, material c, and material d.
  • Step 201 Collect first characteristic information of the user object and second characteristic information of the merchant object;
  • the first characteristic information can be acquired by the trading platform in real time, or it can be customized and transmitted to the trading platform by the user object.
  • the first characteristic information can be acquired by the trading platform in real time, or it can be determined by the historical information read by the trading platform and the user object saved to the server.
  • the first characteristic information may be characteristics of one or more dimensions of the buyer user.
  • the first characteristic information may include, for example, current location, gender, work, and so on.
  • the second feature information may be features of one or more dimensions of the seller user, for example: store location, business category, etc.
  • Step 202 Filter out target advertisement template objects from the plurality of advertisement template objects according to the first characteristic information and the second characteristic information;
  • the advertisement template object with the highest degree of matching with the first characteristic information and the second characteristic information is filtered out, and the advertisement template object is determined It is the target advertising template.
  • the embodiment of the present disclosure may perform step 202 when selecting advertisement template objects at the business layer, thereby determining the target advertisement template object with the highest degree of matching with the first characteristic information and the second characteristic information among the plurality of advertisement template objects, and The advertisement data is generated according to the target advertisement template object.
  • step 202 may include:
  • Sub-step S11 using the first feature information and the scene identification information to determine a set of advertisement template objects to be screened;
  • the advertisement template object may be provided with scene identification information, and a set of advertisement template objects to be filtered is determined according to the first feature information and the scene identification information.
  • the set of advertisement template objects to be screened may consist of one or more advertisement template objects, and the scene identification information of the advertisement template objects in the set of advertisement template objects to be screened matches the first feature information.
  • the first feature information includes one or more feature tags; sub-step S11 may include:
  • Sub-step S111 determining identification information of multiple scenes to be screened that match the first feature information
  • Sub-step S112 calculating the number of matches between the scene tags in the identification information of the scene to be screened and the feature tags, and generating a first matching value
  • Sub-step S113 sorting the first matching values in a descending manner to generate matching value sorting information
  • Sub-step S114 according to the matching value sorting information, select the first preset number of advertisement template objects as the set of advertisement template objects to be filtered.
  • Feature tags include platform identification information (e.g. platform name), channel information (e.g. APP, PC (Personal Computer)), occupational information, gender information, designated location information, intended transaction category, order preference information (e.g. : One or more of discount preference, new product preference).
  • platform identification information e.g. platform name
  • channel information e.g. APP, PC (Personal Computer)
  • occupational information e.g., gender information, designated location information, intended transaction category
  • order preference information e.g. : One or more of discount preference, new product preference.
  • the scene tag in the scene identification information may include platform identification information, channel information, and at least one of gender information, occupation information, and business classification information.
  • the trading platform may be provided with multiple advertising template objects, and the first feature information may have different matching degrees with the scene identification information of the multiple advertising template objects, specifically, the feature tag in the first feature information and the scene in the scene identification information
  • the number of matches for the tags is different. Calculate the number of matches between the scene label and the feature label in the scene information to be filtered, and generate a first matching value. For the generated first matching value, sort in descending order to generate matching value sorting information.
  • the first preset number of advertising template objects are determined in order as the set of advertising template to be filtered (for example, the trading platform may preset 200 advertising template objects, and the first preset number It can be set to 50, and the set of advertisement templates to be screened includes 50 advertisement template objects to be screened).
  • the larger the matching value corresponding to the scene identification information the higher the matching degree between the first feature information and the scene identification information, and the more the scene identification information matches the corresponding user object.
  • the first feature information may include the following feature tags: APP, male, white-collar, and dining.
  • the scene tags of the scene identification information A are: PC, female; the scene tags of the scene identification information B are: PC, male, shopping; the scene tags of the scene identification information C are: APP, male, dining.
  • the first matching values corresponding to the first feature information and the above three scene identification information are 0, 1, and 3, respectively, and the matching degree between the scene identification information C and the first feature information is the highest.
  • the platform identification information and channel information in the scene identification information must match the first feature information, otherwise, the corresponding matching value is directly set to 0, so as to ensure that the advertisement template objects to be screened meet the first feature information
  • the indicated platform identification information and channel information it can be set that the platform identification information and channel information in the scene identification information must match the first feature information, otherwise, the corresponding matching value is directly set to 0, so as to ensure that the advertisement template objects to be screened meet the first feature information The indicated platform identification information and channel information.
  • Sub-step S12 using the second characteristic information to determine the target advertisement template object in the set of advertisement template objects to be screened.
  • Different advertising template objects may be suitable for different business objects (for example: advertising template object A is suitable for business object A and business object B; advertising template object B is suitable for business object A and business object C), and the second characteristic information can be used , Determine the target advertisement template object suitable for the merchant object in the set of advertisement template objects to be screened.
  • sub-step S12 may include:
  • Sub-step S121 sequentially determine whether the advertisement template objects in the set of advertisement template objects to be screened match the second feature information
  • sub-step S122 if the current advertising template object matches the second feature information, it is determined that the current advertising template object is the target advertising template object.
  • the matching value ranking information it is determined in turn whether the advertisement template objects in the set of advertisement template objects to be screened match the second characteristic information. If the current advertisement template object matches the second characteristic information, it indicates the current advertisement If the template object is suitable for the merchant object, it is determined that the current advertisement template object is the target advertisement template object.
  • the second characteristic information may include, but is not limited to, business classification information, platform identification information, and channel information.
  • the scene identification information of the advertisement template object matches the second feature information, it is determined that the current advertisement template object is suitable for the merchant object.
  • the matching value corresponding to each advertisement template object can be calculated according to the first feature information and the scene tag generated offline, and the matching value can be calculated in real time according to the matching value.
  • the advertisement template object is provided with scene identification information, and the scene identification information includes one or more scene tags; the first characteristic information includes one or more characteristic tags;
  • step 202 can also be implemented through the following steps:
  • Sub-step S21 determining a plurality of advertisement template objects to be screened that match the first characteristic information and the second characteristic information
  • Sub-step S22 calculating the number of matches between the scene label corresponding to the advertisement template object to be screened and the feature label, and generating a second matching value
  • Sub-step S23 determining that the advertisement template object corresponding to the maximum value in the second matching value is the target advertisement template object.
  • the advertisement template object corresponding to the scene identification information that matches the first feature information and the second feature information is determined as the advertisement template object to be screened, and then the number of matching scene tags and feature tags corresponding to the advertisement template object to be screened is calculated at a time, and the generated
  • the second matching value determines that the advertisement template object corresponding to the maximum value in the second matching value is the target advertisement template object.
  • an LR Logistic Regression
  • Step 203 Determine multiple material information to be screened corresponding to the target template object
  • the material information corresponding to the target advertisement template object is determined as the material information to be screened, where the material information includes historical exposure and estimated click-through rate.
  • the historical exposure is the number of times the material information has been viewed by the user object for a certain period of time before the current moment.
  • the predicted click-through rate may be generated by using a preset model, combined with the first feature information and scene identification information.
  • Step 204 Generate priority identification information corresponding to the material information to be screened according to the historical exposure and the preset exposure threshold;
  • the historical exposure is compared with the exposure threshold to generate corresponding priority identification information.
  • the preset exposure threshold value includes a first threshold value and a second threshold value less than the first threshold value;
  • the priority identification information includes a first priority identification and a second priority identification And one of the third priority identification;
  • the step 204 may include:
  • Sub-step S31 when the historical exposure of the material information to be screened is greater than the first threshold, the first priority identification is generated;
  • Sub-step S32 when the historical exposure of the material information to be screened is less than the first threshold and greater than the second threshold, generate the second priority identifier
  • Sub-step S33 when the historical exposure of the material information to be screened is less than the second threshold, the third priority identification is generated.
  • the historical exposure of the material information to be screened is compared with the first threshold and the second threshold to determine that the current material information corresponds to one of the first priority identification, the second priority identification, and the third priority identification.
  • the priorities corresponding to the first priority identifier, the second priority identifier, and the third priority identifier are sequentially reduced.
  • Step 205 Determine the target material information according to the priority identification information and the estimated click rate.
  • the target material information After generating the priority identification information, combined with the estimated click-through rate, the target material information is determined.
  • the embodiments of the present disclosure can generate corresponding priority identifications based on the exposure and estimated click-through rate of the material information to be screened in an offline manner, and the priority identifications are used to assign different probabilities of being selected as the target material information to the corresponding material information to be screened, and After the advertisement data is displayed, the estimated click-through rate can be updated according to the user object's selection of the advertisement data.
  • the embodiment of the present disclosure can perform steps 203-205 when selecting material information through the retrieval layer, so as to determine the target material from multiple material information to be screened according to historical exposure rate, exposure rate presets, and estimated click-through rate. information.
  • step 205 may include:
  • Sub-step S41 when there is only material information to be screened corresponding to the first priority identification, determine the material information corresponding to the maximum value of the estimated click-through rate as the target material information;
  • Sub-step S42 when there is no material information to be screened corresponding to the first priority identification, randomly select one of the material information to be screened as the target material information;
  • a number can be assigned to each material information to be screened, and then a random number within the range of the number value will be generated in real time, which will correspond to the random number
  • the material information to be screened is the target material information.
  • Sub-step S43 when there are screening material information corresponding to the first priority identification, the second priority identification, and the third priority identification at the same time, use a preset probability to randomly select one of the material information to be screened as the target material information;
  • the preset probability includes a first probability corresponding to the first identifier, a second probability corresponding to the second identifier, and a third probability corresponding to the third identifier.
  • the first probability is greater than the second probability and the The third probability.
  • the first probability, the second probability, and the third probability respectively represent the probability of being selected corresponding to the first identification.
  • each priority identification information can be numbered separately, and then a random number is generated in real time.
  • One value corresponds to and only corresponds to one priority identification information, and the material information corresponding to the selected priority identification information is determined as the target material information through the random number.
  • Step 206 Fill the target material information into the target advertisement template object to generate advertisement data.
  • the extracted target material information is filled into the target advertisement template object, so that the target material information is displayed according to a certain division rule, and advertisement data is generated.
  • step 206 may include: filling material elements in the target material information into the target advertisement template object; configuring a corresponding display format for each material element in the target advertisement template object.
  • the display format may include, but is not limited to, text specifications, picture specifications, etc.
  • the embodiment of the present disclosure can execute the step of filling the material elements in the target material information into the target advertisement template object through the business layer, so as to realize the real-time combination of the target material information and target advertisement template data;
  • the steps of configuring the corresponding display format for each material element in the target advertisement template object are described, so as to realize the real-time configuration of the material element style and increase the beauty of the advertisement data.
  • Step 207 Sort the predicted click rates in a descending manner to generate predicted click ranking information
  • Step 208 Sort the advertisement data by using the predicted click sort information
  • Step 209 Display the sorted advertisement data.
  • the generated advertisement data is sorted, and when the user object logs into the trading platform, the sorted advertisement data is displayed to the user object.
  • the material information further includes material elements; the material information to be screened is generated by the following method: determining candidate material information that matches the advertising template object; using a preset convolution The neural network scores the material elements of the candidate material information to generate a quality score value; the quality score values are sorted in a descending manner to generate quality ranking information; according to the quality ranking information, select The second preset number of candidate material information is the material information to be screened.
  • the material information to be screened can be generated offline, that is, non-real-time generation, and the generated material information to be screened is stored in the data layer.
  • a neural convolutional network (CNN, Convolutional Neural Network) can be used.
  • the material elements in the material information are used as the input data of the convolutional network, and each material element is scored, and the corresponding quality score value is generated. According to the quality score value from higher To the smallest order, starting from the maximum value of the quality score, the material information whose number is the second preset number is selected as the material information to be filtered.
  • the quality score value in addition to determining the material information to be screened according to the quality ranking information, can also be divided into multiple scoring intervals, and a probability of being selected for each scoring interval can be set, by generating a second preset The random number of the number is determined. From multiple scoring intervals, select the interval to be selected in turn, and then randomly select a material information from the interval to be selected as the material information to be screened. Among them, depending on the order, the interval to be selected can appear repeatedly be chosen.
  • the predicted click-through rate is generated by the following method:
  • the input data includes: the first feature information, the scene information, and the material element;
  • the model input data is sent to a preset logistic regression model to generate the predicted click rate.
  • At least one dimension among the first feature information, the scene information, and the material element is added to determine the estimated click rate, so that the estimated click rate is more accurate and reliable.
  • the first feature information is matched with the scene information in the advertisement template object to obtain the corresponding matching degree
  • the advertisement template object with the highest preset number is the set of advertisement template objects to be screened, which is then used as the set of advertisement template objects to be screened, and the advertisement template objects with the second characteristic information are screened out as the target advertisement target objects.
  • Generate priority identification information corresponding to each material information in the target advertisement template object of the advertisement template determine the target material information corresponding to the target advertisement template object according to the priority identification information and preset click-through rate, and fill the target material information into the target advertisement Template object, so as to realize that the advertisement data matching the user object can be generated according to the first characteristic information of different user objects, and the adaptability of the advertisement data and the user object is improved. Because the first characteristic information, the scene information, The estimated click-through rate is calculated on the dimensions of the material element and the like, so that the estimated click-through rate is more accurate.
  • generating advertisement data suitable for different user objects can promote the success rate of transactions between user objects and merchant objects.
  • FIG. 3 there is shown a structural block diagram of an embodiment of an advertisement data generation system of the present disclosure.
  • the system can be applied to a transaction platform that involves multiple user objects, multiple business objects, and multiple advertising templates.
  • the object, and the advertisement template object has multiple corresponding material information, the system may specifically include the following modules:
  • the collection module 301 is configured to collect the first characteristic information of the user object and the second characteristic information of the merchant object;
  • the template sorting module 302 is configured to filter out target advertisement template objects from the plurality of advertisement template objects according to the first characteristic information and the second characteristic information;
  • the material decision module 303 is used to determine the target material information corresponding to the target advertisement template object
  • the online configuration module 304 is used to fill the target material information into the target advertisement template object to generate advertisement data.
  • the advertisement template object is set with scene identification information;
  • the template sorting module 302 includes:
  • the first object to be screened sub-module is configured to use the first feature information and the scene identification information to determine a set of advertisement template objects to be screened;
  • the first target object determination sub-module is configured to use the second characteristic information to determine the target advertisement template object in the set of advertisement template objects to be screened.
  • the first feature information includes one or more feature tags;
  • the first object to be screened submodule includes:
  • a scene identification unit configured to determine identification information of multiple scenes to be screened that match the first characteristic information
  • the first matching unit is configured to calculate the number of matches between the scene label and the feature label in the scene identification information to be screened, and generate a first matching value
  • the first sorting unit is configured to sort the first matching values in a descending manner to generate matching value sorting information
  • the to-be-screened object unit is configured to sort the information according to the matching value, and select a first preset number of advertisement template objects as the to-be-screened advertisement template object set.
  • the first target object determination submodule includes:
  • the second feature matching unit is configured to sequentially determine whether the advertisement template objects in the set of advertisement template objects to be screened match the second feature information
  • the target object determining unit is configured to determine that the current advertising template object is the target advertising template object if the current advertising template object matches the second characteristic information.
  • the advertisement template object is provided with scene identification information, and the scene identification information includes one or more scene tags; the first feature information includes one or more feature tags;
  • the template sorting module 302 includes:
  • the second object-to-be-screened neutron module is used to determine a plurality of advertising template objects to-be-screened that match the first feature information and the second feature information;
  • the second matching sub-module is used to calculate the number of matches between the scene label corresponding to the advertisement template object to be screened and the feature label, and generate a second matching value
  • the second target object determining sub-module is configured to determine the advertisement template object corresponding to the maximum value in the second matching value as the target advertisement template object.
  • the material information includes historical exposure and estimated click rate; the material decision module 303 includes:
  • the material to be screened sub-module is used to determine multiple material information to be screened corresponding to the target advertisement template object;
  • An identification generating module configured to generate priority identification information corresponding to the material information to be screened according to the historical exposure and the preset exposure threshold;
  • the target material sub-module is used to determine the target material information according to the priority identification information and the estimated click rate.
  • the preset exposure threshold includes a first threshold and a second threshold that is less than the first threshold;
  • the priority identification information includes a first priority identification and a second priority identification And one of the third priority identification;
  • the identification generation module includes:
  • the first priority identification unit is configured to generate the first priority identification when the historical exposure of the material information to be screened is greater than the first threshold;
  • a second priority identification unit configured to generate the second priority identification when the historical exposure of the material information to be screened is less than the first threshold and greater than the second threshold;
  • the third priority identification unit is configured to generate the third priority identification when the historical exposure of the material information to be screened is less than the second threshold.
  • the target material sub-module includes:
  • the first material selection unit is configured to determine that the material information corresponding to the maximum expected click-through rate is the target material information when there is only material information to be screened corresponding to the first priority identification;
  • the second material selection unit is configured to randomly select one of the material information to be screened as the target material information when there is no material information to be screened corresponding to the first priority identification;
  • the third material selection unit is used for when there are screening material information corresponding to the first priority identification, the second priority identification, and the third priority identification at the same time, using a preset probability to randomly select one of the material information to be screened as the Target material information; wherein the preset probability includes a first probability corresponding to the first identifier, a second probability corresponding to the second identifier, and a third probability corresponding to the third identifier, the first probability being greater than the first probability The second probability and the third probability.
  • the material information further includes material elements; the material information to be screened is generated by the following modules:
  • the candidate material module is used to determine candidate material information that matches the advertisement template object
  • the quality scoring module is used to use a preset convolutional neural network to score the material elements of the candidate material information to generate a quality scoring value
  • the quality ranking module is used to rank the quality score values in a descending manner to generate quality ranking information
  • the material selection module is configured to sort the information according to the quality, and select the second preset number of candidate material information as the material information to be screened.
  • the predicted click-through rate is generated by the following modules:
  • the model input module is used to obtain model input data;
  • the input data includes: the first feature information, the scene identification information, and the material element;
  • the click-through rate generating module is used to send the model input data to a preset logistic regression model to generate the predicted click-through rate.
  • system further includes:
  • the click-through rate sorting module is used to sort the predicted click-through rates in a descending manner to generate predicted click-sort information
  • the advertisement data sorting module is used to sort the advertisement data by using the predicted click sort information
  • the display module is used to display the sorted advertisement data.
  • the online configuration module 304 includes:
  • the material content assembly configuration sub-module is used to fill the material elements in the target material information into the target advertisement template object;
  • the material style assembly configuration sub-module is used to configure a corresponding display format for each material element in the target advertisement template object.
  • the description is relatively simple, and the relevant part can be referred to the description of the method embodiment.
  • the embodiment of the present disclosure also provides an electronic device, including: a processor, a memory, and a computer program stored on the memory and capable of running on the processor.
  • the processor executes the computer program when the computer program is executed.
  • the advertisement data generating method of the foregoing embodiment is also provided.
  • the embodiments of the present disclosure also provide a computer program, including computer-readable code, which, when the computer-readable code runs on a computing processing device, causes the computing processing device to execute the advertisement data generation method of the foregoing embodiment.
  • the embodiment of the present disclosure also provides a computer-readable storage medium in which the computer program of the foregoing embodiment is stored.
  • the device embodiments described above are merely illustrative.
  • the units described as separate components may or may not be physically separate, and the components displayed as units may or may not be physical units, that is, they may be located in One place, or it can be distributed to multiple network units. Some or all of the modules may be selected according to actual needs to achieve the purpose of the solution of this embodiment. Those of ordinary skill in the art can understand and implement without creative work.
  • the various component embodiments of the present disclosure may be implemented by hardware, or by software modules running on one or more processors, or by a combination of them.
  • a microprocessor or a digital signal processor (DSP) may be used in practice to implement some or all of the functions of some or all components in the computing processing device according to the embodiments of the present disclosure.
  • DSP digital signal processor
  • the present disclosure can also be implemented as a device or device program (for example, a computer program and a computer program product) for executing part or all of the methods described herein.
  • Such a program for realizing the present disclosure may be stored on a computer-readable storage medium, or may have the form of one or more signals.
  • Such a signal can be downloaded from an Internet website, or provided on a carrier signal, or provided in any other form.
  • FIG. 4 shows a computing processing device that can implement the method according to the present disclosure.
  • the computing processing device traditionally includes a processor 410 and a computer program product in the form of a memory 420 or a computer-readable storage medium.
  • the memory 420 may be an electronic memory such as flash memory, EEPROM (Electrically Erasable Programmable Read Only Memory), EPROM, hard disk, or ROM.
  • the memory 420 has a storage space 430 for executing the program code 431 of any method step in the foregoing method.
  • the storage space 430 for program codes may include various program codes 431 respectively used to implement various steps in the above method. These program codes can be read from or written into one or more computer program products.
  • These computer program products include program code carriers such as hard disks, compact disks (CDs), memory cards or floppy disks.
  • Such a computer program product is usually a portable or fixed storage unit as described with reference to FIG. 5.
  • the storage unit may have storage segments, storage spaces, etc. arranged similarly to the memory 420 in the computing processing device of FIG. 4.
  • the program code can be compressed in a suitable form, for example.
  • the storage unit includes computer-readable codes 431', that is, codes that can be read by, for example, a processor such as 410. These codes, when run by a computing processing device, cause the computing processing device to execute the method described above. The various steps.
  • any reference signs between parentheses should not be constructed as limitations on the claims.
  • the word “comprising” does not exclude the presence of elements or steps not listed in the claims.
  • the word “a” or “an” preceding an element does not exclude the presence of multiple such elements.
  • the present disclosure can be realized by means of hardware including several different elements and by means of a suitably programmed computer. In the unit claims enumerating several devices, several of these devices may be embodied by the same hardware item.
  • the use of the words first, second, and third does not indicate any order. These words can be interpreted as names.

Landscapes

  • Business, Economics & Management (AREA)
  • Engineering & Computer Science (AREA)
  • Accounting & Taxation (AREA)
  • Development Economics (AREA)
  • Strategic Management (AREA)
  • Finance (AREA)
  • Game Theory and Decision Science (AREA)
  • Entrepreneurship & Innovation (AREA)
  • Economics (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)

Abstract

本公开实施例提供了一种广告数据生成方法，涉及互联网技术领域，本公开实施例的广告数据生成方法应用于交易平台，所述交易平台涉及多个用户对象、多个商户对象、多个广告模版对象，以及所述广告模版对象具有对应的多个物料信息，包括：采集所述用户对象的第一特征信息，以及，所述商户对象的第二特征信息；依据所述第一特征信息和第二特征信息，从所述多个广告模板对象中，筛选出目标广告模版对象；确定与所述目标广告模版对象对应的目标物料信息；将所述目标物料信息填充至所述目标广告模版对象，生成广告数据。

Description

广告数据生成
本申请要求在2019年01月16日提交中国专利局、申请号为201910041430.0、发明名称为“一种广告数据生成方法、系统、电子设备及存储介质”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
本公开的实施例涉及互联网技术领域，尤其涉及一种广告数据生成方法、系统、电子设备和存储介质。
背景技术
广告是互联网流量变现的重要手段，合适的广告会促进平台、广告主和用户的共赢。
现有的技术方案中，一般平台预先设置有一个或数个广告模版，在广告主提供了广告内容之后，平台会根据业务逻辑固定地确定当前页面展示的广告模版，并将广告内容填充至该模版中，使得广告的展示只按照业务逻辑进行处理，展示方法和模版较为单调，使得用户容易造成审美疲劳。
发明内容
本公开的实施例提供一种广告数据生成方法、系统、电子设备和存储介质，以解决现有技术广告数据生成的上述问题。
第一方面，本公开的实施例提供了一种广告数据生成方法，应用于交易平台，所述交易平台涉及多个用户对象、多个商户对象、多个广告模版对象，以及所述广告模版对象具有对应的多个物料信息，包括：
采集所述用户对象的第一特征信息，以及，所述商户对象的第二特征信息；
依据所述第一特征信息和第二特征信息，从所述多个广告模板对象中，筛选出目标广告模版对象；
确定与所述目标广告模版对象对应的目标物料信息；
将所述目标物料信息填充至所述目标广告模版对象，生成广告数据。
可选地，所述广告模版对象设置有场景标识信息；所述依据所述第一特征信息和第二特征信息，从所述多个广告模板对象中，筛选出目标广告模版对象的步骤，包括：
采用所述第一特征信息和所述场景标识信息，确定待筛选广告模版对象集；
采用所述第二特征信息，在所述待筛选广告模版对象集中确定所述目标广告模版对象。
可选地，所述第一特征信息包括一个或多个特征标签；所述采用所述第一特征信息和所述场景标识信息，确定待筛选广告模版对象集的步骤，包括：
确定与所述第一特征信息匹配的多个待筛选场景标识信息；
计算所述待筛选场景标识信息中的场景标签与所述特征标签的匹配数量，生成第一匹配值；
采用从大到小的方式，对所述第一匹配值进行排序，生成匹配值排序信息；
按照所述匹配值排序信息，选择第一预置个数的广告模版对象为所述待筛选广告模 版对象集。
可选地，所述采用所述第二特征信息，在所述待筛选广告模版对象集中确定所述目标广告模版对象的步骤，包括：
依次判断所述待筛选广告模版对象集中的广告模版对象与所述第二特征信息是否匹配；
若当前的广告模版对象与所述第二特征信息匹配，则确定所述当前广告模版对象为所述目标广告模版对象。
可选地，所述广告模版对象设置有场景标识信息，所述场景标识信息包括一个或多个场景标签；所述第一特征信息包括一个或多个特征标签；所述依据所述第一特征信息和第二特征信息，从所述多个广告模板对象中，筛选出目标广告模版对象的步骤，包括：
确定与所述第一特征信息和所述第二特征信息均匹配的多个待筛选广告模版对象；
计算待筛选广告模版对象对应的场景标签与所述特征标签的匹配数量，生成第二匹配值；
确定所述第二匹配值中最大值对应的广告模版对象为所述目标广告模版对象。
可选地，所述物料信息包括历史曝光量和预计点击率；所述确定与所述目标广告模版对象对应的目标物料信息的步骤，包括：
确定与所述目标广告模版对象对应的多个待筛选物料信息；
根据所述历史曝光量和预置曝光量阈值，生成与所述待筛选物料信息对应的优先标识信息；
依据所述优先标识信息和所述预计点击率，确定所述目标物料信息。
可选地，所述预置曝光量阈值包括第一阈值，以及小于所述第一阈值的第二阈值；所述优先标识信息包括第一优先标识、第二优先标识和第三优先标识中的一个；所述根据所述历史曝光量和预置曝光量阈值，生成与所述待筛选物料信息对应的优先标识信息的步骤，包括：
当所述待筛选物料信息的历史曝光量大于所述第一阈值时，生成所述第一优先标识；
当所述待筛选物料信息的历史曝光量小于所述第一阈值且大于所述第二阈值时，生成所述第二优先标识；
当所述待筛选物料信息的历史曝光量小于所述第二阈值时，生成所述第三优先标识。
可选地，所述依据所述优先标识信息和所述预计点击率，确定所述目标物料信息的步骤，包括：
当只存在与所述第一优先标识对应的待筛选物料信息时，确定与所述预计点击率最大值对应的物料信息为所述目标物料信息；
当不存在与所述第一优先标识对应的待筛选物料信息时，随机选择待筛选物料信息中的一个为所述目标物料信息；
当同时存在与第一优先标识、第二优先标识、第三优先标识对应的筛选物料信息时，采用预置概率，随机选择待筛选物料信息中的一个为所述目标物料信息；其中所述预置概率包括与第一标识对应的第一概率、与第二标识对应的第二概率、与第三标识对应的第三概率，所述第一概率大于所述第二概率以及所述第三概率。
可选地，所述物料信息还包括物料元素；所述待筛选物料信息通过如下方法生成：
确定与所述广告模版对象匹配的候选物料信息；
采用预置的卷积神经网络，对所述候选物料信息的物料元素进行评分，生成质量评分值；
采用从大到小的方式，对所述质量评分值进行排序，生成质量排序信息；
按照所述质量排序信息，选择第二预置个数的候选物料信息为所述待筛选物料信息。
可选地，所述预计点击率通过如下方法生成：
获取模型输入数据；所述输入数据包括：所述第一特征信息、所述场景标识信息和所述物料元素；
将所述模型输入数据发送至预置的逻辑回归模型，生成所述预计点击率。
可选地，所述方法还包括：
按照从大到小的方式，对所述预计点击率进行排序，生成预计点击排序信息；
采用所述预计点击排序信息，对所述广告数据进行排序；
展示排序后的广告数据。
可选地，所述将所述目标物料信息填充至所述目标广告模版对象，生成广告数据的步骤，包括：
将目标物料信息中的物料元素填入所述目标广告模板对象；
为所述目标广告模板对象中的每一个物料元素配置对应的展示格式。
第二方面，本公开的实施例提供了一种广告数据生成系统，应用于交易平台，所述交易平台涉及多个用户对象、多个商户对象、多个广告模版对象，以及所述广告模版对象具有对应的多个物料信息，包括：
采集模块，用于采集所述用户对象的第一特征信息，以及，所述商户对象的第二特征信息；
模板排序模块，用于依据所述第一特征信息和第二特征信息，从所述多个广告模板对象中，筛选出目标广告模版对象；
物料决策模块，用于确定与所述目标广告模版对象对应的目标物料信息；
在线配置模块，用于将所述目标物料信息填充至所述目标广告模版对象，生成广告数据。
可选地，所述广告模版对象设置有场景标识信息；所述模板排序模块包括：
第一待筛选对象子模块，用于采用所述第一特征信息和所述场景标识信息，确定待筛选广告模版对象集；
第一目标对象确定子模块，用于采用所述第二特征信息，在所述待筛选广告模版对象集中确定所述目标广告模版对象。
可选地，所述第一特征信息包括一个或多个特征标签；所述第一待筛选对象子模块包括：
场景标识单元，用于确定与所述第一特征信息匹配的多个待筛选场景标识信息；
第一匹配单元，用于计算所述待筛选场景标识信息中的场景标签与所述特征标签的匹配数量，生成第一匹配值；
第一排序单元，用于采用从大到小的方式，对所述第一匹配值进行排序，生成匹配值排序信息；
待筛选对象单元，用于按照所述匹配值排序信息，选择第一预置个数的广告模版对象为所述待筛选广告模版对象集。
可选地，所述第一目标对象确定子模块包括：
第二特征匹配单元，用于依次判断所述待筛选广告模版对象集中的广告模版对象与所述第二特征信息是否匹配；
目标对象确定单元，用于若当前的广告模版对象与所述第二特征信息匹配，则确定所述当前广告模版对象为所述目标广告模版对象。
可选地，所述广告模版对象设置有场景标识信息，所述场景标识信息包括一个或多个场景标签；所述第一特征信息包括一个或多个特征标签；所述模板排序模块包括：
第二待筛选对象中子模块，用于确定与所述第一特征信息和所述第二特征信息均匹配的多个待筛选广告模版对象；
第二匹配子模块，用于计算待筛选广告模版对象对应的场景标签与所述特征标签的匹配数量，生成第二匹配值；
第二目标对象确定子模块，用于确定所述第二匹配值中最大值对应的广告模版对象为所述目标广告模版对象。
可选地，所述物料信息包括历史曝光量和预计点击率；所述物料决策模块包括：
待筛选物料子模块，用于确定与所述目标广告模版对象对应的多个待筛选物料信息；
标识生成模块，用于根据所述历史曝光量和预置曝光量阈值，生成与所述待筛选物料信息对应的优先标识信息；
目标物料子模块，用于依据所述优先标识信息和所述预计点击率，确定所述目标物料信息。
可选地，所述预置曝光量阈值包括第一阈值，以及小于所述第一阈值的第二阈值；所述优先标识信息包括第一优先标识、第二优先标识和第三优先标识中的一个；所述标识生成模块包括：
第一优先标识单元，用于当所述待筛选物料信息的历史曝光量大于所述第一阈值时，生成所述第一优先标识；
第二优先标识单元，用于当所述待筛选物料信息的历史曝光量小于所述第一阈值且大于所述第二阈值时，生成所述第二优先标识；
第三优先标识单元，用于当所述待筛选物料信息的历史曝光量小于所述第二阈值时，生成所述第三优先标识。
可选地，所述目标物料子模块包括：
第一物料选择单元，用于当只存在与所述第一优先标识对应的待筛选物料信息时，确定与所述预计点击率最大值对应的物料信息为所述目标物料信息；
第二物料选择单元，用于当不存在与所述第一优先标识对应的待筛选物料信息时，随机选择待筛选物料信息中的一个为所述目标物料信息；
第三物料选择单元，用于当同时存在与第一优先标识、第二优先标识、第三优先标识对应的筛选物料信息时，采用预置概率，随机选择待筛选物料信息中的一个为所述目标物料信息；其中所述预置概率包括与第一标识对应的第一概率、与第二标识对应的第二概率、与第三标识对应的第三概率，所述第一概率大于所述第二概率以及所述第三概 率。
可选地，所述物料信息还包括物料元素；所述待筛选物料信息通过如下模块生成：
候选物料模块，用于确定与所述广告模版对象匹配的候选物料信息；
质量评分模块，用于采用预置的卷积神经网络，对所述候选物料信息的物料元素进行评分，生成质量评分值；
质量排序模块，用于采用从大到小的方式，对所述质量评分值进行排序，生成质量排序信息；
物料选取模块，用于按照所述质量排序信息，选择第二预置个数的候选物料信息为所述待筛选物料信息。
可选地，所述预计点击率通过如下模块生成：
模型输入模块，用于获取模型输入数据；所述输入数据包括：所述第一特征信息、所述场景标识信息和所述物料元素；
点击率生成模块，用于将所述模型输入数据发送至预置的逻辑回归模型，生成所述预计点击率。
可选地，所述系统还包括：
点击率排序模块，用于按照从大到小的方式，对所述预计点击率进行排序，生成预计点击排序信息；
广告数据排序模块，用于采用所述预计点击排序信息，对所述广告数据进行排序；
展示模块，用于展示排序后的广告数据。
可选地，所述在线配置模块包括：
物料内容组装配置子模块，用于将目标物料信息中的物料元素填入所述目标广告模板对象；
物料样式组装配置子模块，用于为所述目标广告模板对象中的每一个物料元素配置对应的展示格式。
第三方面，本公开的实施例提供了一种电子设备，包括：
处理器、存储器以及存储在所述存储器上并可在所述处理器上运行的计算机程序，所述处理器执行所述计算机程序时实现前述广告数据生成方法。
第四方面，本公开的实施例提供了一种计算机程序，包括计算机可读代码，当所述计算机可读代码在计算处理设备上运行时，导致所述计算处理设备执行前述广告数据生成方法。
第五方面，本公开的实施例提供了一种计算机可读存储介质，其中存储了前述计算机程序。
本公开的实施例提供了一种广告数据生成方法，通过采集交易平台的用户对象的第一特征信息，以及商户对象的第二特征信息，根据第一特征信息和第二特征信息，在多个预置广告模版对象中筛选出目标广告目标对象，提取与目标广告模版对象对应的目标物料信息，并将目标物料信息填充至目标广告模版对象，从而实现可以根据不同的用户对象的第一特征信息，生成与用户对象匹配的广告数据，提高广告数据与用户对象的适配性。
进一步的，为不同的用户对象生成符合其特征的广告数据，能够促进用户对象与商 户对象的交易成功率。
上述说明仅是本公开技术方案的概述，为了能够更清楚了解本公开的技术手段，而可依照说明书的内容予以实施，并且为了让本公开的上述和其它目的、特征和优点能够更明显易懂，以下特举本公开的具体实施方式。
附图说明
为了更清楚地说明本公开实施例的技术方案，下面将对本公开的实施例的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公开的一些实施例，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示出了本公开的一种广告数据生成方法实施例一的步骤流程图；
图2示出了本公开的一种广告数据生成方法实施例二的步骤流程图；
图3示出了本公开的一种广告数据生成系统实施例的结构框图；
图4示意性地示出了用于执行根据本公开的方法的电子设备的框图；以及
图5示意性地示出了用于保持或者携带实现根据本公开的方法的程序代码的存储单元。
具体实施例
为使本公开的实施例的目的、技术方案和优点更加清楚，下面将结合本公开的实施例中的附图，对本公开的实施例中的技术方案进行清楚、完整地描述，显然，所描述的实施例是本公开的一部分实施例，而不是全部的实施例。基于本公开中的实施例，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，都属于本公开保护的范围。
实施例一
参照图1，其示出了本公开的一种广告数据生成方法实施例一的步骤流程图，所述方法可以应用于交易平台，所述交易平台涉及多个用户对象、多个商户对象、多个广告模版对象，以及所述广告模版对象具有对应的多个物料信息，包括：
广告模版对象可以是按照设置的划分区域，展示物料信息的模版。
步骤101，采集所述用户对象的第一特征信息，以及，所述商户对象的第二特征信息；
第一特征信息可以为买方用户一个或多个维度的特征，例如：当前位置、性别、职业等。
第二特征信息可以为卖方用户一个或多个维度的特征，例如：店址位置、经营类别等。
步骤102，依据所述第一特征信息和第二特征信息，从所述多个广告模板对象中，筛选出目标广告模版对象；
根据第一特征信息和第二特征信息的匹配程度，从所述多个广告模版对象中，筛选出与第一特征信息、第二特征信息匹配程度最高的广告模版对象，将该广告模版对象确 定为目标广告模版对象。
步骤103，确定与所述目标广告模版对象对应的目标物料信息；
不同的广告模版对象可能关联有不同的物料信息，在确定目标广告模版对象之后，可以提取与目标广告模版对象对应的物料信息为目标物料信息。
步骤104，将所述目标物料信息填充至所述目标广告模版对象，生成广告数据。
将提取到的目标物料信息填充至目标广告模版对象，使得目标物料信息按照一定的划分规则进行显示，生成广告数据。
在本公开的实施例中，通过采集交易平台的用户对象的第一特征信息，以及商户对象的第二特征信息，根据第一特征信息和第二特征信息，在多个预置广告模版对象中筛选出目标广告目标对象，提取与目标广告模版对象对应的目标物料信息，并将目标物料信息填充至目标广告模版对象，从而实现可以根据不同的用户对象的第一特征信息，生成与用户对象匹配的广告数据，提高广告数据与用户对象的适配性。
进一步的，为不同的用户对象生成符合其特征的广告数据，能够促进用户对象与商户对象的交易成功率。
需要说明的是，现有技术中一般是通过接入层、业务层、检索层、索引层和数据层实现广告数据的生成。其中，接入层用于接收前端SDK(Software Development Kit，软件开发工具包)的业务请求，生成请求改写消息，并向业务层请求广告内容；以及获取业务层返回的广告内容后，完成样式组装并发送给SDK。业务层用于在接收到请求改写信息后，依据产品(例如：应用平台)逻辑，全部指定一种广告模版对象；以及向检索层请求广告内容；获取检索层返回结果后，根据产品逻辑获取唯一的物料信息，将指定的广告模板对象和唯一的物料信息组装后，返回给接入层。检索层用于检索符合广告模板对象的商户对象和对应的物料信息。索引层用于将数据层数据建立索引，供检索层检索。数据层用于接入底层数据，包括广告投放数据与物料信息集。
参照图2，示出了本公开的一种广告数据生成方法实施例二的步骤流程图，所述方法可以应用于交易平台，所述交易平台涉及多个用户对象、多个商户对象、多个广告模版对象，以及所述广告模版对象具有对应的多个物料信息，本公开的实施例具体可以包括如下步骤：
交易平台可以通过互联网形式连接多个用户对象、多个商户对象，用户对象可以与商户对象在所述交易平台进行交易。例如：交易平台可以是团购网站、团购APP(application，应用程序)、外卖网站、外卖APP等。
用户对象可以是该交易平台的买方用户，商家对象可以是该交易平台的卖方用户(例如：虚拟商店)。
广告模版对象可以是按照设置的划分区域，展示物料信息的模版。
不同的广告模版对象可以具有对应的有一个或多个物料信息。例如：广告模版对象A对应物料a、物料b、物料c。模版B对应物料a、物料c、物料d。
步骤201，采集所述用户对象的第一特征信息，以及，所述商户对象的第二特征信息；
当用户对象为首次使用该交易平台时，第一特征信息可以由交易平台实时采集获得，也可以由用户对象自定义传输至交易平台获得。当用户对象为非首次使用该交易平台时，第一特征信息可以由交易平台实时采集获得，也可以由交易平台读取与该用户对象保存至服务器的历史信息确定。
第一特征信息可以为买方用户一个或多个维度的特征。第一特征信息可以包括例如：当前位置、性别、工作等。
第二特征信息可以为卖方用户一个或多个维度的特征，例如：店址位置、经营类别等。
步骤202，依据所述第一特征信息和第二特征信息，从所述多个广告模板对象中，筛选出目标广告模版对象；
根据第一特征信息和第二特征信息的匹配程度，从所述多个广告模版对象中，筛选出与第一特征信息、第二特征信息匹配程度最高的广告模版对象，将该广告模版对象确定为目标广告模版对象。
本公开的实施例可以在业务层选择广告模板对象时，通过执行步骤202，从而在多个广告模板对象中，确定与第一特征信息、第二特征信息匹配程度最高的目标广告模板对象，并依据该目标广告模板对象生成广告数据。
在本公开的一种优选实施例中，所述广告模版对象设置有场景标识信息；步骤202可以包括：
子步骤S11，采用所述第一特征信息和所述场景标识信息，确定待筛选广告模版对象集；
广告模版对象可以设置有场景标识信息，根据第一特征信息和场景标识信息，确定待筛选广告模版对象集。所述待筛选广告模版对象集可以由一个或多个广告模版对象组成，并且，待筛选广告模版对象集中的广告模版对象的场景标识信息与第一特征信息匹配。
在本公开的一种优选实施例中，所述第一特征信息包括一个或多个特征标签；子步骤S11可以包括：
子步骤S111，确定与所述第一特征信息匹配的多个待筛选场景标识信息；
子步骤S112，计算所述待筛选场景标识信息中的场景标签与所述特征标签的匹配数量，生成第一匹配值；
子步骤S113，采用从大到小的方式，对所述第一匹配值进行排序，生成匹配值排序信息；
子步骤S114，按照所述匹配值排序信息，选择第一预置个数的广告模版对象为所述待筛选广告模版对象集。
特征标签包括平台标识信息(例如：平台名称)、渠道信息(例如：APP、PC(Personal Computer，个人计算机))、职业信息、性别信息、指定位置信息、意向交易类别、下单偏好信息(例如：折扣偏好、新品偏好)中的一个或多个。
场景标识信息中的场景标签可以包括平台标识信息、渠道信息，以及性别信息、职业信息、经营分类信息中的至少一个。
交易平台可以设置有多个广告模版对象，第一特征信息可能与多个广告模版对象的场景标识信息存在不同匹配程度，具体地，是第一特征信息中的特征标签与场景标识信息中的场景标签的匹配数量不同。计算待筛选场景信息中的场景标签与特征标签的匹配数量，生成第一匹配值。针对生成的第一匹配值，按照从大到小的顺序进行排序，生成匹配值排序信息。从匹配值排序信息的第一个起，按顺序将第一预置数量的广告模版对象确定为待筛选广告模版集(例如，交易平台可以预先设置有200个广告模版对象，第一预置数量可以设置为50，则待筛选广告模版集包括50个待筛选的广告模版对象)。
可以理解的是，场景标识信息对应的匹配值越大，则表示第一特征信息与该场景标识信息的匹配程度越高，该场景标识信息越符合对应的用户对象。
例如：第一特征信息可以包括如下特征标签：APP、男性、白领、餐饮。场景标识信息A的场景标签为：PC、女性；场景标识信息B的场景标签为：PC、男性、购物；场景标识信息C的场景标签为：APP、男性、餐饮。则第一特征信息与上述三个场景标识信息对应的第一匹配值分别为0、1、3，场景标识信息C与第一特征信息的匹配程度最高。
其中，可以设定为场景标识信息中的平台标识信息、渠道信息必须与第一特征信息匹配，否则，直接将其对应匹配值设定为0，从而保证待筛选广告模版对象符合第一特征信息所指示的平台标识信息和渠道信息。
子步骤S12，采用所述第二特征信息，在所述待筛选广告模版对象集中确定所述目标广告模版对象。
不同的广告模版对象可能适用于不同的商户对象(例如：广告模版对象A适用于商户对象A、商户对象B；广告模版对象B适用于商户对象A、商户对象C)，可以采用第二特征信息，确定在所述待筛选广告模版对象集中，适用于商户对象的目标广告模版对象。
在本公开的一种优选实施例中，子步骤S12可以包括：
子步骤S121，依次判断所述待筛选广告模版对象集中的广告模版对象与所述第二特 征信息是否匹配；
子步骤S122，若当前的广告模版对象与所述第二特征信息匹配，则确定所述当前广告模版对象为所述目标广告模版对象。
按照所述匹配值排序信息，依次判断所述待筛选广告模版对象集中的广告模版对象与所述第二特征信息是否匹配，若当前的广告模版对象与所述第二特征信息匹配，表示当前广告模版对象适用于该商户对象，则确定所述当前广告模版对象为所述目标广告模版对象。
具体地，第二特征信息可以包括但不限于经营分类信息、平台标识信息、渠道信息。当广告模版对象的场景标识信息与第二特征信息匹配时，则确定当前广告模版对象适用于该商户对象。
本优选发明实施例可以通过在采集到第一特征信息和第二特征信息后，根据第一特征信息和在先离线生成的场景标签，计算与各个广告模板对象对应的匹配值，根据匹配值实时地确定待筛选广告模板对象集，并实时地根据第二特征信息与待筛选广告模板对象的匹配程度，将待筛选广告模板对象集进行排序，从而实时地确定出当前用户对象匹配程度依次排序的多个广告模板对象。
在本公开的另一种优选实施例中，所述广告模版对象设置有场景标识信息，所述场景标识信息包括一个或多个场景标签；所述第一特征信息包括一个或多个特征标签；步骤202除了可以通过上述子步骤S11-S12实现，还可以通过如下步骤实现：
子步骤S21，确定与所述第一特征信息和所述第二特征信息均匹配的多个待筛选广告模版对象；
子步骤S22，计算待筛选广告模版对象对应的场景标签与所述特征标签的匹配数量，生成第二匹配值；
子步骤S23，确定所述第二匹配值中最大值对应的广告模版对象为所述目标广告模版对象。
将与第一特征信息、第二特征信息均匹配的场景标识信息对应的广告模版对象确定为待筛选广告模版对象，然后一次计算待筛选广告模版对象对应的场景标签与特征标签的匹配数量，生成第二匹配值，确定与第二匹配值中最大值对应的广告模版对象为目标广告模版对象。其中，可以采用LR(Logistic Regression，逻辑回归)模型进行第一特征信息、第二特征信息、场景标识信息的匹配，以及第二匹配值的计算。
步骤203，确定与所述目标模版对象对应的多个待筛选物料信息；
将与目标广告模版对象对应的物料信息确定为待筛选物料信息，其中，所述物料信息包括历史曝光量和预计点击率。历史曝光量为该物料信息在当前时刻之前，被用户对象浏览达到一定时间的次数。所述预计点击率可以采用预置的模型，结合第一特征信息、 场景标识信息生成。
步骤204，根据所述历史曝光量和预置曝光量阈值，生成与所述待筛选物料信息对应的优先标识信息；
将历史曝光量与曝光量阈值进行大小对比，生成相应的优先标识信息。
在本公开的一种优选实施例中，所述预置曝光量阈值包括第一阈值，以及小于所述第一阈值的第二阈值；所述优先标识信息包括第一优先标识、第二优先标识和第三优先标识中的一个；所述步骤204可以包括：
子步骤S31，当所述待筛选物料信息的历史曝光量大于所述第一阈值时，生成所述第一优先标识；
子步骤S32，当所述待筛选物料信息的历史曝光量小于所述第一阈值且大于所述第二阈值时，生成所述第二优先标识；
子步骤S33，当所述待筛选物料信息的历史曝光量小于所述第二阈值时，生成所述第三优先标识。
将待筛选物料信息的历史曝光量与第一阈值、第二阈值进行对比，确定当前物料信息对应第一优先标识、第二优先标识和第三优先标识中的一个。其中，第一优先标识、第二优先标识、第三优先标识对应的优先级依次降低。
步骤205，依据所述优先标识信息和所述预计点击率，确定所述目标物料信息。
在生成优先标识信息之后，结合预计点击率，确定目标物料信息。
本公开的实施例可以通过离线方式，根据曝光量和预计点击率对待筛选物料信息生成对应的优先标识，优先标识用以为对应的待筛选物料信息分配不同的被选中为目标物料信息的概率，并且可以在展示广告数据后，根据用户对象对广告数据的选择情况，更新预计点击率。
本公开的实施例可以通过检索层在选择物料信息时，通过执行步骤203-205，从而根据历史曝光率、曝光率预置和预计点击率，在多个待筛选物料信息中，确定出目标物料信息。
在本公开的一种优选实施例中，步骤205可以包括：
子步骤S41，当只存在与所述第一优先标识对应的待筛选物料信息时，确定与所述预计点击率最大值对应的物料信息为所述目标物料信息；
子步骤S42，当不存在与所述第一优先标识对应的待筛选物料信息时，随机选择待筛选物料信息中的一个为所述目标物料信息；
当不存在与所述第一优先标识对应的待筛选物料信息时，可以为每个待筛选物料信息分配一个编号，然后实时生成一个在编号取值范围内的随机数，将与其随机数对应的待筛选物料信息为目标物料信息。
子步骤S43，当同时存在与第一优先标识、第二优先标识、第三优先标识对应的筛选物料信息时，采用预置概率，随机选择待筛选物料信息中的一个为所述目标物料信息； 其中所述预置概率包括与第一标识对应的第一概率、与第二标识对应的第二概率、与第三标识对应的第三概率，所述第一概率大于所述第二概率以及所述第三概率。
第一概率、第二概率、所述第三概率分别表示对应第一标识被选中的概率，具体地，可以为每一个优先标识信息分别进行编号，然后实时产生一个随机数，所述随机数任一值对应且只对应一个优先标识信息，通过该随机数确定被选中的优先标识信息对应的物料信息为目标物料信息。
步骤206，将所述目标物料信息填充至所述目标广告模版对象，生成广告数据。
通过实时在线配置的方式，将提取到的目标物料信息填充至目标广告模版对象，使得目标物料信息按照一定的划分规则进行显示，生成广告数据。
在一种优选实施例中，步骤206可以包括：将目标物料信息中的物料元素填入所述目标广告模板对象；为所述目标广告模板对象中的每一个物料元素配置对应的展示格式。
其中，展示格式可以包括但不限于文字规格、图片规格等。
本公开的实施例可以通过业务层执行将目标物料信息中的物料元素填入所述目标广告模板对象的步骤，从而实现目标物料信息和目标广告模板数据的实时结合；通过接入层执行为所述目标广告模板对象中的每一个物料元素配置对应的展示格式的步骤，从而实现实时配置物料元素样式，增加广告数据的美观性。
步骤207，按照从大到小的方式，对所述预计点击率进行排序，生成预计点击排序信息；
步骤208，采用所述预计点击排序信息，对所述广告数据进行排序；
步骤209，展示排序后的广告数据。
按照预计点击顺序信息，将生成的广告数据进行排序，当用户对象登录所述交易平台时，向所述用户对象展示排序后的广告数据。
在本公开的一种优选实施例中，所述物料信息还包括物料元素；所述待筛选物料信息通过如下方法生成：确定与所述广告模版对象匹配的候选物料信息；采用预置的卷积神经网络，对所述候选物料信息的物料元素进行评分，生成质量评分值；采用从大到小的方式，对所述质量评分值进行排序，生成质量排序信息；按照所述质量排序信息，选择第二预置个数的候选物料信息为所述待筛选物料信息。
所述待筛选物料信息可以通过离线方式生成，即非实时生成，并将生成的待筛选物料信息存储至数据层中。
可以采用神经卷积网络(CNN，Convolutional Neural Network)，将物料信息中的物料元素作为卷积网络的输入数据，为每一个物料元素进行评分，生成对应的质量评分值，依据质量评分值从大到小的顺序，从质量评分值的最大值开始，选择个数为第二预置个数的物料信息为待筛选物料信息。
在实际应用中，除了可以按照质量排序信息确定待筛选物料信息以外，还可以将质量评分值区分成多个评分区间，为每一个评分区间设定一个被中选概率，通过产生第二 预置个数的随机数确定从多个评分区间中，依次选择出选择出待选区间，再从待选区间中随机选择一个物料信息为待筛选物料信息，其中，依据次序不同，待选区间可以重复出现被选择。
在本公开的一种优选实施例中，所述预计点击率通过如下方法生成：
获取模型输入数据；所述输入数据包括：所述第一特征信息、所述场景信息和所述物料元素；
将所述模型输入数据发送至预置的逻辑回归模型，生成所述预计点击率。
在现有技术的基础上，增加所述第一特征信息、所述场景信息、所述物料元素中的至少一个维度确定预计点击率，使得预计点击率更加准确可靠。
在本公开的实施例中，通过采集交易平台的用户对象的第一特征信息，以及商户对象的第二特征信息，根据第一特征信息与广告模版对象中的场景信息进行匹配，获取对应匹配程度最高的预置个数的广告模版对象为待筛选广告模版对象集，然后充待筛选广告模版对象集中，筛选出与第二特征信息的广告模版对象为目标广告目标对象。生成与广告模版目标广告模版对象中每一个物料信息对应的优先标识信息，依据优先标识信息以及预置点击率，确定与目标广告模版对象对应的目标物料信息，并将目标物料信息填充至目标广告模版对象，从而实现可以根据不同的用户对象的第一特征信息，生成与用户对象匹配的广告数据，提高广告数据与用户对象的适配性，由于添加了第一特征信息、所述场景信息、所述物料元素等维度进行预计点击率的计算，使得预计点击率更加准确。
进一步的，为不同的用户对象生成符合其特征的广告数据，能够促进用户对象与商户对象的交易成功率。
参照图3，示出了本公开的一种广告数据生成系统实施例的结构框图，所述系统可以应用于交易平台，所述交易平台涉及多个用户对象、多个商户对象、多个广告模版对象，以及所述广告模版对象具有对应的多个物料信息，所述系统具体可以包括如下模块：
采集模块301，用于采集所述用户对象的第一特征信息，以及，所述商户对象的第二特征信息；
模板排序模块302，用于依据所述第一特征信息和第二特征信息，从所述多个广告模板对象中，筛选出目标广告模版对象；
物料决策模块303，用于确定与所述目标广告模版对象对应的目标物料信息；
在线配置模块304，用于将所述目标物料信息填充至所述目标广告模版对象，生成广告数据。
在本公开的一种优选实施例中，所述广告模版对象设置有场景标识信息；所述模板排序模块302包括：
第一待筛选对象子模块，用于采用所述第一特征信息和所述场景标识信息，确定待 筛选广告模版对象集；
第一目标对象确定子模块，用于采用所述第二特征信息，在所述待筛选广告模版对象集中确定所述目标广告模版对象。
在本公开的一种优选实施例中，所述第一特征信息包括一个或多个特征标签；所述第一待筛选对象子模块包括：
场景标识单元，用于确定与所述第一特征信息匹配的多个待筛选场景标识信息；
第一匹配单元，用于计算所述待筛选场景标识信息中的场景标签与所述特征标签的匹配数量，生成第一匹配值；
第一排序单元，用于采用从大到小的方式，对所述第一匹配值进行排序，生成匹配值排序信息；
待筛选对象单元，用于按照所述匹配值排序信息，选择第一预置个数的广告模版对象为所述待筛选广告模版对象集。
在本公开的一种优选实施例中，所述第一目标对象确定子模块包括：
第二特征匹配单元，用于依次判断所述待筛选广告模版对象集中的广告模版对象与所述第二特征信息是否匹配；
目标对象确定单元，用于若当前的广告模版对象与所述第二特征信息匹配，则确定所述当前广告模版对象为所述目标广告模版对象。
在本公开的一种优选实施例中，所述广告模版对象设置有场景标识信息，所述场景标识信息包括一个或多个场景标签；所述第一特征信息包括一个或多个特征标签；所述模板排序模块302包括：
第二待筛选对象中子模块，用于确定与所述第一特征信息和所述第二特征信息均匹配的多个待筛选广告模版对象；
第二匹配子模块，用于计算待筛选广告模版对象对应的场景标签与所述特征标签的匹配数量，生成第二匹配值；
第二目标对象确定子模块，用于确定所述第二匹配值中最大值对应的广告模版对象为所述目标广告模版对象。
在本公开的一种优选实施例中，所述物料信息包括历史曝光量和预计点击率；所述物料决策模块303包括：
待筛选物料子模块，用于确定与所述目标广告模版对象对应的多个待筛选物料信息；
标识生成模块，用于根据所述历史曝光量和预置曝光量阈值，生成与所述待筛选物料信息对应的优先标识信息；
目标物料子模块，用于依据所述优先标识信息和所述预计点击率，确定所述目标物料信息。
在本公开的一种优选实施例中，所述预置曝光量阈值包括第一阈值，以及小于所述第一阈值的第二阈值；所述优先标识信息包括第一优先标识、第二优先标识和第三优先 标识中的一个；所述标识生成模块包括：
第一优先标识单元，用于当所述待筛选物料信息的历史曝光量大于所述第一阈值时，生成所述第一优先标识；
第二优先标识单元，用于当所述待筛选物料信息的历史曝光量小于所述第一阈值且大于所述第二阈值时，生成所述第二优先标识；
第三优先标识单元，用于当所述待筛选物料信息的历史曝光量小于所述第二阈值时，生成所述第三优先标识。
在本公开的一种优选实施例中，所述目标物料子模块包括：
第一物料选择单元，用于当只存在与所述第一优先标识对应的待筛选物料信息时，确定与所述预计点击率最大值对应的物料信息为所述目标物料信息；
第二物料选择单元，用于当不存在与所述第一优先标识对应的待筛选物料信息时，随机选择待筛选物料信息中的一个为所述目标物料信息；
第三物料选择单元，用于当同时存在与第一优先标识、第二优先标识、第三优先标识对应的筛选物料信息时，采用预置概率，随机选择待筛选物料信息中的一个为所述目标物料信息；其中所述预置概率包括与第一标识对应的第一概率、与第二标识对应的第二概率、与第三标识对应的第三概率，所述第一概率大于所述第二概率以及所述第三概率。
在本公开的一种优选实施例中，所述物料信息还包括物料元素；所述待筛选物料信息通过如下模块生成：
候选物料模块，用于确定与所述广告模版对象匹配的候选物料信息；
质量评分模块，用于采用预置的卷积神经网络，对所述候选物料信息的物料元素进行评分，生成质量评分值；
质量排序模块，用于采用从大到小的方式，对所述质量评分值进行排序，生成质量排序信息；
物料选取模块，用于按照所述质量排序信息，选择第二预置个数的候选物料信息为所述待筛选物料信息。
在本公开的一种优选实施例中，所述预计点击率通过如下模块生成：
模型输入模块，用于获取模型输入数据；所述输入数据包括：所述第一特征信息、所述场景标识信息和所述物料元素；
点击率生成模块，用于将所述模型输入数据发送至预置的逻辑回归模型，生成所述预计点击率。
在本公开的一种优选实施例中，所述系统还包括：
点击率排序模块，用于按照从大到小的方式，对所述预计点击率进行排序，生成预计点击排序信息；
广告数据排序模块，用于采用所述预计点击排序信息，对所述广告数据进行排序；
展示模块，用于展示排序后的广告数据。
在本公开的一种优选实施例中，所述在线配置模块304包括：
物料内容组装配置子模块，用于将目标物料信息中的物料元素填入所述目标广告模板对象；
物料样式组装配置子模块，用于为所述目标广告模板对象中的每一个物料元素配置对应的展示格式。
对于装置实施例而言，由于其与方法实施例基本相似，所以描述的比较简单，相关之处参见方法实施例的部分说明即可。
本公开的实施例还提供了一种电子设备，包括：处理器、存储器以及存储在所述存储器上并可在所述处理器上运行的计算机程序，所述处理器执行所述计算机程序时实现前述实施例的广告数据生成方法。
本公开的实施例还提供了一种计算机程序，包括计算机可读代码，当所述计算机可读代码在计算处理设备上运行时，导致所述计算处理设备执行前述实施例的广告数据生成方法。
本公开的实施例还提供了一种计算机可读存储介质，其中存储了前述实施例的计算机程序。
以上所描述的装置实施例仅仅是示意性的，其中所述作为分离部件说明的单元可以是或者也可以不是物理上分开的，作为单元显示的部件可以是或者也可以不是物理单元，即可以位于一个地方，或者也可以分布到多个网络单元上。可以根据实际的需要选择其中的部分或者全部模块来实现本实施例方案的目的。本领域普通技术人员在不付出创造性的劳动的情况下，即可以理解并实施。
本公开的各个部件实施例可以以硬件实现，或者以在一个或者多个处理器上运行的软件模块实现，或者以它们的组合实现。本领域的技术人员应当理解，可以在实践中使用微处理器或者数字信号处理器(DSP)来实现根据本公开实施例的计算处理设备中的一些或者全部部件的一些或者全部功能。本公开还可以实现为用于执行这里所描述的方法的一部分或者全部的设备或者装置程序(例如，计算机程序和计算机程序产品)。这样的实现本公开的程序可以存储在计算机可读存储介质上，或者可以具有一个或者多个信号的形式。这样的信号可以从因特网网站上下载得到，或者在载体信号上提供，或者以任何其他形式提供。
例如，图4示出了可以实现根据本公开的方法的计算处理设备。该计算处理设备传统上包括处理器410和以存储器420形式的计算机程序产品或者计算机可读存储介质。存储器420可以是诸如闪存、EEPROM(电可擦除可编程只读存储器)、EPROM、硬盘或者ROM之类的电子存储器。存储器420具有用于执行上述方法中的任何方法步骤的程序代码431的存储空间430。例如，用于程序代码的存储空间430可以包括分别用于实现上面的 方法中的各种步骤的各个程序代码431。这些程序代码可以从一个或者多个计算机程序产品中读出或者写入到这一个或者多个计算机程序产品中。这些计算机程序产品包括诸如硬盘，紧致盘(CD)、存储卡或者软盘之类的程序代码载体。这样的计算机程序产品通常为如参考图5所述的便携式或者固定存储单元。该存储单元可以具有与图4的计算处理设备中的存储器420类似布置的存储段、存储空间等。程序代码可以例如以适当形式进行压缩。通常，存储单元包括计算机可读代码431’，即可以由例如诸如410之类的处理器读取的代码，这些代码当由计算处理设备运行时，导致该计算处理设备执行上面所描述的方法中的各个步骤。
本文中所称的“一个实施例”、“实施例”或者“一个或者多个实施例”意味着，结合实施例描述的特定特征、结构或者特性包括在本公开的至少一个实施例中。此外，请注意，这里“在一个实施例中”的词语例子不一定全指同一个实施例。
在此处所提供的说明书中，说明了大量具体细节。然而，能够理解，本公开的实施例可以在没有这些具体细节的情况下被实践。在一些实例中，并未详细示出公知的方法、结构和技术，以便不模糊对本说明书的理解。
在权利要求中，不应将位于括号之间的任何参考符号构造成对权利要求的限制。单词“包含”不排除存在未列在权利要求中的元件或步骤。位于元件之前的单词“一”或“一个”不排除存在多个这样的元件。本公开可以借助于包括有若干不同元件的硬件以及借助于适当编程的计算机来实现。在列举了若干装置的单元权利要求中，这些装置中的若干个可以是通过同一个硬件项来具体体现。单词第一、第二、以及第三等的使用不表示任何顺序。可将这些单词解释为名称。
最后应说明的是：以上实施例仅用以说明本公开的技术方案，而非对其限制；尽管参照前述实施例对本公开进行了详细的说明，本领域的普通技术人员应当理解：其依然可以对前述各实施例所记载的技术方案进行修改，或者对其中部分技术特征进行等同替换；而这些修改或者替换，并不使相应技术方案的本质脱离本公开各实施例技术方案的精神和范围。

Claims (16)

  1. 一种广告数据生成方法，应用于交易平台，所述交易平台涉及多个用户对象、多个商户对象、多个广告模版对象，以及所述广告模版对象具有对应的多个物料信息，包括：
    采集所述用户对象的第一特征信息，以及，所述商户对象的第二特征信息；
    依据所述第一特征信息和第二特征信息，从所述多个广告模板对象中，筛选出目标广告模版对象；
    确定与所述目标广告模版对象对应的目标物料信息；
    将所述目标物料信息填充至所述目标广告模版对象，生成广告数据。
  2. 根据权利要求1所述的方法，所述广告模版对象设置有场景标识信息；所述依据所述第一特征信息和第二特征信息，从所述多个广告模板对象中，筛选出目标广告模版对象的步骤，包括：
    采用所述第一特征信息和所述场景标识信息，确定待筛选广告模版对象集；
    采用所述第二特征信息，在所述待筛选广告模版对象集中确定所述目标广告模版对象。
  3. 根据权利要求2所述的方法，所述第一特征信息包括一个或多个特征标签；所述采用所述第一特征信息和所述场景标识信息，确定待筛选广告模版对象集的步骤，包括：
    确定与所述第一特征信息匹配的多个待筛选场景标识信息；
    计算所述待筛选场景标识信息中的场景标签与所述特征标签的匹配数量，生成第一匹配值；
    采用从大到小的方式，对所述第一匹配值进行排序，生成匹配值排序信息；
    按照所述匹配值排序信息，选择第一预置个数的广告模版对象为所述待筛选广告模版对象集。
  4. 根据权利要求3所述的方法，所述采用所述第二特征信息，在所述待筛选广告模版对象集中确定所述目标广告模版对象的步骤，包括：
    依次判断所述待筛选广告模版对象集中的广告模版对象与所述第二特征信息是否匹配；
    若当前的广告模版对象与所述第二特征信息匹配，则确定所述当前广告模版对象为所述目标广告模版对象。
  5. 根据权利要求1所述的方法，所述广告模版对象设置有场景标识信息，所述场景标识信息包括一个或多个场景标签；所述第一特征信息包括一个或多个特征标签；所述依据所述第一特征信息和第二特征信息，从所述多个广告模板对象中，筛选出目标广告模版对象的步骤，包括：
    确定与所述第一特征信息和所述第二特征信息均匹配的多个待筛选广告模版对象；
    计算待筛选广告模版对象对应的场景标签与所述特征标签的匹配数量，生成第二匹配值；
    确定所述第二匹配值中最大值对应的广告模版对象为所述目标广告模版对象。
  6. 根据权利要求1所述的方法，所述物料信息包括历史曝光量和预计点击率；所述确定与所述目标广告模版对象对应的目标物料信息的步骤，包括：
    确定与所述目标广告模版对象对应的多个待筛选物料信息；
    根据所述历史曝光量和预置曝光量阈值，生成与所述待筛选物料信息对应的优先标识信息；
    依据所述优先标识信息和所述预计点击率，确定所述目标物料信息。
  7. 根据权利要求6所述的方法，所述预置曝光量阈值包括第一阈值，以及小于所述第一阈值的第二阈值；所述优先标识信息包括第一优先标识、第二优先标识和第三优先标识中的一个；所述根据所述历史曝光量和预置曝光量阈值，生成与所述待筛选物料信息对应的优先标识信息的步骤，包括：
    当所述待筛选物料信息的历史曝光量大于所述第一阈值时，生成所述第一优先标识；
    当所述待筛选物料信息的历史曝光量小于所述第一阈值且大于所述第二阈值时，生成所述第二优先标识；
    当所述待筛选物料信息的历史曝光量小于所述第二阈值时，生成所述第三优先标识。
  8. 根据权利要求7所述的方法，所述依据所述优先标识信息和所述预计点击率，确定所述目标物料信息的步骤，包括：
    当只存在与所述第一优先标识对应的待筛选物料信息时，确定与所述预计点击率最大值对应的物料信息为所述目标物料信息；
    当不存在与所述第一优先标识对应的待筛选物料信息时，随机选择待筛选物料信息中的一个为所述目标物料信息；
    当同时存在与第一优先标识、第二优先标识、第三优先标识对应的筛选物料信息时，采用预置概率，随机选择待筛选物料信息中的一个为所述目标物料信息；其中所述预置概率包括与第一标识对应的第一概率、与第二标识对应的第二概率、与第三标识对应的第三概率，所述第一概率大于所述第二概率以及所述第三概率。
  9. 根据权利要求6所述的方法，所述物料信息还包括物料元素；所述待筛选物料信息通过如下方法生成：
    确定与所述广告模版对象匹配的候选物料信息；
    采用预置的卷积神经网络，对所述候选物料信息的物料元素进行评分，生成质量评分值；
    采用从大到小的方式，对所述质量评分值进行排序，生成质量排序信息；
    按照所述质量排序信息，选择第二预置个数的候选物料信息为所述待筛选物料信息。
  10. 根据权利要求9所述的方法，所述预计点击率通过如下方法生成：
    获取模型输入数据；所述输入数据包括：所述第一特征信息、所述场景标识信息和所述物料元素；
    将所述模型输入数据发送至预置的逻辑回归模型，生成所述预计点击率。
  11. 根据权利要求6所述的方法，还包括：
    按照从大到小的方式，对所述预计点击率进行排序，生成预计点击排序信息；
    采用所述预计点击排序信息，对所述广告数据进行排序；
    展示排序后的广告数据。
  12. 根据权利要求9或10或11所述的方法，所述将所述目标物料信息填充至所述目标广告模版对象，生成广告数据的步骤，包括：
    将目标物料信息中的物料元素填入所述目标广告模板对象；
    为所述目标广告模板对象中的每一个物料元素配置对应的展示格式。
  13. 一种广告数据生成系统，应用于交易平台，所述交易平台涉及多个用户对象、多个商户对象、多个广告模版对象，以及所述广告模版对象具有对应的多个物料信息，包括：
    采集模块，用于采集所述用户对象的第一特征信息，以及，所述商户对象的第二特征信息；
    模板排序模块，用于依据所述第一特征信息和第二特征信息，从所述多个广告模板对象中，筛选出目标广告模版对象；
    物料决策模块，用于确定与所述目标广告模版对象对应的目标物料信息；
    在线配置模块，用于将所述目标物料信息填充至所述目标广告模版对象，生成广告数据。
  14. 一种电子设备，包括：
    处理器、存储器以及存储在所述存储器上并可在所述处理器上运行的计算机程序，其特征在于，所述处理器执行所述程序时实现如权利要求1至12中一个或多个所述的广告数据生成方法。
  15. 一种计算机程序，包括计算机可读代码，当所述计算机可读代码在计算处理设备上运行时，导致所述计算处理设备执行根据权利要求1至12中的任一个所述的广告数据生成方法。
  16. 一种计算机可读存储介质，其中存储了如权利要求15所述的计算机程序。
PCT/CN2019/120677 2019-01-16 2019-11-25 广告数据生成 WO2020147429A1 (zh)

Applications Claiming Priority (2)

Application Number Priority Date Filing Date Title
CN201910041430.0 2019-01-16
CN201910041430.0A CN109903076A (zh) 2019-01-16 2019-01-16 一种广告数据生成方法、系统、电子设备及存储介质

Publications (1)

Publication Number Publication Date
WO2020147429A1 true WO2020147429A1 (zh) 2020-07-23

Family

ID=66943835

Family Applications (1)

Application Number Title Priority Date Filing Date
PCT/CN2019/120677 WO2020147429A1 (zh) 2019-01-16 2019-11-25 广告数据生成

Country Status (2)

Country Link
CN (1) CN109903076A (zh)
WO (1) WO2020147429A1 (zh)

Families Citing this family (7)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN109903076A (zh) * 2019-01-16 2019-06-18 北京三快在线科技有限公司 一种广告数据生成方法、系统、电子设备及存储介质
CN110704763B (zh) * 2019-09-23 2024-02-20 创新先进技术有限公司 弹屏广告展示方法、装置及电子设备
CN113052619B (zh) * 2019-12-27 2024-05-24 北京搜狗科技发展有限公司 一种广告制作方法及装置
CN111325013B (zh) * 2020-02-06 2022-02-25 北京三快在线科技有限公司 自动生成信息卡的方法、装置、通信系统和存储介质
CN111598616B (zh) * 2020-05-11 2023-08-01 百度在线网络技术（北京）有限公司 对象集合筛选的实现方法、装置、设备以及存储介质
CN111833099B (zh) * 2020-06-24 2021-08-31 广州筷子信息科技有限公司 一种生成创意广告的方法和系统
CN111489419B (zh) * 2020-06-28 2021-01-08 广州筷子信息科技有限公司 一种海报生成的方法和系统

Citations (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN103514209A (zh) * 2012-06-27 2014-01-15 百度在线网络技术（北京）有限公司 基于对象信息库生成待推广对象的推广信息的方法与设备
CN106127528A (zh) * 2016-06-30 2016-11-16 北京小米移动软件有限公司 广告投放方法及装置
US20180225721A1 (en) * 2014-09-29 2018-08-09 Pandora Media, Inc. Dynamically generated audio in advertisements
CN109903076A (zh) * 2019-01-16 2019-06-18 北京三快在线科技有限公司 一种广告数据生成方法、系统、电子设备及存储介质

Patent Citations (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN103514209A (zh) * 2012-06-27 2014-01-15 百度在线网络技术（北京）有限公司 基于对象信息库生成待推广对象的推广信息的方法与设备
US20180225721A1 (en) * 2014-09-29 2018-08-09 Pandora Media, Inc. Dynamically generated audio in advertisements
CN106127528A (zh) * 2016-06-30 2016-11-16 北京小米移动软件有限公司 广告投放方法及装置
CN109903076A (zh) * 2019-01-16 2019-06-18 北京三快在线科技有限公司 一种广告数据生成方法、系统、电子设备及存储介质

Also Published As

Publication number Publication date
CN109903076A (zh) 2019-06-18

Similar Documents

Publication Publication Date Title
WO2020147429A1 (zh) 广告数据生成
US20210287250A1 (en) Providing data and analysis for advertising on networked devices
US11995112B2 (en) System and method for information recommendation
Varadarajan et al. Marketing strategy in an internet-enabled environment: a retrospective on the first ten years of JIM and a prospective on the next ten years
US20160125456A1 (en) Advertising campaign targeting using contextual data
JP5662446B2 (ja) リアルタイム広告入札に対して競争評価モデルを使用するための学習システム
CN106570718B (zh) 信息的投放方法及投放系统
CN106557480B (zh) 查询改写的实现方法及装置
CN106982256A (zh) 信息推送方法、装置、设备及存储介质
TW201520936A (zh) 使用者參與度上下文相關之非保證交付自動標價
KR102000076B1 (ko) 온라인 쇼핑몰 통합 관리 시스템에서의 온라인 판매채널을 추천하는 방법 및 그 서버
CN102132301A (zh) 使用用户社会信息进行在线广告的系统和方法
US20230005022A1 (en) Structured informational link annotations
Kudyba et al. Introduction to the big data era
US20200265467A1 (en) Method and apparatus for providing web advertisements to users
CN108429776B (zh) 网络对象的推送方法、装置、客户端、交互设备以及系统
WO2015124024A1 (zh) 一种提升信息的曝光率的方法和装置、确定搜索词的价值的方法和装置
KR102422410B1 (ko) 클라이언트와 인플루언서의 매칭 시스템 및 방법
KR102409719B1 (ko) 공동구매 관리 시스템 및 방법
US20180322513A1 (en) Tracking performance of digital design asset attributes
CN104063799A (zh) 一种推广信息的推送方法及装置
US20110251889A1 (en) Inventory clustering
CN112541111A (zh) 基于去中心化大数据检索市场的商品检索及商品推荐系统
US11562298B1 (en) Predictive analytics using first-party data of long-term conversion entities
CN105469291A (zh) 用户信息提供方法及装置

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: 19909788

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: 19909788

Country of ref document: EP

Kind code of ref document: A1