WO2022245279A1 - 特征构建方法、内容显示方法及相关装置 - Google Patents

特征构建方法、内容显示方法及相关装置 Download PDF

Info

Publication number
WO2022245279A1
WO2022245279A1 PCT/SG2022/050254 SG2022050254W WO2022245279A1 WO 2022245279 A1 WO2022245279 A1 WO 2022245279A1 SG 2022050254 W SG2022050254 W SG 2022050254W WO 2022245279 A1 WO2022245279 A1 WO 2022245279A1
Authority
WO
WIPO (PCT)
Prior art keywords
user
content
conversion data
target
features
Prior art date
Application number
PCT/SG2022/050254
Other languages
English (en)
French (fr)
Inventor
熊泓宇
汪罕
刘臻
张皓程
刘宾
吴云飞
易潇
陆闯
Original Assignee
脸萌有限公司
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 脸萌有限公司 filed Critical 脸萌有限公司
Publication of WO2022245279A1 publication Critical patent/WO2022245279A1/zh

Links

Classifications

    • GPHYSICS
    • G06COMPUTING; CALCULATING OR COUNTING
    • G06FELECTRIC DIGITAL DATA PROCESSING
    • G06F16/00Information retrieval; Database structures therefor; File system structures therefor
    • G06F16/90Details of database functions independent of the retrieved data types
    • G06F16/95Retrieval from the web
    • G06F16/957Browsing optimisation, e.g. caching or content distillation
    • G06F16/9574Browsing optimisation, e.g. caching or content distillation of access to content, e.g. by caching
    • YGENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
    • Y02TECHNOLOGIES OR APPLICATIONS FOR MITIGATION OR ADAPTATION AGAINST CLIMATE CHANGE
    • Y02DCLIMATE CHANGE MITIGATION TECHNOLOGIES IN INFORMATION AND COMMUNICATION TECHNOLOGIES [ICT], I.E. INFORMATION AND COMMUNICATION TECHNOLOGIES AIMING AT THE REDUCTION OF THEIR OWN ENERGY USE
    • Y02D10/00Energy efficient computing, e.g. low power processors, power management or thermal management

Definitions

  • the present disclosure provides a feature construction method, the method includes: acquiring user conversion data corresponding to target content, the user conversion data includes non-attributed conversion data, and the non-attributed conversion data is in the first
  • the content platform displays the target content
  • the operation behavior is attributed to the user data of the second content platform
  • the second content platform displays the content related to all content related to the target content
  • classifying the user conversion data to obtain the user conversion data associated with the same user and the user conversion data corresponding to the same content
  • the target user characteristics and the target content characteristics are used to train a content display model
  • the content display model is used to determine the target user displayed content.
  • the present disclosure provides a content display method, the method comprising: acquiring content information of the content to be displayed; inputting the content information of the content to be displayed into a content display model to determine target users, and the content display model It is obtained by training the target user features and target content features constructed according to the method described in the first aspect; and displaying the content to be displayed to the target user.
  • the present disclosure provides a feature construction device, the device comprising: A data acquisition module, configured to acquire user conversion data corresponding to the target content, the user conversion data includes non-attributed conversion data, and the non-attributed conversion data is when the target content is displayed on the first content platform, When the user generates an operation behavior on the target content, attribute the operation behavior to the user data of the second content platform, and the second content platform displays content related to the target content; the data classification module uses for classifying the user conversion data, so as to obtain the user conversion data associated with the same user and the user conversion data corresponding to the same content; a feature construction module, configured to construct according to the user conversion data associated with the same user target user characteristics, and construct target content characteristics according to the user conversion data corresponding to the same content; the target user characteristics and target content characteristics are used to train a content display model, and the content display model is used to determine the content displayed to the target user content.
  • the present disclosure provides a content display device, the device comprising: an acquisition module, configured to acquire content information of content to be displayed; a determination module, configured to input content information of content to be displayed into a content display model, To determine the target user, the content display model is obtained by training the target user characteristics and target content characteristics constructed according to the method described in the first aspect; a display module, configured to display the content to be displayed to the target user.
  • the present disclosure provides a computer-readable medium on which a computer program is stored, and when the program is executed by a processing device, the steps of the method described in the first aspect or the second aspect are implemented.
  • the present disclosure provides an electronic device, including: a storage device, on which a computer program is stored; a processing device, configured to execute the computer program in the storage device, so as to realize the first aspect or the second aspect steps of the method described in .
  • the target user features and target content features can be constructed in combination with non-attributed conversion data, and the target user features and target content features can be used to train the content display model, so as to train the content display model through more abundant data, not only The result accuracy of the content display model can be improved, the waste of content display resources can be reduced, and the data utilization rate of non-attributed conversion data can also be improved.
  • FIG. 1 is a flowchart of a feature construction method according to an exemplary embodiment of the present disclosure
  • Fig. 2 is a flowchart of a content display method according to an exemplary embodiment of the present disclosure
  • Fig. 1 is a flowchart of a feature construction method according to an exemplary embodiment of the present disclosure
  • Fig. 2 is a flowchart of a content display method according to an exemplary embodiment of the present disclosure
  • Fig. 1 is a flowchart of a feature construction method according to an exemplary embodiment of the present disclosure
  • Fig. 2 is a flowchart of a content display method according to an exemplary embodiment of the present disclosure
  • FIG. 3 is a block diagram of a feature construction device according to an exemplary embodiment of the present disclosure.
  • An exemplary embodiment shows a block diagram of a content display device;
  • FIG. 5 is a block diagram of an electronic device according to an exemplary embodiment of the present disclosure.
  • Behavior is attributed to the data of other content platforms. Related technologies usually analyze user behavior based on attribution conversion data. However, with the emergence of various content platforms and applications, the reasons why users generate subscriptions, downloads, and other behaviors have also become complicated. If the attribution conversion data is used only for analysis, the user behavior trajectory cannot be well restored, thereby affecting the subsequent display of corresponding content to the user, resulting in a waste of content display resources. The inventor found through a large amount of data analysis that compared with attribution conversion data, non-attributed conversion data has significantly improved user activation behavior, user download behavior, etc.
  • the present disclosure provides a feature construction method to construct target user features and target content features in combination with non-attributed conversion data, and
  • the target user features and target content features are used to train the content display model to pass more Rich data training content display model can not only improve the accuracy of the results of the content display model, reduce the waste of content display resources, but also improve the data utilization rate of non-attributed conversion data, and can also improve the user conversion rate in the context of advertising content .
  • the user conversion rate can be understood as the conversion rate from a user clicking an advertisement to becoming an effective active user or a registered user.
  • the acquisition of user conversion data in the present disclosure may first display an authorization interface for data acquisition to the user, such as displaying a prompt box asking the user whether to agree to upload the corresponding user data.
  • the user conversion data corresponding to the user can be obtained for feature construction. That is to say, the user transformation data in this disclosure is obtained under the authorization and consent of the user.
  • Fig. 1 is a flow chart showing a feature construction method according to an exemplary embodiment of the present disclosure. Referring to FIG. 1, the feature construction method includes: Step 101, acquiring user conversion data corresponding to target content, where the user conversion data includes non-attributed conversion data.
  • the non-attributed conversion data is user data that attributes the operation behavior to the second content platform when the user performs an operation behavior on the target content when the first content platform displays the target content, and the second content platform displays There is content that is relevant to the target content.
  • Step 102 classify the user conversion data to obtain user conversion data associated with the same user and user conversion data corresponding to the same content.
  • Step 103 construct target user features according to user conversion data associated with the same user, and construct target content features according to user conversion data corresponding to the same content, target user features and target content features are used to train the content display model, and the content display model uses to determine what content to display to target users.
  • the user conversion data disclosed in this disclosure is obtained under the authorization and consent of the user.
  • an authorization interface for data acquisition may be displayed first, and the authorization interface is used to prompt the user whether to allow the user data corresponding to him to be acquired.
  • step 101 may be: Responding to the user's authorization operation in the authorization interface, obtaining user conversion data corresponding to the target content, the authorization operation is an operation triggered by the user to allow obtaining user data corresponding to itself.
  • a prompt pop-up box is displayed to the user asking whether the user agrees to the content platform obtaining its corresponding user data, and the prompt pop-up box may include controls displayed as "agree” and "disagree”.
  • the user conversion data may be user data used to characterize operations such as subscription and download.
  • user conversion data of different time lengths can be obtained for subsequent feature construction, for example, user conversion data of the past 1 day, 7 days, and 30 days can be respectively obtained for subsequent feature construction.
  • the user conversion data may include non-attribution conversion data.
  • the non-attributed conversion data is user data attributed to other content platforms, so it is difficult for this content platform to obtain non-attributed conversion data, and it is impossible to directly obtain non-attributed conversion data from this content platform.
  • the non-attributed data corresponding to the target content can be obtained from a third-party data platform, and the third-party data platform is used to collect the non-attributed conversion data corresponding to the target content. It should be understood that the non-attributed conversion data obtained from the third-party data platform is relatively messy.
  • user conversion data can be classified first to obtain user conversion data associated with the same user and users corresponding to the same content. Convert data.
  • the user conversion data associated with the same user can be determined through the user device information included in the user conversion data, and the user conversion data corresponding to the same content can be determined through the content identification information included in the user conversion data.
  • the user conversion data may also include attribution conversion data.
  • the attribution conversion data is to attribute the operation behavior when the user performs an operation behavior on the target content when the first content platform displays the target content. Due to the user data of the first content platform, the attributed conversion data associated with the same user can be associated with the non-attributed conversion data to obtain the target user conversion data, and then the target user conversion data can be classified to obtain the same The user conversion data associated with the user and the user conversion data corresponding to the same content.
  • the attribution conversion data corresponding to the target content can be obtained from the first content platform, and then combined with the non-attribution conversion data corresponding to the target content to construct features.
  • the non-attributed conversion data obtained from the third data platform is relatively messy, so in order to facilitate subsequent feature construction, the non-attributed conversion data corresponding to the same user can be associated with the attributable conversion data first,
  • the conversion data of the target user is obtained by classifying the non-attribution conversion data and the attribution conversion data corresponding to the same user.
  • non-attributed conversion data and attributable conversion data corresponding to the same user may also be associated through other user information to obtain target user conversion data, which is not limited in this embodiment of the present disclosure.
  • target user conversion data is further classified to obtain user conversion data associated with the same user and user conversion data corresponding to the same content.
  • target user conversion data can be classified by user equipment information to obtain user conversion data associated with the same user, and target user conversion data can be classified by content identification information to obtain user conversion data corresponding to the same content.
  • attribution conversion data and non-attribution conversion data can be combined for subsequent Feature construction, compared with the method of feature construction only through attribution transformation data, can obtain richer features, so as to train the content display model through richer data, improve the accuracy of the results of the content display model, and reduce the cost of content display resources. Waste, in the advertising scenario can also improve the user conversion rate.
  • the target user characteristics can be constructed according to the user conversion data associated with the same user, and the target content characteristics can be constructed according to the user conversion data corresponding to the same content.
  • constructing target user features based on user conversion data associated with the same user may be: Constructing at least one of the following target user features based on non-attributed conversion data associated with the same user: list user features, numeric values User characteristics and recency characteristics.
  • list-type user features can be used to represent the content platform features or content features visited by the same user before generating the operation behavior
  • numerical user characteristics can be used to represent the quantitative characteristics of the operation behavior generated by the same user
  • the recency feature can be It is used to characterize the time interval between the generation time of the latest operation behavior and the current time.
  • Table 2 shows possible target user features constructed based on non-attributed conversion data associated with the same user, including list-type user features, numeric-type user features, and recency features.
  • Table 2 It should be understood that the target user characteristics shown in Table 2 are only illustrative, and more target user characteristics can be constructed according to actual needs during the specific implementation of the present disclosure. For example, in the case of user authorization, if the user purchases item A, by constructing list-type user characteristics for non-attributed conversion data, it can be determined which content platforms the user has visited before purchasing item A, or which item information has been browsed , that is, it is possible to determine which content the user is interested in.
  • the quantity of item A purchased by the user can be determined, that is, the degree of interest of the user in item A can be determined.
  • the recency feature of the non-attributed conversion data it can be determined when the user purchased item A this time and the last time The time interval between purchasing item A for the first time can determine which contents the user is interested in recently. Therefore, in the case of user authorization, by constructing different types of user characteristics, the behavior track of the user's purchase of item A can be better restored, so that the corresponding content can be displayed to the user more accurately and the waste of content display resources can be reduced.
  • constructing the target content features according to the user conversion data corresponding to the same content may be: constructing the following target content features according to the non-attributed conversion data corresponding to the same content: list content features and/or numerical content features .
  • the list-type content features are used to represent the characteristics of users who generate business operation behaviors for the same content
  • the value-type content features are used to represent the quantitative characteristics of the operation behaviors corresponding to the same content.
  • Table 3 shows possible target content features constructed based on non-attributed conversion data corresponding to the same content, including list-type content features and numerical-type content features.
  • the target content features shown in Table 3 are only illustrative, and more target content features can be constructed according to actual needs during the specific implementation of the present disclosure.
  • the target content features can be constructed according to the content side data in the non-attributed conversion data, so as to build different types of content features, and better restore the user's behavior track for the target content, so that Display the corresponding content to the user more accurately, and reduce the waste of content display resources.
  • constructing the target user characteristics may also be: according to the non-attributed conversion data associated with the same user, determining multiple content data that the same user has generated operation behaviors, and according to multiple Content data, determine the co-occurrence content characteristics corresponding to the user, to obtain the target user characteristics.
  • constructing target content features may also be: according to the non-attribution conversion data corresponding to the same content, determining multiple user identification data that have performed operations on the same content, and according to multiple The user identification data is used to determine the co-occurrence user characteristics corresponding to the content, so as to obtain the target content characteristics.
  • the co-occurrence content characteristics corresponding to the user can be determined according to the content data corresponding to advertisement 1 and advertisement 2 , for example, first extract the content information features from the content data corresponding to advertisement 1 and advertisement 2, and then use the sum of the content information features of advertisement 1 and advertisement 2 as the co-occurrence content feature, or take the same content information of advertisement 1 and advertisement 2 Features are used as co-occurrence content features, etc., to obtain the target user features corresponding to the user.
  • the co-occurrence user characteristics corresponding to the advertisement 3 can be determined according to the user identification data of the first user and the second user, for example, first The user information features are extracted from the user identification data of the first user and the second user, and then the sum of the user information features of the first user and the second user is used as the co-occurrence user feature, or the same User information features are used as co-occurring user features, etc. to obtain target content features corresponding to advertisement 3 .
  • co-occurrence features can be constructed based on multiple content data that the same user has operated in the non-attributed conversion data and multiple user identification data associated with the same content, so as to learn from multiple users or Multi-content dimension construction is used to train the characteristics of the content display model, improve the accuracy of the results of the content display model, thereby more accurately displaying the corresponding content to users, reducing the waste of content display resources, and further improving the accuracy of non-attributed conversion data
  • the data utilization rate can also improve the user conversion rate in the context of advertising content.
  • the present disclosure also provides a content display method, including: Step 201, acquiring content information of the content to be displayed; Step 202, inputting the content information of the content to be displayed into the content display model to determine the target user, the content
  • the display model is obtained by training the target user features and target content features constructed according to any feature construction method provided in the present disclosure; Step 203, displaying the content to be displayed to the target user.
  • the content information is used to represent the basic content such as text and pictures of the content page.
  • the content information of the target content can be obtained, and then the content information can be input into the pre-trained content display model to Obtain the target user whose target content is to be displayed.
  • the content information characteristics of the content in the attribution conversion data are usually input into the content display model for estimation to obtain estimated users, and then the estimated users are combined with The user associated with the content in the attribution conversion data is compared to calculate the loss function. Then backpropagation is performed according to the calculation result of the loss function to update the model parameters.
  • it will repeatedly execute the input of the content information features in the attribution conversion data into the model for estimation, obtain the estimated user, compare the estimated user with the user associated with the content in the attribution conversion data to calculate the loss function, and calculate the loss function according to The calculation result of the loss function is backpropagated to update the process of the model parameters until the loss function is no longer significantly reduced.
  • the content information of the target content can be input into the model to obtain the target users of the target content to be displayed.
  • this method only analyzes the attribution conversion data, and cannot restore the user behavior trajectory well, thereby affecting the accuracy of the final target user, and cannot better realize the push display of the target content. It causes a waste of content display resources.
  • this disclosure proposes a new content display method, which can be combined with non-attribution Transform data to construct target user features and target content features, and use the target user features and target content features to train the content display model, so as to train the content display model through richer data, which can not only improve the result accuracy of the content display model, Reducing the waste of content display resources can also improve the data utilization rate of non-attributed conversion data, and can also increase the user conversion rate in the context of advertising content.
  • the relevant contents of the target user characteristics and the target content characteristics have been described above, and will not be repeated here.
  • the present disclosure also provides a feature construction device, which can become part or all of an electronic device through software, hardware or a combination of both. Referring to FIG.
  • the feature construction device 300 includes: a data acquisition module 301, configured to acquire user conversion data corresponding to target content, the user conversion data includes non-attributed conversion data, and the non-attributed conversion data is When a content platform displays the target content, when the user generates an operation behavior on the target content, attribute the operation behavior to the user data of the second content platform, and the second content platform displays the content related to content related to the target content; a data classification module 302, configured to classify the user conversion data, so as to obtain the user conversion data associated with the same user and the user conversion data corresponding to the same content; a feature construction module 303 , for constructing target user features according to the user conversion data associated with the same user, and constructing target content features according to the user conversion data corresponding to the same content; the target user features and target content features are used for training content display A model, the content display model is used to determine the content displayed to the target user.
  • a data acquisition module 301 configured to acquire user conversion data corresponding to target content
  • the user conversion data includes non-attributed conversion data
  • the apparatus 300 further includes: a display module, configured to display an authorization interface for data acquisition before acquiring the user conversion data corresponding to the target content, and the authorization interface is used to prompt the user whether to allow the acquisition of the user's corresponding conversion data.
  • User data the data acquisition module is used to acquire the user conversion data corresponding to the target content in response to the user's authorization operation in the authorization interface, and the authorization operation is an operation triggered by the user to allow the user to obtain the corresponding user data .
  • the feature construction module 303 is configured to: construct at least one of the following target user features according to the non-attributed conversion data associated with the same user: list-type user features, numerical user features, and recency features .
  • the feature construction module 303 is configured to: construct the following target content features according to the non-attributed conversion data corresponding to the same content: list content features and/or numerical content features, wherein the List-type content features are used to characterize user features that generate business operations for the same content, and the numerical content features Quantitative features used to characterize the operation behavior corresponding to the same content.
  • the feature construction module 303 is configured to: according to the non-attributed conversion data associated with the same user, determine multiple content data in which the same user has generated the operation behavior, and according to the multiple content data data, determining co-occurrence content features corresponding to the user to obtain the target user features; constructing target content features according to the user conversion data corresponding to the same content, including: according to the non-attributed conversion corresponding to the same content data, determining multiple user identification data that have generated the operation behavior on the same content, and determining co-occurrence user characteristics corresponding to the content according to the multiple user identification data, so as to obtain the target content characteristics.
  • the user conversion data further includes attribution conversion data
  • the attribution conversion data is when the target content is displayed on the first content platform, when the user generates the target content
  • the apparatus 300 further includes: an associating module, configured to associate the attribution conversion data corresponding to the same user with the non-attribution Correlating the conversion data to obtain target user conversion data; the data classification module 302 is configured to: classify the target user conversion data, so as to obtain the user conversion data associated with the same user and the same content corresponding to the User conversion data.
  • the present disclosure also provides a content display device, which can become part or all of an electronic device through software, hardware or a combination of both.
  • the content display device 400 includes: an acquisition module 401, configured to acquire content information of the content to be displayed; a determination module 402, configured to input the content information of the content to be displayed into the content display model, so as to determine the target user ,
  • the content display model is obtained by training the target user features and target content features constructed according to the method described in the first aspect; a display module 403, configured to display the content to be displayed to the target user.
  • the electronic device may include the feature constructing device as shown in FIG. 3 and the content display device as shown in FIG. 4 .
  • the target user feature and the target content feature can be constructed by the feature building device, which are used for training the content display model in the content display device.
  • the present disclosure also provides a computer-readable medium on which a computer program is stored, and when the program is executed by a processing device, the steps of any of the above-mentioned feature construction methods or any of the above-mentioned content display methods are implemented.
  • the present disclosure also provides an electronic device, including: a storage device, on which a computer program is stored; a processing device, configured to execute the computer program in the storage device, so as to realize any of the above-mentioned features A method or any of the above shows the steps of a method.
  • FIG. 5 it shows a schematic structural diagram of an electronic device 500 suitable for implementing an embodiment of the present disclosure.
  • the terminal devices in the embodiments of the present disclosure may include but not limited to mobile phones, notebook computers, digital broadcast receivers, PDAs (personal digital assistants), PADs (tablet computers), PMPs (portable multimedia players), vehicle-mounted terminals (eg mobile terminals such as car navigation terminals) and fixed terminals such as digital TVs, desktop computers, and the like.
  • the electronic device shown in FIG. 5 is only an example, and should not limit the functions and scope of use of the embodiments of the present disclosure. As shown in FIG.
  • an electronic device 500 may include a processing device (such as a central processing unit, a graphics processing unit, etc.) 501, which may be randomly accessed according to a program stored in a read-only memory (ROM) 502 or loaded from a storage device 508 Various appropriate actions and processes are executed by programs in the memory (RAM) 503 . In the RAM 503, various programs and data necessary for the operation of the electronic device 500 are also stored.
  • the processing device 501 , ROM 502 and RAM 503 are connected to each other through a bus 504 .
  • An input/output (I/O) interface 505 is also connected to the bus 504 .
  • input devices 506 including, for example, a touch screen, a touch pad, a keyboard, a mouse, a camera, a microphone, an accelerometer, and a gyroscope; including, for example, a liquid crystal display (LCD), a speaker, a vibration output device 507 such as a device; including a storage device 508 such as a magnetic tape, a hard disk, etc.; and a communication device 509.
  • the communication means 509 may allow the electronic device 500 to perform wireless or wired communication with other devices to exchange data. While FIG. 5 shows electronic device 500 having various means, it should be understood that implementing or possessing all of the illustrated means is not a requirement.
  • the processes described above with reference to the flowcharts can be implemented as computer software programs.
  • the embodiments of the present disclosure include a computer program product, which includes a computer program carried on a non-transitory computer readable medium, where the computer program includes program code for executing the method shown in the flowchart.
  • the computer program may be downloaded and installed from a network via communication means 509 , or from storage means 508 , or from ROM 502 .
  • the processing device 501 the above-mentioned functions defined in the methods of the embodiments of the present disclosure are executed.
  • the computer-readable medium mentioned above in the present disclosure may be a computer-readable signal medium or a computer-readable storage medium or any combination of the above two.
  • a computer-readable storage medium may be, for example, but not limited to, an electrical, magnetic, optical, electromagnetic, infrared, or semiconductor system, device, or device, or any combination thereof. More specific examples of computer readable storage media may include, but are not limited to: electrical connections with one or more conductors, portable computer disks, hard disks, random access memory (RAM), read only memory (ROM), erasable Programmable read-only memory (EPROM or flash memory), optical fiber, portable compact disk read-only memory (CD-ROM), optical storage device, magnetic storage device, or any suitable combination of the above.
  • a computer-readable storage medium may be any tangible medium containing or storing a program, and the program may be executed by an instruction execution system, installed configuration or device use or in conjunction with it.
  • a computer-readable signal medium may include a data signal propagated in a baseband or as part of a carrier wave, in which computer-readable program codes are carried. The propagated data signal may take various forms, including but not limited to electromagnetic signal, optical signal, or any suitable combination of the above.
  • the computer-readable signal medium may also be any computer-readable medium other than the computer-readable storage medium, and the computer-readable signal medium may send, propagate or transmit a program for use by or in combination with an instruction execution system, apparatus or device .
  • the program code contained on the computer readable medium can be transmitted by any appropriate medium, including but not limited to: electric wire, optical cable, RF (radio frequency), etc., or any suitable combination of the above.
  • any currently known or future-developed network protocol such as HTTP (HyperText Transfer Protocol, Hypertext Transfer Protocol) can be used for communication, and can communicate with digital data in any form or medium (for example, communication network) interconnection.
  • Examples of communication networks include local area networks (“LANs”), wide area networks (“WANs”), internetworks (e.g., the Internet), and peer-to-peer networks (e.g., ad hoc peer-to-peer networks), as well as any currently known or future developed network of.
  • the above-mentioned computer-readable medium may be included in the above-mentioned electronic device; or it may exist independently without being assembled into the electronic device.
  • the above-mentioned computer-readable medium carries one or more programs, and when the above-mentioned one or more programs are executed by the electronic device, the electronic device: acquires user conversion data corresponding to the target content, and the user conversion data includes non-attributed Conversion data, the non-attribution conversion data is when the user performs an operation behavior on the target content when the target content is displayed on the first content platform, attributing the operation behavior to the second content platform user data, the second content platform displays content related to the target content; classifying the user conversion data to obtain the user conversion data associated with the same user and the user conversion corresponding to the same content data; constructing target user features according to the user conversion data associated with the same user, and constructing target content features according to the user conversion data corresponding to the same content; the target user features and the target content features are used for training content
  • a display model the content display model is used
  • Computer program code for carrying out operations of the present disclosure may be written in one or more programming languages, or combinations thereof, including but not limited to object-oriented programming languages such as Java, Smalltalk, C++, and Included are conventional procedural programming languages such as the "C" language or similar programming languages.
  • the program code may execute entirely on the user's computer, partly on the user's computer, as a stand-alone software package, partly on the user's computer and partly on a remote computer or entirely on the remote computer or server.
  • the remote computer can be connected to the user computer through any kind of network, including a local area network (LAN) or a wide area network (WAN), or it can be connected to an external computer (for example, using an Internet service provider to via Internet connection).
  • LAN local area network
  • WAN wide area network
  • each block in the flowchart or block diagram may represent a module, program segment, or part of code that contains one or more logic functions for implementing the specified executable instructions.
  • the functions noted in the block may occur out of the order noted in the figures. For example, two blocks shown in succession may, in fact, be executed substantially concurrently, or they may sometimes be executed in the reverse order, depending upon the functionality involved.
  • each block in the block diagrams and/or flowcharts, and combinations of blocks in the block diagrams and/or flowcharts can be implemented by a dedicated hardware-based system that performs specified functions or operations. , or may be implemented by a combination of special purpose hardware and computer instructions.
  • the modules involved in the embodiments described in the present disclosure may be implemented by software or by hardware. Wherein, the name of the module does not constitute a limitation on the module itself under certain circumstances.
  • the functions described herein above may be performed at least in part by one or more hardware logic components.
  • exemplary types of hardware logic components include: field programmable gate array (FPGA), application specific integrated circuit (ASIC), application specific standard product (ASSP), system on chip (SOC), complex programmable Logical device (CPLD) and so on.
  • FPGA field programmable gate array
  • ASIC application specific integrated circuit
  • ASSP application specific standard product
  • SOC system on chip
  • CPLD complex programmable Logical device
  • a machine-readable medium may be a tangible medium, which may contain or store a program for use by or in combination with an instruction execution system, device, or device.
  • a machine-readable medium may be a machine-readable signal medium or a machine-readable storage medium.
  • a machine-readable medium may include, but is not limited to, an electronic, magnetic, optical, electromagnetic, infrared, or semiconductor system, device, or device, or any suitable combination of the foregoing.
  • machine-readable storage media would include one or more wire-based electrical connections, portable computer disks, hard disks, Random Access Memory (RAM), Read Only Memory (ROM), Erasable Programmable Read Only Memory (EPROM or flash memory), optical fiber, compact disk read only memory (CD-ROM), optical storage, magnetic storage, or any suitable combination of the foregoing.
  • RAM Random Access Memory
  • ROM Read Only Memory
  • EPROM Erasable Programmable Read Only Memory
  • CD-ROM compact disk read only memory
  • magnetic storage or any suitable combination of the foregoing.
  • Example 1 provides a feature construction method, the method comprising: acquiring user conversion data corresponding to target content, the user conversion data including non-attributed conversion data, the non-attributed conversion data
  • the attribution conversion data is user data that attributes the operation behavior to the second content platform when the user performs an operation behavior on the target content when the target content is displayed on the first content platform, and the The second content platform displays content related to the target content; classifies the user conversion data to obtain the user conversion data associated with the same user and the user conversion data corresponding to the same content; Constructing target user features of the user conversion data, and constructing target content features according to the user conversion data corresponding to the same content; the target user features and the target content features are used to train a content display model, and the content display model Used to determine what content to display to targeted users.
  • Example 2 provides the method of Example 1. Before obtaining the user conversion data corresponding to the target content, the method further includes: displaying an authorization interface for data acquisition, and the authorization interface is used for Prompting the user whether to allow the acquisition of user data corresponding to the user; said obtaining the user conversion data corresponding to the target content includes: responding to the user's authorization operation in the authorization interface, obtaining the user conversion data corresponding to the target content, the authorization operation It is an operation triggered by the user that allows to obtain the user data corresponding to itself.
  • Example 3 provides the method of Example 1, constructing target user characteristics according to the user conversion data associated with the same user, including: according to the non-attributed conversion data associated with the same user , constructing at least one of the following target user features: a list-type user feature, a value-type user feature, and a recency feature.
  • Example 4 provides the method of any one of Examples 1-3, constructing target content features according to the user conversion data corresponding to the same content, including: according to the user conversion data corresponding to the same content For non-attributed conversion data, construct the following target content features: list-type content features and/or numerical content features, wherein the list-type content features are used to characterize user characteristics that generate business operation behaviors for the same content, and the numerical-type content features The content feature is used to characterize the quantitative feature of the operation behavior corresponding to the same content.
  • Example 5 provides the method of any one of Examples 1-3, constructing target user characteristics according to the user conversion data associated with the same user, including: according to the user conversion data associated with the same user For non-attributed conversion data, determine multiple content data in which the same user has generated the operation behavior, and determine co-occurrence content characteristics corresponding to the user according to the multiple content data, so as to obtain the target user characteristics; Constructing target content features for the user conversion data corresponding to the same content, including: according to the non-attributed conversion data corresponding to the same content, determining multiple user identification data that have generated the operation behavior for the same content, and according to the The multiple user identification data are used to determine the co-occurring user characteristics corresponding to the content, so as to obtain the target content characteristics.
  • Example 6 provides the method of any one of Examples 1-3, the user conversion data further includes attribution conversion data, and the attribution conversion data is In the case where the target content is displayed on the platform, when the user generates an operation behavior on the target content, attributing the operation behavior to the user data of the first content platform, the method further includes: assigning the same user The corresponding attributed conversion data is associated with the non-attributed conversion data to obtain the target User conversion data; classifying the user conversion data to obtain the user conversion data associated with the same user and the user conversion data corresponding to the same content, including: classifying the target user conversion data to obtain the same The user conversion data associated with the user and the user conversion data corresponding to the same content.
  • Example 7 provides a content display method, the method comprising: acquiring content information of content to be displayed; inputting the content information of content to be displayed into a content display model to determine For target users, the content display model is obtained by training the target user features and target content features constructed according to the method described in Example 1; and displaying the content to be displayed to the target users.
  • Example 8 provides a feature construction device, the device includes: a data acquisition module, configured to acquire user conversion data corresponding to target content, the user conversion data includes non-attributed Conversion data, the non-attribution conversion data is when the user performs an operation behavior on the target content when the target content is displayed on the first content platform, attributing the operation behavior to the second content platform the user data, the second content platform displays content related to the target content; a data classification module, configured to classify the user conversion data, so as to obtain the user conversion data and the same content associated with the same user The corresponding user conversion data; a feature building module, configured to construct target user features according to the user conversion data associated with the same user, and construct target content features according to the user conversion data corresponding to the same content; the target User features and target content features are used to train a content display model that is used to determine content to be displayed to target users.
  • a data acquisition module configured to acquire user conversion data corresponding to target content
  • the user conversion data includes non-attributed Conversion data
  • the non-attribution conversion data is when the
  • Example 9 provides the device of Example 8, and the device further includes: a display module, configured to display an authorization interface for data acquisition before acquiring user conversion data corresponding to target content, so The authorization interface is used to prompt the user whether to allow the user to obtain the corresponding user data; the data acquisition module is used to obtain the user conversion data corresponding to the target content in response to the user's authorization operation in the authorization interface, and the authorization operation is An operation triggered by a user that allows obtaining its own user data.
  • a display module configured to display an authorization interface for data acquisition before acquiring user conversion data corresponding to target content, so The authorization interface is used to prompt the user whether to allow the user to obtain the corresponding user data
  • the data acquisition module is used to obtain the user conversion data corresponding to the target content in response to the user's authorization operation in the authorization interface
  • the authorization operation is An operation triggered by a user that allows obtaining its own user data.
  • Example 10 provides the device of Example 8, the feature construction module is configured to: construct at least one of the following target user features according to the non-attributed conversion data associated with the same user: List user features, numeric user features, and recency features.
  • Example 11 provides the device of any one of Examples 8-10, and the feature building module is used for: According to the non-attributed conversion data corresponding to the same content, construct the following target content features: list content features and/or numerical content features, wherein the list content features are used to represent the business operation behavior for the same content User characteristics, the numerical content characteristics are used to represent the quantitative characteristics of the operation behavior corresponding to the same content.
  • Example 12 provides the device of any one of Examples 8-10, the feature building module is configured to: determine that the same user generates a plurality of content data of the operation behavior, and according to the plurality of content data, determine co-occurrence content characteristics corresponding to the user, so as to obtain the target user characteristics; according to the user conversion data corresponding to the same content, Constructing target content features, including: according to the non-attributed conversion data corresponding to the same content, determining multiple user identification data that have generated the operation behavior on the same content, and according to the multiple user identification data, determining the co-occurring user features corresponding to the content to obtain the target content features.
  • Example 13 provides the device of any one of Examples 8-10, the user conversion data further includes attribution conversion data, and the attribution conversion data is
  • the platform displays the target content
  • the device further includes: an association module, for associating the attribution conversion data corresponding to the same user with the non-attribution conversion data to obtain target user conversion data;
  • the data classification module is used for: classifying the target user conversion data to The user conversion data associated with the same user and the user conversion data corresponding to the same content are obtained.
  • Example 14 provides a content display device, the device comprising: an acquisition module, configured to acquire content information of content to be displayed; a determination module, configured to convert the content to be displayed Input the content information into the content display model to determine the target user, the content display model is obtained by training the target user characteristics and target content characteristics constructed according to the method described in Example 1; a display module, used to display to the target user The content to be displayed.
  • Example 15 provides a computer-readable medium on which a computer program is stored, and when the program is executed by a processing device, the steps of any one of the methods described in Examples 1-7 are implemented. .
  • Example 16 provides an electronic device, including: A storage device, on which a computer program is stored; a processing device, configured to execute the computer program in the storage device, so as to implement the steps of any one of the methods in Examples 1-7.
  • a storage device on which a computer program is stored
  • a processing device configured to execute the computer program in the storage device, so as to implement the steps of any one of the methods in Examples 1-7.

Landscapes

  • Engineering & Computer Science (AREA)
  • Databases & Information Systems (AREA)
  • Theoretical Computer Science (AREA)
  • Data Mining & Analysis (AREA)
  • Physics & Mathematics (AREA)
  • General Engineering & Computer Science (AREA)
  • General Physics & Mathematics (AREA)
  • Information Transfer Between Computers (AREA)

Abstract

本公开涉及一种特征构建方法、内容显示方法及相关装置,其中,特征构建方法包括:获取目标内容对应的用户转化数据,用户转化数据包括非归因转化数据;对用户转化数据进行分类,以得到同一用户关联的用户转化数据和同一内容对应的用户转化数据;根据同一用户关联的用户转化数据,构建目标用户特征,并根据同一内容对应的用户转5 化数据,构建目标内容特征;目标用户特征和目标内容特征用于训练内容显示模型,内容显示模型用于确定向目标用户显示的内容。

Description

特 征构建方法、 内容显示方法及相关装置 本公开要求于 2021年 05月 21 日提交的, 申请名称为 “特征构建方法、 内容显示 方法及相 关装置 ” 的、 中国专利申请号为 “ 202110558744.5” 的优先权, 该中国专利申 请的全部 内容通过引用结合在本公开中。 技术领域 本公开涉及计算机技术领域 , 具体地, 涉及一种特征构建方法、 内容显示方法及相 关装置。 背景技术 归因转化数据为在内容平 台展示内容, 并将用户产生订阅、 下载等行为归因到该内 容平台 的数据。 相关技术通常是根据归因转化数据分析用户行为。 但随着各种内容平台 和应用程序 的层出不穷, 用户产生订阅、 下载等行为的原因也变得复杂。 如果仅通过归 因转化数据进 行分析, 无法较好的还原用户行为轨迹, 从而影响后续向用户显示对应的 内容, 造成内容显示资源的浪费。 发明内容 提供该发明内容部分 以便以简要的形式介绍构思, 这些构思将在后面的具体实施方 式部分被详细 描述。该发明内容部分并不旨在标识要求保 护的技术方案的关键特征或必 要特征 , 也不旨在用于限制所要求的保护的技术方案的范围。 第一方面,本公开提供一种特征构建方 法,所述方法包括: 获取目标内容对应的用 户转化数据 , 所述用户转化数据包括非归因转化数据, 所述非归因转化数据为在第一内 容平台展示 有所述目标内容的情况下 , 当用户对所述目标内容产生操作行为时, 将所述 操作行为归 因到第二内容平台的用户数据 , 所述第二内容平台展示有与所述目标内容相 关的内容 ; 对所述用户转化数据进行分类, 以得到同一用户关联的所述用户转化数据和 同一内容对应 的所述用户转化数据; 根据同一用户关联的所述用户转化数据, 构建目标 用户特征 , 并根据同一内容对应的所述用户转化数据, 构建目标内容特征; 所述目标用 户特征和所述 目标内容特征用于训练内容显示 模型, 所述内容显示模型用于确定向目标 用户显示的 内容。 第二方面, 本公开提供一种内容显示方法, 所述方法包括: 获取待显示内容的内容信 息; 将所述待显示内容的 内容信息输入内容显示模型, 以确定目标用户, 所述内容显示 模型是根据 第一方面所述方法构建的 目标用户特征和目标内容特征训练得 到的; 向所述目标用户显示所述待 显示内容。 第三方面, 本公开提供一种特征构建装置, 所述装置包括: 数据获取模块, 用于获取目标内容对应的用户转化数据, 所述用户转化数据包括非 归因转化数据 , 所述非归因转化数据为在第一内容平台展示有所 述目标内容的情况下 , 当用户对所述 目标内容产生操作行为时, 将所述操作行为归因到第二内容平台的用户数 据, 所述第二内容平台展示有与所述 目标内容相关的内容; 数据分类模块, 用于对所述用户转化数据进行分类, 以得到同一用户关联的所述用 户转化数据和 同一内容对应的所述用户转化数据 ; 特征构建模块, 用于根据同一用户关联的所述用户转 化数据, 构建目标用户特征, 并根据同一 内容对应的所述用户转化数据, 构建目标内容特征; 所述目标用户特征和 目标内容特征用于训练内容显示模型, 所述内容显示模型用于 确定向 目标用户显示的内容。 第四方面, 本公开提供一种内容显示装置, 所述装置包括: 获取模块, 用于获取待显示内容的内容信息; 确定模块,用于将所述待显示内容的内容信息输 入内容显示模型,以确定目标用户, 所述 内容显示模型是根据第一 方面所述方法构建 的目标用户特征和 目标内容特征训练 得到的; 显示模块, 用于向所述目标用户显示所述待显示内容。 第五方面, 本公开提供一种计算机可读介质, 其上存储有计算机程序, 该程序被处 理装置执行时实 现第一方面或第二方面中所述 方法的步骤。 第六方面, 本公开提供一种电子设备, 包括: 存储装置, 其上存储有计算机程序; 处理装置, 用于执行所述存储装置中的所述计算机程序, 以实现第一方面或第二方 面中所述方法 的步骤。 通过上述技术方案, 可以结合非归因转化数据构建 目标用户特征和目标内容特 征, 并将该 目标用户特征和目标内容特征用于训练 内容显示模型, 以通过更丰富的数据训练 内容显示模 型, 不仅可以提升内容显示模型的结果准确性, 减少内容显示资源的浪费, 还可以提升非 归因转化数据的数据利用率 。 本公开的其他特征和优点将在 随后的具体实施方式部分予 以详细说明。 附图说明 结合附图并参考以下具体实施 方式, 本公开各实施例的上述和其他特征、 优点及方 面将变得更加 明显。 贯穿附图中, 相同或相似的附图标记表示相同或相似的元素。 应当 理解附图是示 意性的, 原件和元素不一定按照比例绘制。 在附图中: 图 1是根据本公开一示例性实施例示 出的一种特征构建方法的流程 图; 图 2是根据本公开一示例性实施例示 出的一种内容显示方法的流程 图; 图 3是根据本公开一示例性实施例示 出的一种特征构建装置的框 图; 图 4是根据本公开一示例性实施例示 出的一种内容显示装置的框 图; 图 5是根据本公开一示例性实施例示 出的一种电子设备的框图。 具体实施方式 下面将参照附图更详细地描述 本公开的实施例。 虽然附图中显示了本公开的某些实 施例, 然而应当理解的是, 本公开可以通过各种形式来实现, 而且不应该被解释为限于 这里阐述的实施 例, 相反提供这些实施例是为了更加透彻和完整地理解本公开 。 应当理 解的是 , 本公开的附图及实施例仅用于示例性作用, 并非用于限制本公开的保护范围 。 应当理解, 本公开的方法实施方式中记载的各个步骤可以按照不 同的顺序执行, 和
/或并行执行。 此外, 方法实施方式可以包括附加的步骤和 /或省略执行示出的步骤。 本 公开的范 围在此方面不受限制。 本文使用的术语“包括”及其变形是开放性包 括, 即 “包括但不限于” 。 术语“基于” 是“至少部分地基 于”。 术语“一个实施例”表示“至少一个实施例”; 术语“另一实施例”表 示“至少一个另外 的实施例”; 术语“一些实施例”表示“至少一些实施例”。 其他术语的相 关定义将在下文 描述中给出。 需要注意, 本公开中提及的“第 “第二”等概念仅用于对不同的装置 、 模块或单 元进行区分 , 并非用于限定这些装置、 模块或单元所执行的功能的顺序或者相互依存关 系。 另外需要注意, 本公开中提及的“一个”、 “多个” 的修饰是示意性而非限制性的, 本领域技术人 员应当理解,除非在上下文另有明确指出,否则应该理解为“一个或多个”。 本公开实施方式 中的多个装置之间所交 互的消息或者信 息的名称仅用于说明性 的 目的, 而并不是用于对这些消息或信息 的范围进行限制。 归因转化数据为在内容平 台展示内容, 并将用户产生购买、 订阅、 下载等行为归因 到该内容平 台的数据。 非归因转化数据为在内容平台展示内容, 并将用户产生订阅、 下 载等行为归 因到其他内容平台的数据。相关技术通常是根 据归因转化数据分析用户行为 。 但随着各种 内容平台和应用程序的层出不 穷, 用户产生订阅、 下载等行为的原因也变得 复杂。 如果仅通过归因转化数据进行分析, 无法较好的还原用户行为轨迹, 从而影响后 续向用户显示对应 的内容, 造成内容显示资源的浪费。 发明人通过大量数据分析 发现, 相较于归因转化数据, 非归因转化数据在用户激活 行为、 用户下载行为等方面有明显提升。 因此, 本公开提供一种特征构建方法, 以结合非归因转化数据构建目标用户特征和 目标内容特 征, 并将该目标用户特征和目标内容特征用于训练内容显示模型, 以通过更 丰富的数据 训练内容显示模型, 不仅可以提升内容显示模型的结果准确性, 减少内容显 示资源的浪 费, 还可以提升非归因转化数据的数据利用率, 在广告内容场景下还可以提 升用户转化率 。 其中, 用户转化率可以理解为用户点击广告到成为有效激活用户或者注 册用户的转化率 。 首先应当理解的是, 本公开中用户转化数据的获取, 可以是先向用户显示数据获取 的授权界面 , 比如显示询问用户是否同意上传自身对应的用户数据的提示弹框等。 当用 户在该授权界面进 行数据获取的授权后, 即用户同意获取自身对应的用户数据后, 则可 以获取该用户对应 的用户转化数据进行特征构建 。 也即是说, 本公开中的用户转化数据 是在用户授权 同意的情况下获取的。 图 1是根据本公开一示例性实施例示 出的一种特征构建方法的流程 图。 参照图 1, 该特征构建方法 包括: 步骤 101, 获取目标内容对应的用户转化数据, 该用户转化数据包括非归因转化数 据。 该非归因转化数据为在第一内容平台展示有 目标内容的情况下, 当用户对目标内容 产生操作行为时 , 将操作行为归因到第二内容平台的用户数据, 该第二内容平台展示有 与目标内容相 关的内容。 步骤 102, 对用户转化数据进行分类, 以得到同一用户关联的用户转化数据和同一 内容对应的用户转化 数据。 步骤 103, 根据同一用户关联的用户转化数据, 构建目标用户特征, 并根据同一内 容对应的用户转化数 据, 构建目标内容特征, 目标用户特征和目标内容特征用于训练内 容显示模型 , 内容显示模型用于确定向目标用户显示的内容。 前文己有说明, 本公开的用户转化数据是在用户授权 同意的情况下获取的 。 因此, 在一些实施例 中, 在步骤 101之前可以先显示数据获取的授权界面, 该授权界面用于提 示用户是否允 许获取自身对应的用户数据。 相应地, 步骤 101可以是: 响应于用户在授 权界面中 的授权操作, 获取目标内容对应的用户转化数据, 该授权操作是用户触发的、 允许获取 自身对应的用户数据的操作。 比如, 向用户显示询问用户是否同意内容平台获取自身对应的用户数据的提示 弹框, 该提示弹框可 以包括显示为 “同意”和 “不同意” 的控件。 若用户通过点击、 长按等操 作选择了显示 为 “同意” 的控件, 则表明用户同意内容平台获取自身对应的用户数据 , 从而内容平 台可以获取该用户对应的用户转化数据 进行特征构建。 示例地, 用户转化数据可以是用于表征产生了订 阅、 下载等操作行为的用户数据。 在一些实施例 中, 可以获取不同时间长度的用户转化数据进行后续的特征构建, 比如可 以分别获取过去 1天、 7天和 30天的用户转化数据进行后续的特征构建。 本公开实施例中, 用户转化数据可以包括非归因转化数据。 该非归因转化数据为归 因到其他 内容平台的用户数据, 因此本内容平台获取非归因转化数据较困难, 无法直接 从本内容平 台获取到非归因转化数据。 示例地, 可以在用户授权的情况下, 从第三方数 据平台获取 目标内容对应的非归因数据, 该第三方数据平台用于收集目标内容对应的非 归因转化数据 。 应当理解的是, 从第三方数据平台获取的非归因转化数据较杂乱 , 为了便于后续特 征构建, 可以先对用户转化数据进行分类, 以得到同一用户关联的用户转化数据和同一 内容对应的用户转化 数据。 示例地, 可以通过用户转化数据包括的用户设备信息确定同 一用户关联的用户转 化数据, 同时通过用户转化数据包括的内容标识信息确定同一内容 对应的用户转化数据 。 在一些实施例中, 用户转化数据还可以包括归因转化数据, 该归因转化数据为在第 一内容平 台展示有目标内容的情况下, 当用户对目标内容产生操作行为时, 将操作行为 归因到第一 内容平台的用户数据, 还可以先将同一用户关联的归因转化数据与非归因转 化数据进行关联 , 以得到目标用户转化数据, 然后对目标用户转化数据进行分类, 以得 到同一用户关联 的用户转化数据和同一内容对应 的用户转化数据。 示例地, 可以在用户授权的情况下, 从第一内容平台获取目标内容对应的归因转化 数据, 然后结合目标内容对应的非归因转化 数据共同构建特征 。 其中, 上文己有说明, 从第三数据平 台获取的非归因转化数据较杂乱 , 因此为了便于后续特征构建, 可以先将 同一用户对应 的非归因转化数据与归因转化数据进行 关联, 以归类同一用户对应的非归 因转化数据和 归因转化数据, 得到目标用户转化数据。 在一些实施例中, 考虑到同一用户使用的用户设备信息通常是相 同的, 因此可以通 过用户设备信息将 同一用户对应的非归因转化数据与 归因转化数据进行关联, 以得到目 标用户转化数据 。 当然, 在其他可能的方式中, 也可以通过其他用户信息将同一用户对 应的非归因转化数 据与归因转化数据进行关联, 以得到目标用户转化数据, 本公开实施 例对此不作 限定。 通过上述关联处理后, 可以将同一用户对应的非归因转化数据和归因转化数据进 行 关联, 但是关联后的数据中还可能存在同一用户对应多个数 据, 因此为了便于后续特征 构建, 还可以进一步对目标用户转化数据进行分类, 以得到同一用户关联的用户转化数 据和同一 内容对应的用户转化数据。 比如, 可以通过用户设备信息对目标用户转化数据 进行分类, 以得到同一用户关联的用户转化数据, 同时可以通过内容标识信息对目标用 户转化数据进行 分类, 以得到同一内容对应的用户转化数据。 由此, 在用户授权的情况下, 可以结合归因转化数据和非归因转化数据进行后续的 特征构建 , 相较于仅通过归因转化数据进行特征构建的方式 , 可以得到更丰富的特征, 从而通过更 丰富的数据训练内容显示模型 , 提升内容显示模型的结果准确性, 减少内容 显示资源的浪 费, 在广告场景下还可以提升用户转化率。 在得到同一用户关联的用户转化 数据和同一内容对应的用户转化 数据之后, 可以根 据同一用户 关联的用户转化数据, 构建目标用户特征, 并根据同一内容对应的用户转化 数据, 构建目标内容特征。 在一些实施例中,根据同一用户关联的用户转化数据 ,构建目标用户特征,可以是: 根据同一用 户关联的非归因转化数据, 构建以下至少一种目标用户特征: 列表类用户特 征、 数值类用户特征和新近程度特征。 示例地, 列表类用户特征可以用于表征同一用户在产生操作行为之前访 问过的内容 平台特征或 内容特征, 数值类用户特征可以用于表征同一用户产生的操作行为的数量 特 征, 新近程度特征可以用于表征最近一次的操作 行为的产生时间与当前时间之间的时 间 间隔特征 。 示例地, 表 2示出了根据同一用户关联的非归因转化数据构建的可能的 目标用户特 征, 包括列表类用户特征、 数值类用户特征和新近程度特征。 表 2
Figure imgf000008_0001
应当理解的是, 表 2所示的目标用户特征仅作示意, 在本公开具体实施时, 可以根 据实际需求构 建更多的目标用户特征。 例如, 在用户授权的情况下, 若用户购买了物品 A, 则通过对非归因转化数据构建 列表类用户特 征, 可以确定用户在购买物品 A之前访问过哪些内容平台, 或者浏览过哪 些物品信 息, 即可以确定用户对哪些内容感兴趣。 通过对非归因转化数据构建数值类用 户特征,可以确定用户购买的物 品 A的数量,即可以确定用户对物品 A的感兴趣程度。 通过对非归 因转化数据构建新近程度特征 ,可以确定用户本次购买物品 A的时间与上一 次购买物品 A的时间间隔, 即可以确定用户最近对哪些内容感兴趣。 由此, 在用户授权 的情况下 , 通过构建不同类型的用户特征, 可以较好还原用户购买物品 A的行为轨迹, 从而后续可 以更准确的向用户显示对应的 内容, 减少内容显示资源的浪费。 在一些实施例中, 根据同一内容对应的用户转化数 据, 构建目标内容特征可以是: 根据同一 内容对应的非归因转化数据,构建以下 目标内容特征: 列表类内容特征和 /或数 值类内容特征 。 其中, 列表类内容特征用于表征针对同一内容产生业务操作行为的用户 特征, 数值类内容特征用于表征同一 内容对应的操作行为的数量特征 。 示例地, 表 3示出了根据同一内容对应的非归因转化数据构建的可能的 目标内容特 征, 包括列表类内容特征和数值类内容特征 。 表 3
Figure imgf000009_0001
应当理解的是, 表 3所示的目标内容特征仅作示意, 在本公开具体实施时, 可以根 据实际需求构 建更多的目标内容特征。 通过上述方式, 在用户授权的情况下, 可以根据非归因转化数据中的内容侧数据构 建 目标内容特征, 从而构建不同类型的内容特征, 更好地还原用户针对目标内容的行为 轨迹, 从而可以更准确地向用户显示对应的 内容, 减少内容显示资源的浪费。 在一些实施例中,根据同一用户关联的用户转化数 据,构建目标用户特征还可以是: 根据同一 用户关联的非归因转化数据 , 确定同一用户产生过操作行为的多个内容数据, 并根据多个 内容数据,确定该用户对应的共现内容特征,以得到目标用户特征。同样地, 根据同一 内容对应的用户转化数据, 构建目标内容特征还可以是: 根据同一内容对应的 非归因转化数据 , 确定对同一内容产生过操作行为的多个用户标识数据, 并根据多个用 户标识数据 , 确定该内容对应的共现用户特征, 以得到目标内容特征。 例如, 在用户授权的情况下, 获取到某一用户对内容平台上的广告 1和广告 2产生 过点击行为 , 则可以根据广告 1和广告 2对应的内容数据确定该用户对应的共现内容特 征, 比如先从广告 1和广告 2对应的内容数据中提取内容信息特 征, 然后将广告 1和广 告 2的内容信息特征的总和作为共现内容特征, 或者将广告 1和广告 2的相同内容信息 特征作为共现 内容特征等, 以得到该用户对应的目标用户特征。 或者, 在用户授权的情 况下, 获取到第一用户和第二用户均点击过广 告 3, 则可以根据第一用户和第二用户的 用户标识数据确 定广告 3对应的共现用户特征, 比如先从第一用户和第二用户的用户标 识数据中提取用 户信息特征, 然后将第一用户和第二用户的用户信息特征的总和作为共 现用户特征, 或者将第一用户和第二用户的相同用户信 息特征作为共现用户特征等 , 以 得到广告 3对应的目标内容特征。 通过上述方式, 在用户授权的情况下, 可以根据非归因转化数据中同一用户产生过 操作行为的多个 内容数据和同一内容关联 的多个用户标识数据构建共现特征 , 以从多用 户或多内容 的维度构建用于训练内容显示模型 的特征, 提升内容显示模型的结果准确性, 从而更准确地 向用户显示对应的内容, 减少内容显示资源的浪费, 并进一步提升非归因 转化数据的数据 利用率, 在广告内容场景下还可以提升用户转化率。 基于同一发明构思, 本公开还提供一种内容显示方法, 包括: 步骤 201, 获取待显示内容的内容信息; 步骤 202, 将待显示内容的内容信息输入内容显示模型, 以确定目标用户, 该内容 显示模型 是根据本公开提供 的任一特征构建方法 构建的目标用户特征和 目标内容特征 训练得到的; 步骤 203, 向目标用户显示待显示内容。 示例地, 内容信息用于表征内容页面的文字、 图片等基本内容, 在确定目标内容的 情况下, 可以获取该目标内容的内容信息, 然后将该内容信息输入预训练的内容显示模 型中, 以得到待显示该目标内容的目标用户。 应当理解的是, 相关技术中, 在初始化内容显示模型参数后, 通常是将归因转化数 据中内容 的内容信息特征输入内容显示模 型进行预估, 得到预估用户, 再将该预估用户 与归因转化数据 中该内容关联的用户做比较计算损失 函数。然后根据该损失函数的计算 结果进行反 向传播, 以更新模型参数。 并且, 会重复执行将归因转换数据中的内容信息 特征输入模型进 行预估, 得到预估用户、 将该预估用户与归因转化数据中该内容关联的 用户做比较计算损 失函数、 以及根据该损失函数的计算结果进行反向传播, 以更新模型 参数的过程 , 直到损失函数不再有显著的降低。 之后, 在模型应用阶段, 可以向模型输 入目标内容 的内容信息, 得到待显示目标内容的目标用户。 但是, 前文己有说明, 此种方式仅通过归因转化数据进行分析, 无法较好的还原用 户行为轨迹, 从而影响最终确定的目标用户的准确性 , 无法较好的实现目标内容的推送 显示, 造成内容显示资源的浪费。 因此, 本公开提出一种新的内容显示方式, 可以在用户授权的情况下, 结合非归因 转化数据构建 目标用户特征和目标内容特征 , 并将该目标用户特征和目标内容特征用于 训练内容显示模 型, 以通过更丰富的数据训练内容显示模型, 不仅可以提升内容显示模 型的结果准确 性,减少内容显示资源的浪费,还可以提升非归因转化数据的数据利用 率, 在广告 内容场景下还可以提升用户转化率 。 其中, 目标用户特征和目标内容特征的相关 内容己在上文 进行说明, 这里不再赘述。 具体地, 经测试, 相较于仅通过归因转化数据 训练的内容显示模 型,本公开实施例中结合非归因转化数据训练的内容显示模型 的 A U C ( area under the curve ) 可以提升 0.2%, 在广告内容场景下, 可以更准确的向用户推送 广告, 从而提升用户转化率。 基于同一发明构思, 本公开还提供一种特征构建装置, 该装置可以通过软件、 硬件 或者两者结合 的方式成为电子设备的部分或全部 。 参照图 3, 该特征构建装置 300, 包 括: 数据获取模块 301, 用于获取目标内容对应的用户转化数据, 所述用户转化数据包 括非归因转化数 据, 所述非归因转化数据为在第一内容平台展示有所述目标内容的情况 下, 当用户对所述目标内容产生操作行为时, 将所述操作行为归因到第二内容平台的用 户数据, 所述第二内容平台展示有与所述 目标内容相关的内容; 数据分类模块 302, 用于对所述用户转化数据进行分类, 以得到同一用户关联的所 述用户转化数据 和同一内容对应的所述用户转化数 据; 特征构建模块 303, 用于根据同一用户关联的所述用户转化数据, 构建目标用户特 征, 并根据同一内容对应的所述用户转化数据 , 构建目标内容特征; 所述目标用户特征和 目标内容特征用于训练内容显示模型, 所述内容显示模型用于 确定向 目标用户显示的内容。 在一些实施例中, 所述装置 300还包括: 显示模块, 用于在获取目标内容对应的用户转化数据之前, 显示数据获取的授权界 面, 所述授权界面用于提示用户是否允许获取 自身对应的用户数据; 所述数据获取模块用于响应于用 户在所述授权界面中的授权操 作, 获取目标内容对 应的用户转化数据 ,所述授权操作是用户触发的、允许获取自身对应的用户数据的操作 。 在一些实施例中, 所述特征构建模块 303用于: 根据同一用户关联的所述非归因转 化数据, 构建以下至少一种目标用户特征: 列表类用户特征、 数值类用户特征和新近程 度特征。 在一些实施例中, 所述特征构建模块 303用于: 根据同一内容对应的所述非归因转 化数据, 构建以下目标内容特征: 列表类内容特征和 /或数值类内容特征, 其中, 所述列 表类内容特征用 于表征针对同一内容产生业务操作 行为的用户特征, 所述数值类内容特 征用于表征 同一内容对应的所述操作行为 的数量特征。 在一些实施例中, 所述特征构建模块 303用于: 根据同一用户关联的所述非归因转 化数据,确定同一用户产生过所述操 作行为的多个内容数据,并根据所述多个 内容数据, 确定所述用户对 应的共现内容特征, 以得到所述目标用户特征; 根据同一内容对应的所 述用户转化数据 ,构建目标内容特征,包括:根据同一内容对应的所述非归因转化数据, 确定对同一 内容产生过所述操作行为的多个用 户标识数据, 并根据所述多个用户标识数 据, 确定所述内容对应的共现用户特征, 以得到所述目标内容特征。 在一些实施例中, 所述用户转化数据还包括归因转化数据, 所述归因转化数据为在 所述第一 内容平台展示有所述目标内容 的情况下, 当用户对所述目标内容产生操作行为 时, 将所述操作行为归因到所述第一内容平 台的用户数据, 所述装置 300还包括: 关联模块, 用于将同一用户对应的所述归因转化数据与所述非归因转化数据进行 关 联, 以得到目标用户转化数据; 所述数据分类模块 302用于: 对所述目标用户转化数据进行分类, 以得到同一用户 关联的所述用户转化 数据和同一内容对应的所述用 户转化数据。 基于同一发明构思, 本公开还提供一种内容显示装置, 该装置可以通过软件、 硬件 或者两者结合 的方式成为电子设备的部分或全部 。 参照图 4, 该内容显示装置 400, 包 括: 获取模块 401, 用于获取待显示内容的内容信息; 确定模块 402, 用于将所述待显示内容的内容信息输入内容显示模型, 以确定目标 用户, 所述内容显示模型是根据第一方面所述方法构 建的目标用户特征和目标内容特征 训练得到的; 显示模块 403, 用于向所述目标用户显示给所述待显示内容。 应当理解的是, 在一些实施例中, 电子设备可以包括如图 3所示的特征构建装置和 如图 4所示的内容显示装置。 其中, 通过特征构建装置可以构建目标用户特征和目标内 容特征, 用于内容显示装置中内容显示模 型的训练。 关于上述实施例中的装置 , 其中各个模块执行操作的具体方式己经在有关该方法的 实施例中进行 了详细描述, 此处将不做详细阐述说明。 基于同一发明构思, 本公开还提供一种计算机可读介 质, 其上存储有计算机程序, 该程序被处理 装置执行时实现上述任一特 征构建方法或上述任一 内容显示方法的步骤。 基于同一发明构思, 本公开还提供一种电子设备, 包括: 存储装置, 其上存储有计算机程序; 处理装置, 用于执行所述存储装置中的所述计算机程序, 以实现上述任一特征构建 方法或上述 任一内容显示方法的步骤。 下面参考图 5,其示出了适于用来实现本公开实施例的 电子设备 500的结构示意图。 本公开实施例 中的终端设备可以包括但不限于诸 如移动电话、 笔记本电脑、 数字广播接 收器、 PDA (个人数字助理) 、 PAD (平板电脑) 、 PMP (便携式多媒体播放器) 、 车 载终端 (例如车载导航终端) 等等的移动终端以及诸如数字 TV、 台式计算机等等的固 定终端。 图 5示出的电子设备仅仅是一个示例, 不应对本公开实施例的功能和使用范围 带来任何限制 。 如图 5所示, 电子设备 500可以包括处理装置 (例如中央处理器、 图形处理器等) 501, 其可以根据存储在只读存储器 (ROM) 502 中的程序或者从存储装置 508加载到 随机访问存储器 (RAM) 503中的程序而执行各种适当的动作和处理。在 RAM 503中, 还存储有电子设备 500操作所需的各种程序和数据。处理装置 501、 ROM 502以及 RAM 503通过总线 504彼此相连。 输入 /输出 (I/O) 接口 505也连接至总线 504。 通常, 以下装置可以连接至 I/O接口 505: 包括例如触摸屏、 触摸板、 键盘、 鼠标、 摄像头、 麦克风、 加速度计、 陀螺仪等的输入装置 506; 包括例如液晶显示器(LCD)、 扬声器、 振动器等的输出装置 507; 包括例如磁带、 硬盘等的存储装置 508; 以及通信 装置 509。 通信装置 509可以允许电子设备 500与其他设备进行无线或有线通信以交换 数据。 虽然图 5示出了具有各种装置的电子设备 500, 但是应理解的是, 并不要求实施 或具备所有示 出的装置。 可以替代地实施或具备更多或更少的装置。 特别地, 根据本公开的实施例, 上文参考流程图描述的过程可以被实现为计算机软 件程序 。 例如, 本公开的实施例包括一种计算机程序产品, 其包括承载在非暂态计算机 可读介质上 的计算机程序, 该计算机程序包含用于执行流程 图所示的方法的程序代码。 在这样的实施例 中, 该计算机程序可以通过通信装置 509从网络上被下载和安装, 或者 从存储装置 508被安装, 或者从 ROM 502被安装。 在该计算机程序被处理装置 501执 行时, 执行本公开实施例的方法中限定的上述 功能。 需要说明的是, 本公开上述的计算机可读介质可以是计算机可读信号介质或 者计算 机可读存储介质 或者是上述两者的任意组合 。 计算机可读存储介质例如可以是一一但不 限于一一 电、 磁、 光、 电磁、 红外线、 或半导体的系统、 装置或器件, 或者任意以上的 组合。 计算机可读存储介质的更具体的例子可以包括 但不限于: 具有一个或多个导线的 电连接、 便携式计算机磁盘、 硬盘、 随机访问存储器(RAM) 、 只读存储器(ROM) 、 可擦式可编程 只读存储器 (EPROM或闪存 ) 、 光纤、 便携式紧凑磁盘只读存储器(CD- ROM ) 、 光存储器件、 磁存储器件、 或者上述的任意合适的组合。 在本公开中, 计算机 可读存储介质可 以是任何包含或存储程序 的有形介质, 该程序可以被指令执行系统、 装 置或者器件使 用或者与其结合使用。 而在本公开中, 计算机可读信号介质可以包括在基 带中或者作为载波 一部分传播的数据信号, 其中承载了计算机可读的程序代码。 这种传 播的数据信 号可以采用多种形式, 包括但不限于电磁信号、 光信号或上述的任意合适的 组合。 计算机可读信号介质还可以是计算机 可读存储介质以外的任何 计算机可读介质, 该计算机可读信 号介质可以发送、 传播或者传输用于由指令执行系统、 装置或者器件使 用或者与其 结合使用的程序。计算机可读介质上包含 的程序代码可以用任何适当的介质 传输, 包括但不限于: 电线、 光缆、 RF (射频) 等等, 或者上述的任意合适的组合。 在一些实施方式中, 可以利用诸如 HTTP(HyperText Transfer Protocol, 超文本传输 协议)之类的任何 当前己知或未来研发的网络协议进行通 信, 并且可以与任意形式或介 质的数字数 据通信 (例如, 通信网络) 互连。 通信网络的示例包括局域网 (“LAN”) , 广域网 (“WAN”) , 网际网 (例如, 互联网) 以及端对端网络 (例如, ad hoc端对端网 络) , 以及任何当前己知或未来研发的网络。 上述计算机可读介质可以是上述 电子设备中所包含的; 也可以是单独存在, 而未装 配入该电子设备 中。 上述计算机可读介质承载有一个 或者多个程序, 当上述一个或者多个程序被该电子 设备执行时, 使得该电子设备: 获取目标内容对应的用户转化数据, 所述用户转化数据 包括非归因转化 数据, 所述非归因转化数据为在第一内容平台展示有所述目标内容 的情 况下, 当用户对所述目标内容产生操作行为时, 将所述操作行为归因到第二内容平台的 用户数据, 所述第二内容平台展示有与所述 目标内容相关的内容; 对所述用户转化数据 进行分类, 以得到同一用户关联的所述用户转化数据和同一 内容对应的所述用户转化数 据; 根据同一用户关联的所述用户转化数据, 构建目标用户特征, 并根据同一内容对应 的所述用户转化 数据, 构建目标内容特征; 所述目标用户特征和所述目标内容特征用于 训练内容显示模 型, 所述内容显示模型用于确定向目标用户显示的内容。 可以以一种或多种程 序设计语言或其组合 来编写用于执行本公 开的操作的计算机 程序代 码, 上述程序设计语言包括但 不限于面向对象 的程序设计语言一 诸如 Java、 Smalltalk、 C++, 还包括常规的过程式程序设计语言一一诸如 “C”语言或类似的程序设计 语言。 程序代码可以完全地在用户计算机上执行、 部分地在用户计算机上执行、 作为一 个独立的软件包 执行、 部分在用户计算机上部分在远程计算机上执行、 或者完全在远程 计算机或服务器 上执行。 在涉及远程计算机的情形中, 远程计算机可以通过任意种类的 网络一一包括局域 网 (LAN) 或广域网 (WAN) —一连接到用户计算机, 或者, 可以连 接到外部计算机 (例如利用因特网服务提供商来通过因特网连接) 。 附图中的流程图和框图, 图示了按照本公开各种实施例的系统、 方法和计算机程序 产品的可能实现 的体系架构、 功能和操作。 在这点上, 流程图或框图中的每个方框可以 代表一个模 块、 程序段、 或代码的一部分, 该模块、 程序段、 或代码的一部分包含一个 或多个用于 实现规定的逻辑功能的可执行指令 。也应当注意,在有些作为替换的实现中, 方框中所标注 的功能也可以以不同于附 图中所标注的顺序发生。 例如, 两个接连地表示 的方框实际上可 以基本并行地执行, 它们有时也可以按相反的顺序执行, 这依所涉及的 功能而定 。 也要注意的是, 框图和 /或流程图中的每个方框、 以及框图和 /或流程图中的 方框的组合 , 可以用执行规定的功能或操作的专用的基于硬件的系统来实现, 或者可以 用专用硬件与 计算机指令的组合来实现。 描述于本公开实施例中所涉 及到的模块可以通过软件的方式 实现, 也可以通过硬件 的方式来实现 。 其中, 模块的名称在某种情况下并不构成对该模块本身的限定。 本文中以上描述的功 能可以至少部分地由一个 或多个硬件逻辑部件来执 行。 例如, 非限制性地 , 可以使用的示范类型的硬件逻辑部件包括: 现场可编程门阵列(FPGA)、 专用集成 电路(ASIC) 、 专用标准产品 (ASSP) 、 片上系统(SOC) 、 复杂可编程逻辑 设备 (CPLD) 等等。 在本公开的上下文中, 机器可读介质可以是有形的介质, 其可以包含或存储以供指 令执行系统、 装置或设备使用或与指令执行系统、 装置或设备结合地使用的程序。 机器 可读介质可 以是机器可读信号介质或机器可读储存 介质。机器可读介质可以包括但不限 于电子的、 磁性的、 光学的、 电磁的、 红外的、 或半导体系统、 装置或设备, 或者上述 内容的任何 合适组合。机器可读存储介质的更具体示例 会包括基于一个或多个线 的电气 连接、 便携式计算机盘、 硬盘、 随机存取存储器 (RAM) 、 只读存储器 (ROM) 、 可擦 除可编程只读存 储器(EPROM 或快闪存储器 ) 、 光纤、 便捷式紧凑盘只读存储器(CD- ROM ) 、 光学储存设备、 磁储存设备、 或上述内容的任何合适组合。 根据本公开的一个或多个 实施例,示例 1提供了一种特征构建方法,所述方法包括: 获取目标内容对应的用户转化数据 , 所述用户转化数据包括非归因转化数据, 所述 非归因转化数据 为在第一内容平台展示有所述 目标内容的情况下, 当用户对所述目标内 容产生操作行 为时, 将所述操作行为归因到第二内容平台的用户数据, 所述第二内容平 台展示有与 所述目标内容相关的内容; 对所述用户转化数据进行分类 , 以得到同一用户关联的所述用户转化数据和同一内 容对应的所述用 户转化数据; 根据同一用户关联的所述用户转 化数据, 构建目标用户特征, 并根据同一内容对应 的所述用户转化 数据, 构建目标内容特征; 所述目标用户特征和所述 目标内容特征用于训练内容显示模 型, 所述内容显示模型 用于确定 向目标用户显示的内容。 根据本公开的一个或多个实施 例, 示例 2提供了示例 1的方法, 在获取目标内容对 应的用户转化数据 之前, 所述方法还包括: 显示数据获取的授权界面, 所述授权界面用于提示用户是否允许获取 自身对应的用 户数据; 所述获取目标内容对应的用户转化数 据, 包括: 响应于用户在所述授权界面 中的授权操作, 获取目标内容对应的用户转化数据, 所 述授权操作是用 户触发的、 允许获取自身对应的用户数据的操作。 根据本公开的一个或多个实施 例, 示例 3提供了示例 1的方法, 根据同一用户关联 的所述用户转化 数据, 构建目标用户特征, 包括: 根据同一用户关联的所述非归 因转化数据, 构建以下至少一种目标用户特征: 列表 类用户特征、 数值类用户特征和新近程度特征 。 根据本公开的一个或多个实施 例, 示例 4提供了示例 1-3任一项的方法, 根据同一 内容对应的所述 用户转化数据, 构建目标内容特征, 包括: 根据同一内容对应的所述非 归因转化数据, 构建以下目标内容特征: 列表类内容特 征和 /或数值类内容特征, 其中,所述列表类内容特征用于表征针对同一内容产生业务操 作行为的用户特 征, 所述数值类内容特征用于表征同一内容对应的所述操作行为的数量 特征。 根据本公开的一个或多个实施 例, 示例 5提供了示例 1-3任一项的方法, 根据同一 用户关联的所述用 户转化数据, 构建目标用户特征, 包括: 根据同一用户关联的所述非归 因转化数据, 确定同一用户产生过所述操作行为的多 个内容数据 , 并根据所述多个内容数据, 确定所述用户对应的共现内容特征, 以得到所 述目标用户特征 ; 根据同一内容对应的所述用户转 化数据, 构建目标内容特征, 包括: 根据同一内容对应的所述非 归因转化数据, 确定对同一内容产生过所述操作行为的 多个用户标识数据 ,并根据所述多个用户标识数据,确定所述内容对应的共现用户特征 , 以得到所述 目标内容特征。 根据本公开的一个或多个实施 例, 示例 6提供了示例 1-3任一项的方法, 所述用户 转化数据还包括 归因转化数据, 所述归因转化数据为在所述第一内容平台展示有所述 目 标内容的情况 下, 当用户对所述目标内容产生操作行为时, 将所述操作行为归因到所述 第一内容平 台的用户数据, 所述方法还包括: 将同一用户对应的所述归因转化数 据与所述非归因转化数据进行 关联, 以得到目标 用户转化数据 ; 对所述用户转化数据进行分类 , 以得到同一用户关联的所述用户转化数据和同一内 容对应的所述用 户转化数据, 包括: 对所述目标用户转化数据进行分类 , 以得到同一用户关联的所述用户转化数据和同 一内容对应的所 述用户转化数据。 根据本公开的一个或多个实施 例,示例 7提供了一种内容显示方法,所述方法包括: 获取待显示内容的内容信息 ; 将所述待显示内容的内容信 息输入内容显示模型, 以确定目标用户, 所述内容显示 模型是根据 示例 1所述方法构建的目标用户特征和 目标内容特征训练得到的; 向所述目标用户显示所述待显示 内容。 根据本公开的一个或多个实施 例,示例 8提供了一种特征构建装置,所述装置包括: 数据获取模块, 用于获取目标内容对应的用户转化数据, 所述用户转化数据包括非 归因转化数据 , 所述非归因转化数据为在第一内容平台展示有所 述目标内容的情况下, 当用户对所述 目标内容产生操作行为时, 将所述操作行为归因到第二内容平台的用户数 据, 所述第二内容平台展示有与所述 目标内容相关的内容; 数据分类模块, 用于对所述用户转化数据进行分类, 以得到同一用户关联的所述用 户转化数据和 同一内容对应的所述用户转化数据 ; 特征构建模块, 用于根据同一用户关联的所述用户转 化数据, 构建目标用户特征, 并根据同一 内容对应的所述用户转化数据, 构建目标内容特征; 所述目标用户特征和 目标内容特征用于训练内容显示模型, 所述内容显示模型用于 确定向 目标用户显示的内容。 根据本公开的一个或多个实施 例, 示例 9提供了示例 8的装置, 所述装置还包括: 显示模块, 用于在获取目标内容对应的用户转化数据之前, 显示数据获取的授权界 面, 所述授权界面用于提示用户是否允许获取 自身对应的用户数据; 所述数据获取模块用于响应于用 户在所述授权界面中的授权操 作, 获取目标内容对 应的用户转化数据 ,所述授权操作是用户触发的、允许获取自身对应的用户数据的操作 。 根据本公开的一个或多个实施 例, 示例 10提供了示例 8的装置, 所述特征构建模 块用于: 根据同一用户关联的所述非归 因转化数据, 构建以下至少一种目标用户特征: 列表 类用户特征、 数值类用户特征和新近程度特征 。 根据本公开的一个或多个实施 例, 示例 11提供了示例 8-10任一项的装置, , 所述 特征构建模块用 于: 根据同一内容对应的所述非 归因转化数据, 构建以下目标内容特征: 列表类内容特 征和 /或数值类内容特征, 其中,所述列表类内容特征用于表征针对同一内容产生业务操 作行为的用户特 征, 所述数值类内容特征用于表征同一内容对应的所述操作行为的数量 特征。 根据本公开的一个或多个实施 例, 示例 12提供了示例 8-10任一项的装置, 所述特 征构建模块用 于: 根据同一用户关联的所述非归 因转化数据, 确定同一用户产生过所述操作行为的多 个内容数据 , 并根据所述多个内容数据, 确定所述用户对应的共现内容特征, 以得到所 述目标用户特征 ; 根据同一内容对应的所述用户转 化数据, 构建目标内容特征, 包括: 根据同一内容对应的所述非 归因转化数据, 确定对同一内容产生过所述操作行为的 多个用户标识数据 ,并根据所述多个用户标识数据,确定所述内容对应的共现用户特征 , 以得到所述 目标内容特征。 根据本公开的一个或多个实施 例, 示例 13提供了示例 8-10任一项的装置, 所述用 户转化数据还包括 归因转化数据, 所述归因转化数据为在所述第一内容平台展示有所述 目标内容的情况 下, 当用户对所述目标内容产生操作行为时, 将所述操作行为归因到所 述第一 内容平台的用户数据, 所述装置还包括: 关联模块, 用于将同一用户对应的所述归因转化数据与所述非归因转化数据进行 关 联, 以得到目标用户转化数据; 所述数据分类模块用于: 对所述目标用户转化数据进行分类 , 以得到同一用户关联的所述用户转化数据和同 一内容对应的所 述用户转化数据。 根据本公开的一个或多个实施 例, 示例 14提供了一种内容显示装置, 所述装置包 括: 获取模块, 用于获取待显示内容的内容信息; 确定模块,用于将所述待显示内容的内容信息输 入内容显示模型,以确定目标用户, 所述内容 显示模型是根据示例 1 所述方法构建的目标用户特征和目标 内容特征训练得 到的; 显示模块, 用于向所述目标用户显示给所述待显示内容。 根据本公开的一个或多个实施 例, 示例 15 提供了一种计算机可读介质, 其上存储 有计算机程序 , 该程序被处理装置执行时实现示例 1-7中任一项所述方法的步骤。 根据本公开的一个或多个实施 例, 示例 16提供了一种电子设备, 包括: 存储装置, 其上存储有计算机程序; 处理装置, 用于执行所述存储装置中的所述计算机程序, 以实现示例 1-7中任一项 所述方法 的步骤。 以上描述仅为本公开的较佳实施例 以及对所运用技术原理的说 明。本领域技术人员 应当理解, 本公开中所涉及的公开范围, 并不限于上述技术特征的特定组合而成的技术 方案, 同时也应涵盖在不脱离上述公开构思的情况 下, 由上述技术特征或其等同特征进 行任意组合而 形成的其它技术方案。 例如上述特征与本公开中公开的 (但不限于) 具有 类似功能的技 术特征进行互相替换而形成的技 术方案。 此外, 虽然采用特定次序描绘了各操作, 但是这不应当理解为要求这些操作以所示 出的特定次序 或以顺序次序执行来执行。 在一定环境下, 多任务和并行处理可能是有利 的。 同样地, 虽然在上面论述中包含了若干具体实现细节, 但是这些不应当被解释为对 本公开的范 围的限制。在单独的实施例的上下文中描述 的某些特征还可以组合地实现在 单个实施例 中。 相反地, 在单个实施例的上下文中描述的各种特征也可以单独地或以任 何合适的子 组合的方式实现在多个实施例 中。 尽管己经采用特定于 结构特征和 /或方法逻辑动作的语言描述了本主题, 但是应当 理解所附权利 要求书中所限定的主题未必局 限于上面描述的特定特征或动作 。 相反, 上 面所描述 的特定特征和动作仅仅是实现权利要求书 的示例形式。关于上述实施例中的装 置, 其中各个模块执行操作的具体方 式己经在有关该方法的实施例 中进行了详细描述, 此处将不做详细 阐述说明。

Claims

权利要求书
1、 一种特征构建方法, 其包括: 获取目标内容对应的用户转化数据 , 所述用户转化数据包括非归因转化数据, 所述 非归因转化数据 为在第一内容平台展示有所述 目标内容的情况下, 当用户对所述目标内 容产生操作行 为时, 将所述操作行为归因到第二内容平台的用户数据, 所述第二内容平 台展示有与所 述目标内容相关的内容; 对所述用户转化数据进行分类 , 以得到同一用户关联的所述用户转化数据和同一内 容对应的所述用 户转化数据; 根据同一用户关联的所述用户转化 数据, 构建目标用户特征, 并根据同一内容对应 的所述用户转化 数据, 构建目标内容特征; 所述目标用户特征和所述 目标内容特征用于训练内容显示模 型, 所述内容显示模型 用于确定 向目标用户显示的内容。
2、 根据权利要求 1 所述的方法, 其中, 在所述获取目标内容对应的用户转化数据 之前, 所述方法还包括: 显示数据获取的授权界面, 所述授权界面用于提示用户是否允许获取 自身对应的用 户数据; 所述获取目标内容对应的用户转化 数据, 包括: 响应于用户在所述授权界面 中的授权操作, 获取目标内容对应的用户转化数据, 所 述授权操作是用 户触发的、 允许获取自身对应的用户数据的操作。
3、 根据权利要求 1 所述的方法, 其中, 所述根据同一用户关联的所述用户转化数 据, 构建目标用户特征, 包括: 根据同一用户关联的所述非归 因转化数据, 构建以下至少一种目标用户特征: 列表 类用户特征、 数值类用户特征和新近程度特征 。
4、根据权利要求 1-3任一项所述的方法, 其中, 所述根据同一内容对应的所述用户 转化数据, 构建目标内容特征, 包括: 根据同一内容对应的所述非 归因转化数据, 构建以下目标内容特征: 列表类内容特 征和 /或数值类内容特征, 其中,所述列表类内容特征用于表征针对同一内容产生业务操 作行为的用户特 征, 所述数值类内容特征用于表征同一内容对应的所述操作行为的数量 特征。
5、根据权利要求 1-3任一项所述的方法, 其中, 所述根据同一用户关联的所述用户 转化数据, 构建目标用户特征, 包括: 根据同一用户关联的所述非归 因转化数据, 确定同一用户产生过所述操作行为的多 个内容数据 , 并根据所述多个内容数据, 确定所述用户对应的共现内容特征, 以得到所 述目标用户特征 ; 所述根据同一内容对应的所述 用户转化数据, 构建目标内容特征, 包括: 根据同一内容对应的所述非 归因转化数据, 确定对同一内容产生过所述操作行为的 多个用户标识数据 ,并根据所述多个用户标识数据,确定所述内容对应的共现用户特征 , 以得到所述 目标内容特征。
6、根据权利要求 1-3任一项所述的方法, 其中, 所述用户转化数据还包括归因转化 数据, 所述归因转化数据为在所述第一内容平台展 示有所述目标内容的情况下 , 当用户 对所述 目标内容产生操作行为时 , 将所述操作行为归因到所述第一内容平 台的用户数 据, 所述方法还包括: 将同一用户对应的所述归因转化数 据与所述非归因转化数据进行 关联, 以得到目标 用户转化数据 ; 所述对所述用户转化数据进行分 类, 以得到同一用户关联的所述用户转化数据和同 一内容对应的所 述用户转化数据, 包括: 对所述目标用户转化数据进行分类 , 以得到同一用户关联的所述用户转化数据和同 一内容对应的所 述用户转化数据。
7、 一种内容显示方法, 其包括: 获取待显示内容的内容信息 ; 将所述待显示内容的内容信 息输入内容显示模型, 以确定目标用户, 所述内容显示 模型是根据权利 要求 1-6任一项所述方法构建的目标用户特征和 目标内容特征训练得到 的; 向所述目标用户显示所述待显示 内容。
8、 一种特征构建装置, 其包括: 数据获取模块, 用于获取目标内容对应的用户转化数据, 所述用户转化数据包括非 归因转化数据 , 所述非归因转化数据为在第一内容平台展示有所述 目标内容的情况下, 当用户对所述 目标内容产生操作行为时, 将所述操作行为归因到第二内容平台的用户数 据, 所述第二内容平台展示有与所述 目标内容相关的内容; 数据分类模块, 用于对所述用户转化数据进行分类, 以得到同一用户关联的所述用 户转化数据和 同一内容对应的所述用户转化数据 ; 特征构建模块, 用于根据同一用户关联的所述用户转化数据 , 构建目标用户特征, 并根据同一 内容对应的所述用户转化数据, 构建目标内容特征; 所述目标用户特征和 目标内容特征用于训练内容显示模型, 所述内容显示模型用于 确定向 目标用户显示的内容。
9、 一种内容显示装置, 其包括: 获取模块, 用于获取待显示内容的内容信息; 确定模块,用于将所述待显示内容的内容信息输 入内容显示模型,以确定目标用户, 所述内容显示模 型是根据权利要求 1-6任一项所述方法构建的目标用户特征和目标内容 特征训练得到 的; 显示模块, 用于向所述目标用户显示所述待显示内容。
10、 一种计算机可读介质, 其上存储有计算机程序, 其中, 该程序被处理装置执行 时实现权利要求 1-7中任一项所述方法的步骤。
11、 一种电子设备, 其包括: 存储装置, 其上存储有计算机程序; 处理装置, 用于执行所述存储装置中的所述计算机程序, 以实现权利要求 1-7中任 一项所述方法 的步骤。
PCT/SG2022/050254 2021-05-21 2022-04-28 特征构建方法、内容显示方法及相关装置 WO2022245279A1 (zh)

Applications Claiming Priority (2)

Application Number Priority Date Filing Date Title
CN202110558744.5A CN113177174B (zh) 2021-05-21 2021-05-21 特征构建方法、内容显示方法及相关装置
CN202110558744.5 2021-05-21

Publications (1)

Publication Number Publication Date
WO2022245279A1 true WO2022245279A1 (zh) 2022-11-24

Family

ID=76929620

Family Applications (1)

Application Number Title Priority Date Filing Date
PCT/SG2022/050254 WO2022245279A1 (zh) 2021-05-21 2022-04-28 特征构建方法、内容显示方法及相关装置

Country Status (2)

Country Link
CN (1) CN113177174B (zh)
WO (1) WO2022245279A1 (zh)

Citations (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US20160210657A1 (en) * 2014-12-30 2016-07-21 Anto Chittilappilly Real-time marketing campaign stimuli selection based on user response predictions
US20180005261A9 (en) * 2012-06-08 2018-01-04 Anto Chittilappilly A method , computer readable medium and system for determining touchpoint attribution
US20190278378A1 (en) * 2018-03-09 2019-09-12 Adobe Inc. Utilizing a touchpoint attribution attention neural network to identify significant touchpoints and measure touchpoint contribution in multichannel, multi-touch digital content campaigns
CN112529634A (zh) * 2020-12-18 2021-03-19 恩亿科(北京)数据科技有限公司 基于大数据的转化链路分析方法、系统和计算机设备

Family Cites Families (17)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US8655907B2 (en) * 2011-07-18 2014-02-18 Google Inc. Multi-channel conversion path position reporting
WO2013173626A2 (en) * 2012-05-18 2013-11-21 Clipfile Corporation Using content
WO2013176654A1 (en) * 2012-05-22 2013-11-28 Empire Technology Development, Llc Insurance adjustment for cloud based services
CN103605736B (zh) * 2013-11-19 2017-06-30 北京国双科技有限公司 转化数据的处理方法及装置
CN104133837B (zh) * 2014-06-24 2017-10-31 上海交通大学 一种基于分布式计算的互联网信息投放渠道优化系统
US9818134B2 (en) * 2015-04-02 2017-11-14 Vungle, Inc. Systems and methods for dynamic ad selection of multiple ads or ad campaigns on devices
CN108171528B (zh) * 2016-12-07 2022-02-25 阿里巴巴集团控股有限公司 一种归因方法及归因系统
US20190066154A1 (en) * 2017-08-24 2019-02-28 Conversant, Llc Escrow Agent for Conversion Verification
CN110209919B (zh) * 2018-02-23 2022-05-27 北京国双科技有限公司 归因的多维度剖析方法和装置
CN111127051B (zh) * 2018-10-30 2023-08-29 百度在线网络技术(北京)有限公司 多渠道动态归因方法、装置、服务器及存储介质
CN110059925A (zh) * 2019-03-15 2019-07-26 微梦创科网络科技(中国)有限公司 一种广告数据处理方法及装置
CN110287180B (zh) * 2019-06-25 2021-12-28 上海诚数信息科技有限公司 一种基于深度学习的风控建模方法
CN110569429B (zh) * 2019-08-08 2023-11-24 创新先进技术有限公司 一种内容选择模型的生成方法、装置和设备
CN111737575B (zh) * 2020-06-19 2023-11-14 北京字节跳动网络技术有限公司 内容分发方法、装置、可读介质及电子设备
CN111949890B (zh) * 2020-09-27 2021-02-05 平安科技(深圳)有限公司 基于医学领域的数据推荐方法、设备、服务器及存储介质
CN112434818B (zh) * 2020-11-19 2023-09-26 脸萌有限公司 模型构建方法、装置、介质及电子设备
CN112633945B (zh) * 2020-12-31 2024-05-10 北京达佳互联信息技术有限公司 落地页投放方法、投放数据处理方法、装置、设备及介质

Patent Citations (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US20180005261A9 (en) * 2012-06-08 2018-01-04 Anto Chittilappilly A method , computer readable medium and system for determining touchpoint attribution
US20160210657A1 (en) * 2014-12-30 2016-07-21 Anto Chittilappilly Real-time marketing campaign stimuli selection based on user response predictions
US20190278378A1 (en) * 2018-03-09 2019-09-12 Adobe Inc. Utilizing a touchpoint attribution attention neural network to identify significant touchpoints and measure touchpoint contribution in multichannel, multi-touch digital content campaigns
CN112529634A (zh) * 2020-12-18 2021-03-19 恩亿科(北京)数据科技有限公司 基于大数据的转化链路分析方法、系统和计算机设备

Also Published As

Publication number Publication date
CN113177174B (zh) 2024-02-06
CN113177174A (zh) 2021-07-27

Similar Documents

Publication Publication Date Title
JP2021103506A (ja) 情報を生成するための方法及び装置
CN111782968A (zh) 内容推荐方法、装置、可读介质及电子设备
CN110619078A (zh) 用于推送信息的方法和装置
WO2020199749A1 (zh) 基于反馈的信息推送方法、装置及电子设备
CN113934938A (zh) 信息展示的方法、装置、可读介质和电子设备
CN112214665A (zh) 内容展示方法、装置、电子设备及计算机可读存储介质
CN112669816B (zh) 模型的训练方法、语音识别方法、装置、介质及设备
CN112309390B (zh) 信息交互方法和装置
CN113240027A (zh) 图像分类方法、装置、可读介质及电子设备
WO2024198920A1 (zh) 内容推送模型的训练方法、装置、设备及存储介质
CN112148962B (zh) 用于推送信息的方法和装置
WO2022245280A1 (zh) 特征构建方法、内容显示方法及相关装置
CN117034959A (zh) 数据处理方法、装置、电子设备及存储介质
WO2022245279A1 (zh) 特征构建方法、内容显示方法及相关装置
CN112287171A (zh) 信息处理方法、装置和电子设备
CN113240107B (zh) 图像处理方法、装置和电子设备
CN116109374A (zh) 资源位展示方法、装置、电子设备和计算机可读介质
CN114595393A (zh) 一种数据处理方法及装置
CN114912039A (zh) 一种搜索特效展示方法、装置、设备及介质
CN115269978A (zh) 一种视频标签生成方法、装置、设备及介质
CN112187945A (zh) 信息推送方法、装置和电子设备
CN117726480B (zh) 基于数字化虚拟评标室的电力设备获取方法
CN118797157A (zh) 内容确定方法、装置、介质、设备以及计算机程序产品
CN116415092A (zh) 信息处理方法、装置和电子设备
CN116364066A (zh) 分类模型生成方法、音频分类方法、装置、介质及设备

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: 22805081

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: 22805081

Country of ref document: EP

Kind code of ref document: A1

122 Ep: pct application non-entry in european phase

Ref document number: 22805081

Country of ref document: EP

Kind code of ref document: A1