WO2020134571A1 - 页面展示方法、装置、终端设备及存储介质 - Google Patents

页面展示方法、装置、终端设备及存储介质 Download PDF

Info

Publication number
WO2020134571A1
WO2020134571A1 PCT/CN2019/115205 CN2019115205W WO2020134571A1 WO 2020134571 A1 WO2020134571 A1 WO 2020134571A1 CN 2019115205 W CN2019115205 W CN 2019115205W WO 2020134571 A1 WO2020134571 A1 WO 2020134571A1
Authority
WO
WIPO (PCT)
Prior art keywords
page
displayed
display
evaluation
image
Prior art date
Application number
PCT/CN2019/115205
Other languages
English (en)
French (fr)
Inventor
张博文
赵致辰
姜宇宁
徐力
Original Assignee
北京字节跳动网络技术有限公司
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 北京字节跳动网络技术有限公司 filed Critical 北京字节跳动网络技术有限公司
Publication of WO2020134571A1 publication Critical patent/WO2020134571A1/zh

Links

Images

Classifications

    • GPHYSICS
    • G06COMPUTING; CALCULATING OR COUNTING
    • G06FELECTRIC DIGITAL DATA PROCESSING
    • G06F9/00Arrangements for program control, e.g. control units
    • G06F9/06Arrangements for program control, e.g. control units using stored programs, i.e. using an internal store of processing equipment to receive or retain programs
    • G06F9/44Arrangements for executing specific programs
    • G06F9/451Execution arrangements for user interfaces
    • GPHYSICS
    • G06COMPUTING; CALCULATING OR COUNTING
    • G06FELECTRIC DIGITAL DATA PROCESSING
    • G06F16/00Information retrieval; Database structures therefor; File system structures therefor
    • G06F16/20Information retrieval; Database structures therefor; File system structures therefor of structured data, e.g. relational data

Definitions

  • Embodiments of the present disclosure relate to the field of data technology, for example, to a page display method, device, terminal device, and storage medium.
  • an online evaluation needs to be performed based on the identification (ID) characteristics of the page creativity and the user's ID characteristics, and for any newly created page to be displayed, the accurate ID of the page must be obtained
  • ID identification
  • the newly created pages to be displayed need to consume display opportunities. If there are a large number of newly created pages to be displayed with poor display effects, a large number of display opportunities and display time will be wasted. At the same time, it will also cause a decline in user experience.
  • Embodiments of the present disclosure provide a page display method, device, terminal device, and storage medium, which can improve the generation efficiency and quality of pages to be displayed, reduce display costs, and improve user experience.
  • An embodiment of the present disclosure provides a page display method.
  • the method includes:
  • the display prediction evaluation result and the preset display strategy select a target page to be displayed from the set of pages to be displayed for display.
  • an embodiment of the present disclosure also provides a page display device, which includes:
  • a module for obtaining a page to be displayed configured to obtain at least one page to be displayed
  • the display prediction evaluation result acquisition module is configured to input each page to be displayed into the page evaluation model for each page to be displayed to obtain the display prediction evaluation result of each page to be displayed; wherein, the page
  • the evaluation model includes a neural network model, and the neural network model includes a feature extraction layer and a fully connected layer;
  • a module for generating a set of pages to be displayed configured to generate a set of pages to be displayed according to the at least one page to be displayed and the display prediction evaluation result corresponding to the at least one page to be displayed;
  • the page to-be-displayed module is configured to select a target page to be displayed from the set of pages to be displayed for display according to the display prediction evaluation result and a preset display strategy.
  • an embodiment of the present disclosure also provides a terminal device.
  • the terminal device includes:
  • One or more processors are One or more processors;
  • Memory set to store one or more programs
  • the one or more processors When the one or more programs are executed by the one or more processors, the one or more processors implement the page display method as described in the embodiments of the present disclosure.
  • an embodiment of the present disclosure also provides a computer-readable storage medium that stores a computer program, and when the program is executed by a processor, a page display method as described in an embodiment of the present disclosure is implemented.
  • FIG. 1a is a flowchart of a page display method provided in Embodiment 1 of the present disclosure
  • FIG. 1b is a functional block diagram of a page evaluation model provided in Embodiment 1 of the present disclosure
  • FIG. 1c is a schematic diagram of a page to be displayed according to Embodiment 1 of the present disclosure.
  • FIG. 1d is a training flowchart of a page evaluation model provided in Embodiment 1 of the present disclosure
  • FIG. 2a is a flowchart of a page display method provided in Embodiment 2 of the present disclosure
  • FIG. 2b is a schematic diagram showing a prediction evaluation result provided by Embodiment 2 of the present disclosure.
  • FIG. 2c is another schematic diagram showing the prediction evaluation result provided by Embodiment 2 of the present disclosure.
  • FIG. 2d is a schematic structural diagram of an evaluation system for pages to be provided according to Embodiment 2 of the present disclosure
  • FIG. 3 is a schematic structural diagram of a page display device provided in Embodiment 3 of the present disclosure.
  • FIG. 4 is a schematic structural diagram of a terminal device according to Embodiment 4 of the present disclosure.
  • FIG. 1a is a flowchart of a page display method according to Embodiment 1 of the present disclosure. This embodiment is applicable to the case of displaying pages.
  • the method may be performed by a page display device, which may use software and/or hardware.
  • the device can be configured in a terminal device, such as a computer. As shown in Figure 1a, the method includes the following steps:
  • the page to be displayed may refer to a page used for display in the display interface of the terminal device, for example, a display interface of a browser in the terminal device, or a display interface of an application program in the terminal device.
  • the page to be displayed may include at least one of the following: a title page, an image page, and a video page.
  • the title page may refer to a page containing only text content
  • the image page may refer to a page containing static text content and image content Pages
  • video pages can refer to pages that contain dynamic text content, image content, and audio content.
  • the page to be displayed is a page that is not displayed, and also refers to a newly created page designed by a user.
  • the page to be displayed is not displayed on the network, so that all information of the page to be displayed is obtained offline, that is, the page to be displayed is evaluated according to the offline information.
  • the page to be displayed is not displayed, in fact, the display object of the page to be displayed is unknown, that is, during the evaluation of the page to be displayed, the user characteristics of the actual display object of the page to be displayed cannot be obtained, Therefore, it is impossible to effectively evaluate the displayed page based on online data (user characteristics of the display object).
  • the page evaluation model is used to evaluate the page to be displayed, and the page evaluation model receives the page to be displayed and outputs the display prediction evaluation result corresponding to the page to be displayed.
  • the evaluation result of the display prediction is data for predicting the display effect of the page to be displayed after the display.
  • the display effect data may be user feedback.
  • the display performance data may be at least one of the click rate and conversion rate fed back by the terminal device used by the user.
  • the page evaluation model may use a neural network model, where the neural network model may include, but is not limited to, word vector network (Word to Vec Network, Word 2vec Network), convolutional neural network (Convolution Neural Network, CNN) and fully connected network (Fully Connected Network, FCN), etc.
  • the network structure of the neural network model may include a feature extraction layer and a fully connected layer.
  • the feature extraction layer is used to extract image features and text features of the page to be displayed.
  • Each node in the fully connected layer is connected to all nodes in the layer above the fully connected layer, and is used to synthesize the features extracted by the feature extraction layer.
  • the fully connected layer is after the feature extraction layer.
  • the page evaluation model can avoid manual evaluation of the subjectivity of the page, and at the same time can improve the accuracy and efficiency of page evaluation and reduce labor costs.
  • the inputting each page to be displayed into the page evaluation model to obtain the display prediction evaluation result corresponding to each page to be displayed may include: inputting each page to be displayed into the page evaluation model In the image extraction layer of the feature extraction layer, the image features of each page to be displayed are obtained; the word embedding layer of the feature extraction layer is used to obtain the text features of each page to be displayed; Describe the image features of each page to be displayed and the text features of each page to be displayed to generate a feature vector of each page to be displayed; according to the feature vector of each page to be displayed, through a fully connected layer, Obtain the display prediction evaluation result of each page to be displayed.
  • the page evaluation model 101 includes an image embedding layer 102, a word embedding layer 103, and a fully connected layer 104.
  • the image embedding layer 102 and the word embedding layer 103 may be collectively referred to as a feature extraction layer.
  • the image page includes text and images, and the text includes text and titles in the images.
  • the image embedding layer 102 can identify the image features of the image in the image page.
  • the word embedding layer 103 can obtain the text in the image through optical character recognition (OCR) technology, and the title text through the text recognition technology, so as to obtain the text features of the image page.
  • OCR optical character recognition
  • the feature vector acquisition module stitches the image features and text features in the image page to form a high-dimensional vector as the high-dimensional feature of the image page. According to the high-dimensional features of the image page, the display prediction evaluation result is obtained through the fully connected layer 104 and output in a numerical form.
  • the image embedding layer may use a mobile terminal network (MobileNet).
  • the image features corresponding to the image are represented by a 1024-dimensional vector of the penultimate layer of the image in MobileNet.
  • the word embedding layer can use the output results of the multi-scale convolutional neural network as text features.
  • the presentation form of the image page may also be that the title is below the image.
  • the presentation form of the image page may have other forms, which are not limited in the embodiments of the present disclosure.
  • the text recognition method in the image may also be other methods, and the image embedding layer and the word embedding layer may also have other structures, and this embodiment of the present disclosure does not limit this.
  • This embodiment is merely an example of providing a page evaluation model for evaluating image pages, but is not limited to this example.
  • the feature extraction layer of the page evaluation model used to evaluate the title page may be a word embedding layer, and the word embedding layer is used to extract text features in the title.
  • the video can be split into a series of image frames, and each image frame is input into the page evaluation model for evaluating the image page for evaluation, and finally, the prediction prediction evaluation of multiple image frames is obtained The sum of the results, divided by the time, and the obtained results are used as the prediction prediction evaluation results of the video page.
  • there are other network structures to which the embodiments of the present disclosure are not limited.
  • the page evaluation model Before using the page evaluation model, the page evaluation model needs to be trained in advance. In an embodiment, before acquiring at least one page to be displayed, it may further include: acquiring a historical display page and evaluation data of the historical display page, the evaluation data including display effect data and page information; additional evaluation data will be added
  • the history display page is used as a training sample to train the page evaluation model.
  • the history display page may refer to a page that has been displayed in the terminal device.
  • Page information can include page type information (such as video pages), page attribute information (such as page theme content), and page source information (such as source address or client name), etc., and can train different pages according to page information. Evaluate the model.
  • the display performance data may refer to the behavior data of the user on the historical display page, and may include at least one of click rate and conversion rate.
  • the click-through rate may refer to the probability that the user group clicks on the page, and the user group may be a user group with a set range or type (such as a female white-collar worker group of 28-32 years old); the conversion rate may refer to After the user clicks the page, the ratio of the number of users who perform the set behavior operation to the number of all users who click the page.
  • the evaluation data may further include at least one of the following: location information of the historical display page on the display interface, display time information, display weather information, user information, display context information, and client terminal displaying the historical display page operating system.
  • the user information may refer to at least one of the gender, age, and preference information of the user who issued the behavior operation on the history display page.
  • the operating system of the client can be an iOS system, an Android system, or a Windows system.
  • the presentation context information may refer to at least one item of information such as the current language environment, background, and page content of the page.
  • the display time information may be time information such as year, month, day, hour, minute, and second.
  • Displaying weather information may refer to at least one of information such as temperature, humidity, solar terms, and seasons.
  • the evaluation data may also include festival information.
  • the representativeness of the training samples is increased, thereby improving the accuracy of the evaluation results of the page evaluation model.
  • training samples are provided in pairs to train the page evaluation model, as shown in FIG. 1d.
  • the first image page and the second image page are respectively input into the page evaluation model for evaluation, and the first impression prediction evaluation result and the second impression prediction evaluation result are respectively obtained correspondingly.
  • the page evaluation model is shared weights.
  • the prediction difference which can represent the prediction results of the first image page and the second image page. For example, in the case where the prediction difference is greater than zero, the first image page is better than the second image page; When the difference is equal to zero, the first image page and the second image page have the same advantages and disadvantages; when the predicted difference is less than zero, the first image page is inferior to the second image page.
  • a nonlinear action function such as the Sigmoid function
  • the actual difference between the first image page and the second image page is determined according to the actual conversion rate of the first image page and the actual conversion rate of the second image page, and is used to represent the first image page and the second image page.
  • the actual advantages and disadvantages of the second image page are used as the loss function, and the convex optimization (convex optimization) method is used to solve the minimum value of the loss function to complete the training of the model.
  • the page to be displayed can be obtained online by building an evaluation service platform, and the pre-trained page evaluation model can be loaded through the evaluation service platform to evaluate the displayed page through the page evaluation model, and sent out to be displayed through the evaluation service platform
  • the display of the page predicts the evaluation results. It can also be a page evaluation model that directly obtains the matching according to the needs, and offline evaluation is performed on the pages to be displayed with the set requirements, for example, the pages to be displayed that are attached to the display interface and used to display location information and display time information can be performed. Evaluation to obtain the best display time of the page to be displayed and the best display position on the display interface.
  • the set of pages to be displayed includes at least one page to be displayed and display prediction evaluation results of at least one page to be displayed matching, and in addition, the set of pages to be displayed also includes evaluation data of at least one page to be displayed matching.
  • the set of pages to be displayed can be generated based on all the pages to be displayed evaluated by the page evaluation model, or the set of pages to be displayed can be generated according to at least one page to be displayed that the display prediction evaluation result exceeds a set threshold.
  • S140 Select a target page to be displayed from the set of pages to be displayed for display according to the display prediction evaluation result and a preset display strategy.
  • the display strategy may refer to a display method or display form of at least one page to be displayed, for example, display the top 10 pages to be displayed in the prediction evaluation result, and display online in order from the highest to the lowest in the ranking order.
  • the evaluation data of at least one page to be displayed for each page to be displayed, at a display position that matches the page to be displayed in the display interface, and within a display period that matches the page to be displayed
  • the page to be displayed is displayed.
  • the target page to be displayed may refer to the page to be displayed currently being displayed.
  • the embodiment of the present disclosure evaluates at least one page to be displayed through the page evaluation model, obtains the display prediction evaluation result corresponding to the at least one page to be displayed, and selects the target page to be displayed according to the preset display strategy according to the display prediction evaluation result , Solve the problem of low efficiency of page evaluation in the related technology that requires online data for manual evaluation of the page.
  • the page evaluation through the page evaluation model can avoid the subjectivity of page evaluation, while improving the accuracy and efficiency of page evaluation, and the evaluation does not require online Collecting data can save display opportunities and display time.
  • the method may further include: obtaining evaluation data of the target page to be displayed; adding evaluation data The target page to be displayed is used as a training sample to train the page evaluation model.
  • the evaluation data of the target page to be displayed is collected, and the target page to be displayed with the additional evaluation data is used as a new training sample to train the page evaluation model to update the page evaluation in real time
  • the training samples of the model improve the representativeness of the training samples and modify the evaluation results of the page evaluation model, thereby improving the evaluation accuracy of the page evaluation model.
  • FIG. 2a is a flowchart of a page display method according to Embodiment 2 of the present disclosure. This embodiment is based on the optional solutions in the above embodiments.
  • the method before acquiring at least one page to be displayed, the method further includes: acquiring a historical display page and evaluation data of the historical display page.
  • the evaluation data includes display effect data and page information;
  • the history display page is used as a training sample to train the page evaluation model.
  • the method further includes: obtaining evaluation data of the target page to be displayed; using the target page to be displayed with additional evaluation data as a training sample to train the page evaluation model.
  • S2010 Obtain a historical display page and evaluation data of the historical display page, where the evaluation data includes display effect data and page information.
  • a historical display page with additional evaluation data is used as a training sample to train a page evaluation model.
  • the page to be displayed can all Refer to the description of the above embodiment.
  • the page evaluation model includes a neural network model and the neural network model Including feature extraction layer and fully connected layer.
  • FIG. 2b and FIG. 2c respectively show two pages to be displayed and corresponding display prediction evaluation results.
  • the content shown in FIG. 2b is obtained through the page evaluation model according to the image page shown in FIG. 1c.
  • the two pages to be displayed are image pages, and from top to bottom are the image of the page to be displayed, the title of the page to be displayed, the optical character recognition result and the page in the image of the page to be displayed
  • the score given by the evaluation model that is, showing the predicted evaluation result.
  • the image page shown in FIG. 2b can be displayed preferentially.
  • S2050 Generate a set of pages to be displayed according to the at least one page to be displayed and a display prediction evaluation result corresponding to the at least one page to be displayed.
  • S2060 Select a target page to be displayed from the set of pages to be displayed for display according to the display prediction evaluation result and a preset display strategy.
  • FIG. 2d is a schematic structural diagram of a page evaluation system to be displayed according to Embodiment 2 of the present disclosure.
  • the page evaluation system to be displayed includes a posterior data collection and storage module 201, and model training.
  • the posterior data collection and storage module 201 collects all data required for page evaluation model training such as page information, user information, context information, and user behavior on the page displayed on the line, And store the data in the database; the model training module 202 preprocesses the stored data into the input format required by the page evaluation model, and performs page evaluation model training, offline evaluation, and selects the best model according to the offline evaluation index; to be displayed
  • the page effect prediction module 203 is set to evaluate the pages to be displayed online and offline, and each page to be displayed is scored, and the score result is output;
  • the online display module 204 is set to be based on the score of each page to be displayed, combined with the display strategy
  • FIG. 3 is a schematic structural diagram of a page display device according to an embodiment of the present disclosure. This embodiment can be applied to the case of generating a display page.
  • the device can be implemented in software and/or hardware, and the device can be configured in the terminal device.
  • the apparatus may include: a page to be displayed obtaining module 310, a display prediction evaluation result obtaining module 320, a page to be displayed set generation module 330, and a page to be displayed module 340.
  • the page to be displayed obtaining module 310 is configured to obtain at least one page to be displayed; the display prediction evaluation result obtaining module 320 is configured to input each page to be displayed into the page evaluation model for each page to be displayed to obtain The display prediction evaluation result of each page to be displayed; wherein, the page evaluation model includes a neural network model, the neural network model includes a feature extraction layer and a fully connected layer; and the page collection to be displayed generation module 330 is set to The at least one page to be displayed and the display prediction evaluation result corresponding to the at least one page to be displayed generate a set of pages to be displayed; the page to be displayed module 340 is set to be based on the display prediction evaluation result and a preset display strategy, from Select the target page to be displayed from the set of pages to be displayed for display.
  • the embodiment of the present disclosure evaluates at least one page to be displayed through the page evaluation model, obtains the display prediction evaluation result corresponding to the at least one page to be displayed, and selects the target page to be displayed according to the preset display strategy according to the display prediction evaluation result , Solve the problem of low efficiency of page evaluation in the related technology that requires online data for manual evaluation of the page.
  • the page evaluation through the page evaluation model can avoid the subjectivity of page evaluation, while improving the accuracy and efficiency of page evaluation, and the evaluation does not require online Collecting data can save display opportunities and display time.
  • the display prediction evaluation result acquisition module 320 includes: an image feature acquisition module configured to acquire image features of each page to be displayed through an image embedding layer in the feature extraction layer; a text feature acquisition module, Set to acquire the text features of each page to be displayed through the word embedding layer in the feature extraction layer; the feature vector acquisition module is set to be based on the image features of each page to be displayed and the each to be displayed The text feature of the page generates the feature vector of the page to be displayed; the display prediction evaluation result determination module is set to obtain the feature of each page to be displayed through the fully connected layer according to the feature vector of each page to be displayed Show the forecast evaluation results.
  • the page display device further includes: a historical display page acquisition module configured to acquire a historical display page and evaluation data of the historical display page, the evaluation data including display effect data and page information;
  • the evaluation model training module is configured to train the page evaluation model by using a historical display page of additional evaluation data as a training sample.
  • the display performance data includes at least one of click-through rate and conversion rate.
  • the evaluation data further includes at least one of the following: location information of the historical display page on the display interface, display time information, display weather information, user information, display context information, and customers displaying the historical display page Operating system.
  • the page to be displayed includes at least one of the following: a title page, an image page, and a video page.
  • the page display device further includes: an evaluation data acquisition module configured to obtain evaluation data of the target page to be displayed; a training sample acquisition module configured to use the target evaluation page with additional evaluation data as training Sample, train the page evaluation model.
  • the page display device provided by the embodiment of the present disclosure belongs to the same inventive concept as the page display method provided by the embodiment 1.
  • the embodiment 1 For technical details not described in detail in the embodiments of the present disclosure, please refer to the embodiment 1, and the embodiments and the embodiments of the present disclosure One has the same beneficial effect.
  • FIG. 4 shows a schematic structural diagram of a terminal device (such as a client or a server) 400 suitable for implementing the embodiment of the present disclosure.
  • the terminal devices in the embodiments of the present disclosure may include, but are not limited to, mobile phones, notebook computers, digital broadcast receivers, personal digital assistants (Personal Digital Assistant (PDA), tablet computers (Portable Android Device, PAD), portable multimedia players (Portable Media Player, PMP), mobile terminals such as in-vehicle terminals (such as in-vehicle navigation terminals), and fixed terminals such as digital televisions (Television, TV), desktop computers, and so on.
  • PDA Personal Digital Assistant
  • PMP portable multimedia players
  • mobile terminals such as in-vehicle terminals (such as in-vehicle navigation terminals)
  • fixed terminals such as digital televisions (Television, TV), desktop computers, and so on.
  • the terminal device shown in FIG. 4 is only an example, and should not bring any limitation to the functions and use scope of the embodiments of
  • the terminal device 400 may include a processing device (such as a central processor, a graphics processor, etc.) 401, and the processing device may be stored in a read-only memory (Read-only Memory, ROM) 402 program or from a storage device.
  • the device 408 loads the program in the random access memory (Random Access Memory, RAM) 403 to perform at least one appropriate action and process.
  • RAM Random Access Memory
  • the processing device 401, ROM 402, and RAM 403 are connected to each other via a bus 404.
  • An input/output (Input/Output, I/O) interface 405 is also connected to the bus 404.
  • the following devices can be connected to the I/O interface 405: including an input device 406 such as a touch screen, a touch pad, a keyboard, a mouse, a camera, a microphone, an accelerometer, a gyroscope, etc.; including, for example, a liquid crystal display (Liquid Crystal Display) , LCD), speakers, vibrators, and other output devices 407; including storage devices 408, such as magnetic tape, hard disk, etc.; and a communication device 409.
  • the communication device 409 may allow the terminal device 400 to perform wireless or wired communication with other devices to exchange data.
  • FIG. 4 shows a terminal device 400 having various devices, it is not required to implement or have all the devices shown. More or fewer devices may be implemented or provided instead.
  • the process described above with reference to the flowchart may be implemented as a computer software program.
  • embodiments of the present disclosure include a computer program product that includes a computer program carried on a computer-readable medium, the computer program containing program code for performing the method shown in the flowchart.
  • the computer program may be downloaded and installed from the network through the communication device 409, or from the storage device 408, or from the ROM 402.
  • the processing device 401 When the computer program is executed by the processing device 401, the above-mentioned functions defined in the method of the embodiments of the present disclosure are executed.
  • Embodiments of the present disclosure also provide a computer-readable storage medium.
  • the computer-readable medium may be a computer-readable signal medium or a computer-readable storage medium, or any combination of the two.
  • the 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 of the above.
  • Examples of computer-readable storage media may include, but are not limited to: electrical connections with one or more wires, portable computer diskettes, hard drives, RAM, ROM, erasable programmable read-only memory (Electrically Programmable Read-Only-Memory, EPROM or flash memory), optical fiber, portable compact disk read-only memory (Compact Disc Read-Only Memory, CD-ROM), optical storage device, magnetic storage device, or any suitable combination of the above.
  • the computer-readable storage medium may be any tangible medium containing or storing a program, which may be used by or in combination with an instruction execution system, apparatus, or device.
  • the computer-readable signal medium may include a data signal that is propagated in baseband or as part of a carrier wave, and the computer-readable signal medium carries computer-readable program code.
  • This propagated data signal can take many forms, including but not limited to electromagnetic signals, optical signals, or any suitable combination of the foregoing.
  • the computer-readable signal medium may also be any computer-readable medium other than a 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 may be transmitted using any appropriate medium, including but not limited to: electric wires, optical cables, radio frequency (RF), etc., or any suitable combination of the foregoing.
  • the above-mentioned computer-readable medium may be included in the above-mentioned terminal device; or it may exist alone without being assembled into the terminal device.
  • the computer-readable medium carries one or more programs.
  • the terminal device When the one or more programs are executed by the terminal device, the terminal device: obtain at least one page to be displayed; input each page to be displayed into the page evaluation model In the method, a display prediction evaluation result corresponding to each page to be displayed is obtained; wherein, the page evaluation model includes a neural network model, and the neural network model includes a feature extraction layer and a fully connected layer; according to the at least one A display page and a display prediction evaluation result corresponding to the at least one page to be displayed generate a page set to be displayed; according to the display prediction evaluation result and a preset display strategy, select a target page to be displayed from the page to be displayed set Show.
  • the computer program code for performing the operations of the present disclosure can be written in one or more programming languages or a combination thereof.
  • the programming languages include object-oriented programming languages such as Java, Smalltalk, C++, and also include conventional Procedural programming language-such as "C" language or similar programming language.
  • the program code may execute entirely on the user's computer, partly on the user's computer, as an independent 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's computer through any kind of network, including a local area network (Local Area Network, LAN) or a wide area network (Wide Area Network, WAN), or it can be connected to an external computer (for example, using an Internet service provider to connect through the Internet).
  • LAN Local Area Network
  • WAN Wide Area Network
  • each block in the flowchart or block diagram may represent a module, a program segment, or a part of code, and the module, program segment, or a part of code contains one or more executable instructions for implementing a prescribed logical function.
  • the functions marked in the blocks may also occur in an order different from that marked in the drawings. For example, two blocks represented in succession may actually be executed in parallel, and they may sometimes be executed in the reverse order, depending on the functions involved.
  • the modules described in the embodiments of the present disclosure may be implemented in software or hardware.
  • the name of the module does not constitute a limitation on the module itself.
  • the module for acquiring a page to be displayed may also be described as “a module for acquiring at least one page to be displayed”.

Landscapes

  • Engineering & Computer Science (AREA)
  • Software Systems (AREA)
  • Theoretical Computer Science (AREA)
  • Physics & Mathematics (AREA)
  • General Engineering & Computer Science (AREA)
  • General Physics & Mathematics (AREA)
  • Human Computer Interaction (AREA)
  • Data Mining & Analysis (AREA)
  • Databases & Information Systems (AREA)
  • User Interface Of Digital Computer (AREA)

Abstract

一种页面展示方法、装置、终端设备及存储介质。方法包括:获取至少一个待展示页面(S110);将每个待展示页面输入到页面评估模型中,得到与所述每个待展示页面对应的展示预测评估结果(S120);其中,所述页面评估模型包括神经网络模型,所述神经网络模型包括特征提取层和全连接层;根据所述至少一个待展示页面以及所述至少一个待展示页面对应的展示预测评估结果生成待展示页面集合(S130);根据所述展示预测评估结果以及预设的展示策略,从所述待展示页面集合中选择目标待展示页面进行展示(S140)。

Description

页面展示方法、装置、终端设备及存储介质
本申请要求在2018年12月27日提交中国专利局、申请号为201811614139.X的中国专利申请的优先权,该申请的全部内容通过引用结合在本申请中。
技术领域
本公开实施例涉及数据技术领域,例如涉及一种页面展示方法、装置、终端设备及存储介质。
背景技术
随着通信技术和终端设备的发展,多种终端设备例如安卓手机、苹果手机和平板电脑等已经成为了人们工作和生活中不可或缺的一部分。为了满足人们对信息的获取需求,通常在为终端设备开发的应用程序上展示大量页面。
在相关技术中的页面展示方法中,需要根据页面创意的识别(Identification,ID)特征和用户的ID特征,进行在线评估,而针对任何一个新建的待展示页面,要得到该页面的准确的ID特征,必须经历冷启动过程。但在冷启动期间,新建的待展示页面需要消耗展示机会,如果存在大量展示效果差的新建待展示页面,则会浪费大量展示机会和展示时间。同时,还会导致用户体验下降。
发明内容
本公开实施例提供一种页面展示方法、装置、终端设备及存储介质,可以提高待展示页面的生成效率和质量,降低展示成本,提高用户体验。
本公开实施例提供了一种页面展示方法,该方法包括:
获取至少一个待展示页面;
将每个待展示页面输入到页面评估模型中,得到与所述每个待展示页面对应的展示预测评估结果;其中,所述页面评估模型包括神经网络模型,所述神经网络模型包括特征提取层和全连接层;
根据所述至少一个待展示页面以及所述至少一个待展示页面对应的展示预测评估结果生成待展示页面集合;
根据所述展示预测评估结果以及预设的展示策略,从所述待展示页面集合中选择目标待展示页面进行展示。
第二方面,本公开实施例还提供了一种页面展示装置,该装置包括:
待展示页面获取模块,设置为获取至少一个待展示页面;
展示预测评价结果获取模块,设置为针对每个待展示页面,将所述每个待展示页面输入到页面评估模型中,得到所述每个待展示页面的展示预测评估结果;其中,所述页面评估模型包括神经网络模型,所述神经网络模型包括特征提取层和全连接层;
待展示页面集合生成模块,设置为根据所述至少一个待展示页面以及所述至少一个待展示页面对应的展示预测评估结果生成待展示页面集合;
待展示页面展示模块,设置为根据所述展示预测评估结果以及预设的展示策略,从所述待展示页面集合中选择目标待展示页面进行展示。
第三方面,本公开实施例还提供了一种终端设备,该终端设备包括:
一个或多个处理器;
存储器,设置为存储一个或多个程序;
当所述一个或多个程序被所述一个或多个处理器执行,使得所述一个或多个处理器实现如本公开实施例所述的页面展示方法。
第四方面,本公开实施例还提供了一种计算机可读存储介质,存储有计算机程序,该程序被处理器执行时实现如本公开实施例所述的页面展示方法。
附图说明
图1a是本公开实施例一提供的一种页面展示方法的流程图;
图1b是本公开实施例一提供的一种页面评估模型的功能框图;
图1c是本公开实施例一提供的一种待展示页面的示意图;
图1d是本公开实施例一提供的一种页面评估模型的训练流程图;
图2a是本公开实施例二提供的一种页面展示方法的流程图;
图2b是本公开实施例二提供的一种展示预测评估结果的示意图;
图2c是本公开实施例二提供的另一种展示预测评估结果的示意图;
图2d是本公开实施例二提供的一种待展示页面评估系统的结构示意图;
图3是本公开实施例三提供的一种页面展示装置的结构示意图;
图4是本公开实施例四提供的一种终端设备的结构示意图。
具体实施方式
下面结合附图和实施例对本公开进行说明。此处所描述的具体实施例仅仅用于解释本公开,而非对本公开的限定。另外,为了便于描述,附图中仅示出了与本公开相关的部分而非全部结构。
实施例一
图1a为本公开实施例一提供的一种页面展示方法的流程图,本实施例可适用于展示页面的情况,该方法可以由页面展示装置来执行,该装置可以采用软件和/或硬件的方式实现,该装置可以配置于终端设备中,例如计算机等。如图1a所示,该方法包括如下步骤:
S110,获取至少一个待展示页面。
一实施例中,待展示页面可以是指在终端设备的显示界面中用于展示的页面,例如终端设备中浏览器的显示界面,或者是终端设备中应用程序的显示界面。待展示页面可以包括下述至少一项:标题页面、图片页面和视频页面,一实施例中,标题页面可以是指仅包含文字内容的页面;图片页面可以是指包含静态的文字内容和图像内容的页面;视频页面可以是指包含动态的文字内容、图像内容和音频内容的页面。
在本公开实施例中,待展示页面是未进行展示的页面,也是指由用户设计的新建页面。
一实施例中,待展示页面并未在网络上进行展示,从而待展示页面的所有信息都是离线获取的,也即根据离线信息对待展示页面进行评估。例如,由于待展示页面未展示,实际上该待展示页面的展示对象是未知的,也就是说,在对待展示页面进行评估的过程中,无法获取该待展示页面的实际展示对象的用户特征,从而无法根据在线数据(展示对象的用户特征)对待展示页面进行有效评估。
S120,将每个待展示页面输入到页面评估模型中,得到与所述每个待展示页面对应的展示预测评估结果;其中,所述页面评估模型包括神经网络模型,所述神经网络模型包括特征提取层和全连接层。
一实施例中,页面评估模型用于评估待展示页面,页面评估模型接收待展示页面并输出该待展示页面对应的展示预测评估结果。一实施例中,展示预测评估结果为预测待展示页面在展示之后的展示效果数据。一实施例中,展示效果数据可以是用户的反馈情况。一实施例中,展示效果数据可以是用户使用的终端设备反馈的点击率和转化率中的至少之一。
页面评估模型可以采用神经网络模型,其中,神经网络模型可以包括但不限于词向量网络(Word to Vec Network,Word2vec Network)、卷积神经网络 (Convolution Neural Network,CNN)和全连接网络(Fully Connected Network,FCN)等。神经网络模型的网络结构可以包括特征提取层和全连接层。本实施例中,特征提取层用于提取待展示页面的图像特征和文字特征。全连接层中的每个结点都与全连接层的上一层的所有结点相连接,用于将特征提取层提取的特征综合起来。一实施例中,全连接层在特征提取层的后一层。
通过页面评估模型可以避免人工评估页面的主观性,同时可以提高页面评估的准确性和效率,降低人工成本。
一实施例中,所述将每个待展示页面输入到页面评估模型中,得到与所述每个待展示页面对应的展示预测评估结果,可以包括:将每个待展示页面输入到页面评估模型中,通过所述特征提取层中的图像嵌入层获取所述每个待展示页面的图像特征;通过所述特征提取层中的字嵌入层获取所述每个待展示页面的文字特征;根据所述每个待展示页面的图像特征和所述每个待展示页面的文字特征,生成所述每个待展示页面的特征向量;根据所述每个待展示页面的特征向量,通过全连接层,获取所述每个待展示页面的展示预测评估结果。
一实施例中,如图1b所示,页面评估模型101包括图像嵌入层102、字嵌入层103和全连接层104,图像嵌入层102和字嵌入层103可以合称为特征提取层。一实施例中,如图1c所示,图像页面中包括文字和图像,文字包括图像中的文字和标题。通过图像嵌入层102可以识别出图像页面中图像的图像特征。一实施例中,字嵌入层103可以通过光字符识别(Optical Character Recognition,OCR)技术获得图像中的文字,并通过文字识别技术获取标题文字,从而可以获取该图像页面的文字特征。特征向量获取模块将图像页面中的图像特征和文字特征进行拼接,形成高维度的向量作为该图像页面的高维特征。根据图像页面的高维特征,通过全连接层104得到展示预测评估结果,以数值形式输出。
一实施例中,图像嵌入层可以采用移动端网络(MobileNet),一实施例中,以图像在MobileNet中倒数第二层的1024维的向量来表示该图像对应的图像特征。字嵌入层可以采用多尺度的卷积神经网络的输出结果作为文字特征。
一实施例中,图像页面的展现形式还可以是标题在图像的下方,此外,图像页面的展现形式还有其他形式,对此,本公开实施例不作限制。同时,图像中的文字识别方法还可以是其他方法,以及图像嵌入层和字嵌入层均还可以是其他结构,对此,本公开实施例不作限制。
本实施例仅仅是提供了一个评估图像页面的页面评估模型的实例,但并不限于本实例。例如,针对标题页面,仅包括文字,从而用于评估标题页面的页面评估模型的特征提取层可以为字嵌入层,字嵌入层用于提取标题中的文字特征。又如,针对视频页面,可以将视频拆分成一系列图像帧,并将每一图像帧 输入到该用于评估图像页面的页面评估模型中进行评估,最后求取多个图像帧的展示预测评估结果之和,并除以时间,将得到的结果作为该视频页面的展示预测评估结果。此外,还有其他网络结构,对此,本公开实施例不作限制。
在使用页面评估模型之前,需要预先对该页面评估模型进行训练。在一实施例中,在获取至少一个待展示页面之前,还可以包括:获取历史展示页面,以及所述历史展示页面的评估数据,所述评估数据包括展示效果数据和页面信息;将附加评估数据的历史展示页面作为训练样本,训练所述页面评估模型。
一实施例中,历史展示页面可以是指在终端设备中已展示的页面。页面信息可以包括页面的类型信息(如视频页面)、页面的属性信息(如页面主题内容)和页面的来源信息(如源地址或客户端名称)等,而且可以根据页面信息对应训练不同的页面评估模型。展示效果数据可以是指用户对该历史展示页面的行为数据,可以包括点击率和转化率中的至少之一。一实施例中,点击率可以是指用户群体点击页面的概率,用户群体可以是设定范围或设定类型的用户群体(如28岁-32岁的女白领群体);转化率可以是指在用户点击页面之后,执行设定行为操作的用户数目,与所有点击该页面的用户的数目的比值。
一实施例中,评估数据还可以包括下述至少一项:历史展示页面在展示界面的位置信息、展示时间信息、展示天气信息、用户信息、展示上下文信息和显示所述历史展示页面的客户端的操作系统。
一实施例中,用户信息可以是指对历史展示页面发出行为操作的用户的性别、年龄和偏好信息等中的至少一项。客户端的操作系统可以是iOS系统、Android系统或Windows系统等。展示上下文信息可以是指页面的当前语言环境、背景和页面内容等信息中的至少一项。展示时间信息可以是年、月、日、时、分和秒等时间信息。展示天气信息可以是指温度、湿度、节气和季节等信息中的至少一种。此外,评估数据还可以包括节日信息。
通过收集历史展示页面,以及历史展示页面匹配的评估数据,作为页面评估模型的训练样本,增加训练样本的代表性,从而提高页面评估模型的评估结果的准确性。
一实施例中,成对提供训练样本对页面评估模型进行训练,如图1d所示。将第一图像页面和第二图像页面分别输入到页面评估模型中进行评估,并分别对应得到第一展示预测评估结果和第二展示预测评估结果。一实施例中,在评估第一图像页面和第二图像页面的过程中,该页面评估模型是共享权重(shared weights)的。
获取第一展示预测评估结果和第二展示预测评估结果的差值,并通过非线 性作用函数(如Sigmoid函数)将差值映射到[0,1]的范围中,并作为该页面评估模型的预测差值,该预测差值可以表示第一图像页面和第二图像页面的预测优劣结果,例如,在预测差值大于零的情况下,第一图像页面好于第二图像页面;在预测差值等于零的情况下,第一图像页面与第二图像页面的优劣程度相同;在预测差值小于零的情况下,第一图像页面劣于第二图像页面。
一实施例中,根据第一图像页面的实际转化率和第二图像页面的实际转化率,确定第一图像页面和第二图像页面之间的实际差值,用于表示第一图像页面和第二图像页面的实际优劣结果。将预测差值与实际差值之间的交叉熵(cross entropy)作为损失函数(loss function),并使用凸优化(convex optimization)方法求解损失函数的最小值即可完成对模型的训练。
一实施例中,可以通过搭建评估服务平台在线获取待展示页面,以及通过评估服务平台加载预先训练的页面评估模型以通过页面评估模型对待展示页面进行评估,并通过评估服务平台向外发送待展示页面的展示预测评估结果。还可以是根据需求直接获取匹配的页面评估模型,对有设定需求的待展示页面进行离线评估,例如,可以对附加在展示界面中的用于展示位置信息和展示时间信息的待展示页面进行评估,以获取该待展示页面的最佳展示时间和在展示界面的最佳展示位置。
S130,根据所述至少一个待展示页面以及所述至少一个待展示页面对应的展示预测评估结果生成待展示页面集合。
待展示页面集合中包括至少一个待展示页面以及至少一个待展示页面匹配的展示预测评估结果,此外,待展示页面集合中还包括至少一个待展示页面匹配的评估数据。
可以根据所有经过页面评估模型评估的待展示页面生成待展示页面集合,或者还可以根据展示预测评估结果超过设定阈值的至少一个待展示页面生成待展示页面集合。
S140,根据所述展示预测评估结果以及预设的展示策略,从所述待展示页面集合中选择目标待展示页面进行展示。
展示策略可以是指至少一个待展示页面的展示方法或展示形式,例如展示预测评估结果中排名前10的待展示页面,按照排名顺序从高到低依次循环在线展示。或者,还可以根据至少一个待展示页面的评估数据,针对每个待展示页面,在展示界面中与该待展示页面匹配的展示位置处,并在与该待展示页面匹配的展示时间段内对该待展示页面进行展示。目标待展示页面可以是指当前正在展示的待展示页面。
本公开实施例通过页面评估模型对至少一个待展示页面进行评估,获取至少一个待展示页面对应的展示预测评估结果,并根据展示预测评估结果,按照预设的展示策略选择目标待展示页面进行展示,解决了相关技术中需要在线数据进行人工评估页面导致页面评估效率低的问题,通过页面评估模型实现页面评估可以避免页面评估的主观性,同时提高页面评估的准确性和效率,而且评估无需在线采集数据,可以节省展示机会和展示时间。
一实施例中,在上述实施例的基础上,在从所述待展示页面集合中选择目标待展示页面进行展示之后,还可以包括:获取所述目标待展示页面的评估数据;将附加评估数据的目标待展示页面作为训练样本,训练所述页面评估模型。
一实施例中,在目标待展示页面展示之后,采集目标待展示页面的评估数据,并将附加评估数据的目标待展示页面作为新的训练样本,对页面评估模型进行训练,从而实时更新页面评估模型的训练样本,提高训练样本的代表性,以及对页面评估模型的评估结果进行修正,从而提高页面评估模型的评估准确性。
实施例二
图2a为本公开实施例二提供的一种页面展示方法的流程图。本实施例以上述实施例中可选方案为基础。在本实施例中,在获取至少一个待展示页面之前,还包括:获取历史展示页面,以及所述历史展示页面的评估数据,所述评估数据包括展示效果数据和页面信息;将附加评估数据的历史展示页面作为训练样本,训练所述页面评估模型。在从所述待展示页面集合中选择目标待展示页面进行展示之后,还包括:获取所述目标待展示页面的评估数据;将附加评估数据的目标待展示页面作为训练样本,训练所述页面评估模型。
本实施例的方法可以包括:
S2010,获取历史展示页面,以及所述历史展示页面的评估数据,所述评估数据包括展示效果数据和页面信息。
S2020,将附加评估数据的历史展示页面作为训练样本,训练页面评估模型。
S2030,获取至少一个待展示页面。
本实施例中的待展示页面、页面评估模型、神经网络模型、特征提取层、全连接层、历史展示页面、展示效果数据、页面信息、展示预测评估结果、待展示页面集合和展示策略均可以参考上述实施例的描述。
S2040,将每个待展示页面输入到所述页面评估模型中,得到与所述每个待 展示页面对应的展示预测评估结果;其中,所述页面评估模型包括神经网络模型,所述神经网络模型包括特征提取层和全连接层。
一实施例中,图2b和图2c分别展示了两个待展示页面以及对应的展示预测评估结果。一实施例中,图2b所示的内容是根据图1c所示的图像页面通过页面评估模型得到的。如图2b和图2c所示,两个待展示页面均为图像页面,自上到下分别是待展示页面的图像、待展示页面的标题、待展示页面的图像中的光字符识别结果和页面评估模型给出的得分(即展示预测评估结果)。根据两个图像页面的得分结果,可以优先展示图2b所示的图像页面。
S2050,根据所述至少一个待展示页面以及所述至少一个待展示页面对应的展示预测评估结果生成待展示页面集合。
S2060,根据所述展示预测评估结果以及预设的展示策略,从所述待展示页面集合中选择目标待展示页面进行展示。
S2070,获取所述目标待展示页面的评估数据。
S2080,将附加评估数据的目标待展示页面作为训练样本,训练所述页面评估模型。
一实施例中,图2d为本公开实施例二提供的一种待展示页面评估系统的结构示意图,如图2d所示,该待展示页面评估系统包括后验数据收集及存储模块201、模型训练模块202、待展示页面效果预测模块203、在线展示模块204和在线学习(online learning)模块205。一实施例中,后验数据收集及存储模块201,收集线上展示过的所有页面的页面信息、用户信息、上下文信息和用户的对于页面的行为信息等页面评估模型训练所需的所有数据,并将数据落地存储于数据库中;模型训练模块202将存储的数据预处理为页面评估模型所需的输入格式,并进行页面评估模型训练、离线评估、根据离线评估指标选取最佳模型;待展示页面效果预测模块203设置为在线和离线评估待展示页面,对每个待展示页面打分,将打分结果输出;在线展示模块204,设置为根据每个待展示页面的打分情况,结合展示策略将待展示页面展现给用户;在线学习模块205,设置为在线上实时收集用户对待展示页面的行为数据,并将该行为数据作为该展示页面的实际值,使页面评估模型进行在线学习,从而实现对评估结果进行修正。
实施例三
图3为本公开实施例提供的一种页面展示装置的结构示意图,本实施例可适用于生成展示页面的情况。该装置可以采用软件和/或硬件的方式实现,该装 置可以配置于终端设备中。如图3所示,该装置可以包括:待展示页面获取模块310、展示预测评价结果获取模块320、待展示页面集合生成模块330和待展示页面展示模块340。
待展示页面获取模块310,设置为获取至少一个待展示页面;展示预测评价结果获取模块320,设置为针对每个待展示页面,将所述每个待展示页面输入到页面评估模型中,得到所述每个待展示页面的展示预测评估结果;其中,所述页面评估模型包括神经网络模型,所述神经网络模型包括特征提取层和全连接层;待展示页面集合生成模块330,设置为根据所述至少一个待展示页面以及所述至少一个待展示页面对应的展示预测评估结果生成待展示页面集合;待展示页面展示模块340,设置为根据所述展示预测评估结果以及预设的展示策略,从所述待展示页面集合中选择目标待展示页面进行展示。
本公开实施例通过页面评估模型对至少一个待展示页面进行评估,获取至少一个待展示页面对应的展示预测评估结果,并根据展示预测评估结果,按照预设的展示策略选择目标待展示页面进行展示,解决了相关技术中需要在线数据进行人工评估页面导致页面评估效率低的问题,通过页面评估模型实现页面评估可以避免页面评估的主观性,同时提高页面评估的准确性和效率,而且评估无需在线采集数据,可以节省展示机会和展示时间。
一实施例中,所述展示预测评价结果获取模块320,包括:图像特征获取模块,设置为通过所述特征提取层中的图像嵌入层获取每个待展示页面的图像特征;文字特征获取模块,设置为通过所述特征提取层中的字嵌入层获取所述每个待展示页面的文字特征;特征向量获取模块,设置为根据所述每个待展示页面的图像特征和所述每个待展示页面的文字特征,生成所述待展示页面的特征向量;展示预测评估结果确定模块,设置为根据所述每个待展示页面的特征向量,通过全连接层,获取所述每个待展示页面的展示预测评估结果。
一实施例中,所述页面展示装置,还包括:历史展示页面获取模块,设置为获取历史展示页面,以及所述历史展示页面的评估数据,所述评估数据包括展示效果数据和页面信息;页面评估模型训练模块,设置为将附加评估数据的历史展示页面作为训练样本,训练所述页面评估模型。
一实施例中,所述展示效果数据包括点击率和转化率中的至少一种。
一实施例中,所述评估数据还包括下述至少一项:历史展示页面在展示界面的位置信息、展示时间信息、展示天气信息、用户信息、展示上下文信息和显示所述历史展示页面的客户端的操作系统。
一实施例中,所述待展示页面包括下述至少一项:标题页面、图片页面和 视频页面。
一实施例中,所述页面展示装置,还包括:评估数据获取模块,设置为获取所述目标待展示页面的评估数据;训练样本获取模块,设置为将附加评估数据的目标待展示页面作为训练样本,训练所述页面评估模型。
本公开实施例提供的页面展示装置,与实施例一提供的页面展示方法属于同一发明构思,未在本公开实施例中详尽描述的技术细节可参见实施例一,并且本公开实施例与实施例一具有相同的有益效果。
实施例四
本公开实施例提供了一种终端设备,下面参考图4,其示出了适于用来实现本公开实施例的终端设备(例如客户端或服务器端)400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、个人数字助理(Personal Digital Assistant,PDA)、平板电脑(Portable Android Device,PAD)、便携式多媒体播放器(Portable Media Player,PMP)、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电视机(Television,TV)、台式计算机等等的固定终端。图4示出的终端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4所示,终端设备400可以包括处理装置(例如中央处理器、图形处理器等)401,处理装置可以根据存储在只读存储器(Read-only Memory,ROM)402中的程序或者从存储装置408加载到随机访问存储器(Random Access Memory,RAM)403中的程序而执行至少一种适当的动作和处理。在RAM 403中,还存储有终端设备400操作所需的至少一种程序和数据。处理装置401、ROM 402以及RAM 403通过总线404彼此相连。输入/输出(Input/Output,I/O)接口405也连接至总线404。
一实施例中,以下装置可以连接至I/O接口405: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406;包括例如液晶显示器(Liquid Crystal Display,LCD)、扬声器、振动器等的输出装置407;包括例如磁带、硬盘等的存储装置408;以及通信装置409。通信装置409可以允许终端设备400与其他设备进行无线或有线通信以交换数据。虽然图4示出了具有多种装置的终端设备400,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。
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,该计算机程序产品 包括承载在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409从网络上被下载和安装,或者从存储装置408被安装,或者从ROM 402被安装。在该计算机程序被处理装置401执行时,执行本公开实施例的方法中限定的上述功能。
实施例五
本公开实施例还提供了一种计算机可读存储介质,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——但不限于——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、RAM、ROM、可擦式可编程只读存储器(Electrically Programmable Read-Only-Memory,EPROM或闪存)、光纤、便携式紧凑磁盘只读存储器(Compact Disc Read-Only Memory,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,计算机可读信号介质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、射频(Radio Frequency,RF)等等,或者上述的任意合适的组合。
上述计算机可读介质可以是上述终端设备中所包含的;也可以是单独存在,而未装配入该终端设备中。
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终端设备执行时,使得该终端设备:获取至少一个待展示页面;将每个待展示页面输入到页面评估模型中,得到与所述每个待展示页面对应的展示预测评估结果;其中,所述页面评估模型包括神经网络模型,所述神经网络模型包括特征提取层和全连接层;根据所述至少一个待展示页面以及所述至少一个待展示页面对应的展示预测评估结果生成待展示页面集合;根据所述展示预测评估结果以及预设的展示策略,从所述待展示页面集合中选择目标待展示页面进 行展示。
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面向对象的程序设计语言—诸如Java、Smalltalk、C++,还包括常规的过程式程序设计语言—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——包括局域网(Local Area Network,LAN)或广域网(Wide Area Network,WAN)—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,图示了本公开至少一种实施例的系统、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。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。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。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,例如,待展示页面获取模块还可以被描述为“获取至少一个待展示页面的模块”。

Claims (16)

  1. 一种页面展示方法,包括:
    获取至少一个待展示页面;
    将每个待展示页面输入到页面评估模型中,得到与所述每个待展示页面对应的展示预测评估结果;其中,所述页面评估模型包括神经网络模型,所述神经网络模型包括特征提取层和全连接层;
    根据所述至少一个待展示页面以及所述至少一个待展示页面对应的展示预测评估结果生成待展示页面集合;
    根据所述展示预测评估结果以及预设的展示策略,从所述待展示页面集合中选择目标待展示页面进行展示。
  2. 根据权利要求1所述的方法,其中,所述将每个待展示页面输入到页面评估模型中,得到与所述每个待展示页面对应的展示预测评估结果,包括:
    将每个待展示页面输入到页面评估模型中,通过所述特征提取层中的图像嵌入层获取所述每个待展示页面的图像特征;
    通过所述特征提取层中的字嵌入层获取所述每个待展示页面的文字特征;
    根据所述每个待展示页面的图像特征和所述每个待展示页面的文字特征,生成所述每个待展示页面的特征向量;
    根据所述每个待展示页面的特征向量,通过所述全连接层,获取所述每个待展示页面的展示预测评估结果。
  3. 根据权利要求1或2所述的方法,在所述获取至少一个待展示页面之前,还包括:
    获取历史展示页面,以及所述历史展示页面的评估数据,所述评估数据包括展示效果数据和页面信息;
    将附加评估数据的历史展示页面作为训练样本,训练所述页面评估模型。
  4. 根据权利要求3所述的方法,其中,所述展示效果数据包括点击率和转化率中的至少之一。
  5. 根据权利要求3或4所述的方法,其中,所述评估数据还包括下述至少一项:
    所述历史展示页面在展示界面的位置信息、展示时间信息、展示天气信息、用户信息、展示上下文信息和显示所述历史展示页面的客户端的操作系统。
  6. 根据权利要求1-5任一项所述的方法,其中,所述待展示页面包括下述至少一项:标题页面、图片页面和视频页面。
  7. 根据权利要求1-6任一项所述的方法,在所述从所述待展示页面集合中选择目标待展示页面进行展示之后,还包括:
    获取所述目标待展示页面的评估数据;
    将附加评估数据的目标待展示页面作为训练样本,训练所述页面评估模型。
  8. 一种页面展示装置,包括:
    待展示页面获取模块,设置为获取至少一个待展示页面;
    展示预测评价结果获取模块,设置为将每个所述待展示页面输入到页面评估模型中,得到与所述每个待展示页面对应的展示预测评估结果;其中,所述页面评估模型包括神经网络模型,所述神经网络模型包括特征提取层和全连接层;
    待展示页面集合生成模块,设置为根据所述至少一个待展示页面以及所述至少一个待展示页面对应的展示预测评估结果生成待展示页面集合;
    待展示页面展示模块,设置为根据所述展示预测评估结果以及预设的展示策略,从所述待展示页面集合中选择目标待展示页面进行展示。
  9. 根据权利要求8所述的装置,其中,所述展示预测评价结果获取模块,包括:
    图像特征获取模块,设置为通过所述特征提取层中的图像嵌入层获取每个待展示页面的图像特征;
    文字特征获取模块,设置为通过所述特征提取层中的字嵌入层获取所述每个待展示页面的文字特征;
    特征向量获取模块,设置为根据所述每个待展示页面的图像特征和所述每个待展示页面的文字特征,生成所述每个待展示页面的特征向量;
    展示预测评估结果确定模块,设置为根据所述每个待展示页面的特征向量,通过所述全连接层,获取所述每个待展示页面的展示预测评估结果。
  10. 根据权利要求8或9所述的装置,还包括:
    历史展示页面获取模块,设置为获取历史展示页面,以及所述历史展示页面的评估数据,所述评估数据包括展示效果数据和页面信息;
    页面评估模型训练模块,设置为将附加评估数据的历史展示页面作为训练样本,训练所述页面评估模型。
  11. 根据权利要求10所述的装置,其中,所述展示效果数据包括点击率和转化率中的至少之一。
  12. 根据权利要求10或11所述的装置,其中,所述评估数据还包括下述至少一项:
    所述历史展示页面在展示界面的位置信息、展示时间信息、展示天气信息、用户信息、展示上下文信息和显示所述历史展示页面的客户端的操作系统。
  13. 根据权利要求8-12任一项所述的装置,其特征在于,所述待展示页面包括下述至少一项:标题页面、图片页面和视频页面。
  14. 根据权利要求8-13任一项所述的装置,还包括:
    评估数据获取模块,设置为获取所述目标待展示页面的评估数据;
    训练样本获取模块,设置为将附加评估数据的目标待展示页面作为训练样本,训练所述页面评估模型。
  15. 一种终端设备,包括:
    至少一个处理器;
    存储器,设置为存储至少一个程序;
    当所述至少一个程序被所述至少一个处理器执行,使得所述至少一个处理器实现如权利要求1-7任一所述的页面展示方法。
  16. 一种计算机可读存储介质,存储有计算机程序,所述计算机程序被处理器执行时实现如权利要求1-7任一所述的页面展示方法。
PCT/CN2019/115205 2018-12-27 2019-11-04 页面展示方法、装置、终端设备及存储介质 WO2020134571A1 (zh)

Applications Claiming Priority (2)

Application Number Priority Date Filing Date Title
CN201811614139.XA CN111381909B (zh) 2018-12-27 2018-12-27 一种页面展示方法、装置、终端设备及存储介质
CN201811614139.X 2018-12-27

Publications (1)

Publication Number Publication Date
WO2020134571A1 true WO2020134571A1 (zh) 2020-07-02

Family

ID=71128406

Family Applications (1)

Application Number Title Priority Date Filing Date
PCT/CN2019/115205 WO2020134571A1 (zh) 2018-12-27 2019-11-04 页面展示方法、装置、终端设备及存储介质

Country Status (2)

Country Link
CN (1) CN111381909B (zh)
WO (1) WO2020134571A1 (zh)

Cited By (9)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN112256537A (zh) * 2020-11-12 2021-01-22 腾讯科技(深圳)有限公司 模型运行状态的展示方法、装置、计算机设备和存储介质
CN112734523A (zh) * 2021-01-11 2021-04-30 北京城市网邻信息技术有限公司 信息展示方法、装置、电子设备和计算机可读介质
CN113158102A (zh) * 2021-02-23 2021-07-23 北京三快在线科技有限公司 页面配置方法、装置、电子设备和计算机可读介质
CN113378097A (zh) * 2021-07-08 2021-09-10 北京安天网络安全技术有限公司 一种web页面展示方法、装置、电子设备及存储介质
CN113713378A (zh) * 2021-09-02 2021-11-30 北京百度网讯科技有限公司 内容生成方法和装置
CN113760272A (zh) * 2020-08-24 2021-12-07 北京沃东天骏信息技术有限公司 信息展示方法、装置、设备及存储介质
CN113794604A (zh) * 2021-09-09 2021-12-14 北京恒安嘉新安全技术有限公司 一种网络安全态势展示方法、装置、设备及存储介质
CN113835582A (zh) * 2021-09-27 2021-12-24 青岛海信移动通信技术股份有限公司 一种终端设备、信息显示方法和存储介质
CN114690996A (zh) * 2020-12-31 2022-07-01 Oppo广东移动通信有限公司 内容显示方法、装置以及电子设备

Families Citing this family (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN112051953B (zh) * 2020-09-29 2021-09-14 中国银行股份有限公司 一种页面栏位的输出控制方法、装置及电子设备
CN112307966B (zh) * 2020-10-30 2023-09-05 京东方科技集团股份有限公司 事件展示方法及装置、存储介质及电子设备
CN113077305B (zh) * 2021-03-23 2024-03-29 上海尊溢商务信息咨询有限公司 页面处理方法、系统、电子设备及存储介质
CN115035192B (zh) * 2022-06-21 2023-04-14 北京远舢智能科技有限公司 一种烟叶布料车和传送带的位置确定方法及装置

Citations (6)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US9298763B1 (en) * 2013-03-06 2016-03-29 Google Inc. Methods for providing a profile completion recommendation module
CN106484913A (zh) * 2016-10-26 2017-03-08 腾讯科技(深圳)有限公司 一种目标图片确定的方法以及服务器
CN106557556A (zh) * 2016-11-08 2017-04-05 北京奇虎科技有限公司 一种网页页面的展示方法、装置、服务器和系统
CN106682144A (zh) * 2016-12-20 2017-05-17 上海亿账通互联网科技有限公司 页面展示方法和装置
CN108021626A (zh) * 2017-11-22 2018-05-11 阿里巴巴集团控股有限公司 页面排版方法、装置及设备
CN108959619A (zh) * 2018-07-17 2018-12-07 武汉市冰盒网络科技有限公司 内容屏蔽方法、用户设备、存储介质及装置

Family Cites Families (13)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US7822636B1 (en) * 1999-11-08 2010-10-26 Aol Advertising, Inc. Optimal internet ad placement
US8533043B2 (en) * 2010-03-31 2013-09-10 Yahoo! Inc. Clickable terms for contextual advertising
WO2012075600A1 (en) * 2010-12-06 2012-06-14 Yahoo! Inc. System and method for list ranking and ads placement using interaction features
CN104992348B (zh) * 2015-06-24 2018-07-10 深圳市腾讯计算机系统有限公司 一种信息展示的方法和装置
CN105208113A (zh) * 2015-08-31 2015-12-30 北京百度网讯科技有限公司 信息推送的方法和装置
CN106897892A (zh) * 2015-12-18 2017-06-27 北京奇虎科技有限公司 广告投放方法及装置
CN105678335B (zh) * 2016-01-08 2019-07-02 车智互联(北京)科技有限公司 预估点击率的方法、装置及计算设备
CN107545301B (zh) * 2016-06-23 2020-10-20 阿里巴巴集团控股有限公司 页面展示方法及装置
CN107391538B (zh) * 2017-04-26 2020-08-14 阿里巴巴集团控股有限公司 点击数据采集、处理和展示方法、装置、设备及存储介质
CN108804469B (zh) * 2017-05-04 2021-10-29 腾讯科技(深圳)有限公司 一种网页识别方法以及电子设备
CN108182472A (zh) * 2018-01-30 2018-06-19 百度在线网络技术(北京)有限公司 用于生成信息的方法和装置
CN108416625A (zh) * 2018-02-28 2018-08-17 阿里巴巴集团控股有限公司 营销产品的推荐方法和装置
CN109086439B (zh) * 2018-08-15 2022-02-25 腾讯科技(深圳)有限公司 信息推荐方法及装置

Patent Citations (6)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US9298763B1 (en) * 2013-03-06 2016-03-29 Google Inc. Methods for providing a profile completion recommendation module
CN106484913A (zh) * 2016-10-26 2017-03-08 腾讯科技(深圳)有限公司 一种目标图片确定的方法以及服务器
CN106557556A (zh) * 2016-11-08 2017-04-05 北京奇虎科技有限公司 一种网页页面的展示方法、装置、服务器和系统
CN106682144A (zh) * 2016-12-20 2017-05-17 上海亿账通互联网科技有限公司 页面展示方法和装置
CN108021626A (zh) * 2017-11-22 2018-05-11 阿里巴巴集团控股有限公司 页面排版方法、装置及设备
CN108959619A (zh) * 2018-07-17 2018-12-07 武汉市冰盒网络科技有限公司 内容屏蔽方法、用户设备、存储介质及装置

Cited By (12)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN113760272A (zh) * 2020-08-24 2021-12-07 北京沃东天骏信息技术有限公司 信息展示方法、装置、设备及存储介质
CN112256537A (zh) * 2020-11-12 2021-01-22 腾讯科技(深圳)有限公司 模型运行状态的展示方法、装置、计算机设备和存储介质
CN112256537B (zh) * 2020-11-12 2024-03-29 腾讯科技(深圳)有限公司 模型运行状态的展示方法、装置、计算机设备和存储介质
CN114690996A (zh) * 2020-12-31 2022-07-01 Oppo广东移动通信有限公司 内容显示方法、装置以及电子设备
CN112734523A (zh) * 2021-01-11 2021-04-30 北京城市网邻信息技术有限公司 信息展示方法、装置、电子设备和计算机可读介质
CN113158102A (zh) * 2021-02-23 2021-07-23 北京三快在线科技有限公司 页面配置方法、装置、电子设备和计算机可读介质
CN113378097A (zh) * 2021-07-08 2021-09-10 北京安天网络安全技术有限公司 一种web页面展示方法、装置、电子设备及存储介质
CN113378097B (zh) * 2021-07-08 2024-05-24 北京安天网络安全技术有限公司 一种web页面展示方法、装置、电子设备及存储介质
CN113713378A (zh) * 2021-09-02 2021-11-30 北京百度网讯科技有限公司 内容生成方法和装置
CN113794604A (zh) * 2021-09-09 2021-12-14 北京恒安嘉新安全技术有限公司 一种网络安全态势展示方法、装置、设备及存储介质
CN113835582A (zh) * 2021-09-27 2021-12-24 青岛海信移动通信技术股份有限公司 一种终端设备、信息显示方法和存储介质
CN113835582B (zh) * 2021-09-27 2024-03-15 青岛海信移动通信技术有限公司 一种终端设备、信息显示方法和存储介质

Also Published As

Publication number Publication date
CN111381909B (zh) 2022-03-04
CN111381909A (zh) 2020-07-07

Similar Documents

Publication Publication Date Title
WO2020134571A1 (zh) 页面展示方法、装置、终端设备及存储介质
US20200125919A1 (en) Virtual conversation method or system
JP2022058915A (ja) 画像認識モデルをトレーニングするための方法および装置、画像を認識するための方法および装置、電子機器、記憶媒体、並びにコンピュータプログラム
WO2021135319A1 (zh) 基于深度学习的文案生成方法、装置及电子设备
CN111666416B (zh) 用于生成语义匹配模型的方法和装置
WO2021031917A1 (zh) 视频处理方法、装置、电子设备及可读介质
CN113723087B (zh) 信息处理方法、装置、设备、可读存储介质及产品
WO2020107625A1 (zh) 视频分类方法、装置、电子设备及计算机可读存储介质
CN111432282B (zh) 一种视频推荐方法及装置
WO2020042376A1 (zh) 用于输出信息的方法和装置
US20230334880A1 (en) Hot word extraction method and apparatus, electronic device, and medium
CN111738010B (zh) 用于生成语义匹配模型的方法和装置
CN109767257B (zh) 基于大数据分析的广告投放方法、系统及电子设备
CN111737620A (zh) 一种页面访问方法、装置、设备及介质
DE112016005358T5 (de) Informations-Ranking basierend auf Eigenschaften einer Berechnungsvorrichtung
CN110475158B (zh) 视频学习素材的提供方法、装置、电子设备及可读介质
WO2022156468A1 (zh) 模型数据处理方法、装置、电子设备以及计算机可读介质
CN114417169A (zh) 信息推荐优化方法、设备、介质及程序产品
WO2024012255A1 (zh) 语义分割模型训练方法、装置、电子设备及存储介质
WO2023202361A1 (zh) 视频生成方法、装置、介质及电子设备
CN112381236A (zh) 联邦迁移学习的数据处理方法、装置、设备及存储介质
CN116522012A (zh) 用户兴趣挖掘方法、系统、电子设备以及介质
WO2023056850A1 (zh) 页面显示方法、装置、设备及存储介质
CN116958852A (zh) 视频与文本的匹配方法、装置、电子设备和存储介质
CN113706663B (zh) 图像生成方法、装置、设备及存储介质

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: 19902688

Country of ref document: EP

Kind code of ref document: A1

NENP Non-entry into the national phase

Ref country code: DE

32PN Ep: public notification in the ep bulletin as address of the adressee cannot be established

Free format text: NOTING OF LOSS OF RIGHTS PURSUANT TO RULE 112(1) EPC (EPO FORM 1205A DATED 05/10/2021)

122 Ep: pct application non-entry in european phase

Ref document number: 19902688

Country of ref document: EP

Kind code of ref document: A1