CN111639257A - Information display method, information display device, storage medium and electronic equipment - Google Patents
Information display method, information display device, storage medium and electronic equipment Download PDFInfo
- Publication number
- CN111639257A CN111639257A CN202010388476.2A CN202010388476A CN111639257A CN 111639257 A CN111639257 A CN 111639257A CN 202010388476 A CN202010388476 A CN 202010388476A CN 111639257 A CN111639257 A CN 111639257A
- Authority
- CN
- China
- Prior art keywords
- information
- display
- displayed
- user
- determining
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Withdrawn
Links
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
- G06F16/90—Details of database functions independent of the retrieved data types
- G06F16/95—Retrieval from the web
- G06F16/953—Querying, e.g. by the use of web search engines
- G06F16/9538—Presentation of query results
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
- G06F16/90—Details of database functions independent of the retrieved data types
- G06F16/95—Retrieval from the web
- G06F16/953—Querying, e.g. by the use of web search engines
- G06F16/9535—Search customisation based on user profiles and personalisation
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/21—Design or setup of recognition systems or techniques; Extraction of features in feature space; Blind source separation
- G06F18/214—Generating training patterns; Bootstrap methods, e.g. bagging or boosting
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Databases & Information Systems (AREA)
- Data Mining & Analysis (AREA)
- Physics & Mathematics (AREA)
- General Engineering & Computer Science (AREA)
- General Physics & Mathematics (AREA)
- Life Sciences & Earth Sciences (AREA)
- Artificial Intelligence (AREA)
- Evolutionary Computation (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Computational Linguistics (AREA)
- Bioinformatics & Computational Biology (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Health & Medical Sciences (AREA)
- Biomedical Technology (AREA)
- Biophysics (AREA)
- Evolutionary Biology (AREA)
- General Health & Medical Sciences (AREA)
- Molecular Biology (AREA)
- Computing Systems (AREA)
- Mathematical Physics (AREA)
- Software Systems (AREA)
- User Interface Of Digital Computer (AREA)
Abstract
The present disclosure relates to an information display method, apparatus, storage medium, and electronic device, the method comprising: responding to a page access request of a user, and acquiring user characteristic data of the user; inputting user characteristic data and characteristic data of information to be displayed into a parameter determination model aiming at each information to be displayed of a page, and obtaining various display characteristic information corresponding to the information to be displayed and output by the parameter determination model; determining a display sequence of a plurality of information to be displayed according to a plurality of display characteristic information corresponding to the information to be displayed; and displaying the plurality of information to be displayed according to the display sequence. Therefore, the model can be determined through the parameters, various display characteristic information can be output simultaneously, the accuracy of information display can be effectively guaranteed, the efficiency of calculating and displaying the characteristic information is improved, the data calculation amount is reduced, and the efficiency of information display is improved.
Description
Technical Field
The present disclosure relates to the field of computer technologies, and in particular, to an information display method and apparatus, a storage medium, and an electronic device.
Background
With the development of computer technology, various application programs are used more and more widely in daily life, and the use requirements of users are higher and higher. For example, a user may trigger a display request, and in response to the display request, the application may display corresponding information in the display interface for the user to meet the use requirement of the user. In the related art, it is common to display each piece of information by predicting factors affecting user-selected information based on a model. However, in the above process, modeling is generally performed for each factor to perform target optimization to obtain a model. Therefore, when information is displayed in response to a display request, if the information is displayed based on a single factor, the accuracy of information display is low, and if the information display needs to be performed in consideration of multiple factors, calculation needs to be performed according to multiple models, which increases the amount of resources and calculation, and reduces the response efficiency of information display.
Disclosure of Invention
An object of the present disclosure is to provide an information display method, apparatus, storage medium, and electronic device for improving efficiency of information display and response speed of a page access request.
In order to achieve the above object, according to a first aspect of the present disclosure, there is provided an information display method including:
responding to a page access request of a user, and acquiring user characteristic data of the user;
for each piece of information to be displayed of a page, inputting the user characteristic data and the characteristic data of the information to be displayed into a parameter determination model, and obtaining multiple pieces of display characteristic information corresponding to the information to be displayed output by the parameter determination model, wherein the page is provided with multiple pieces of information to be displayed, the display characteristic information is used for representing the interest degree of a user on the information to be displayed, the parameter determination model comprises multiple sub models, and each sub model is respectively used for determining one piece of display characteristic information;
determining a display sequence of the plurality of information to be displayed according to a plurality of display characteristic information corresponding to the information to be displayed;
and displaying the plurality of information to be displayed according to the display sequence.
Optionally, the determining, according to a plurality of display feature information corresponding to the information to be displayed, a display order of the plurality of information to be displayed includes:
determining display parameters corresponding to the information to be displayed according to the various display characteristic information;
and sequencing the display parameters corresponding to each piece of information to be displayed from large to small to obtain the display sequence of the plurality of pieces of information to be displayed.
Optionally, the plurality of display characteristic information includes: click rate, conversion rate, unit price.
Optionally, the parameter determination model is obtained by:
acquiring a plurality of training samples, wherein the training samples comprise historical user characteristic data, characteristic data of historical display information and actual values of various display characteristic information corresponding to the historical display information;
and training a multitask neural network model according to the training sample to obtain the parameter determination model.
Optionally, the training a multitask neural network model according to the training sample to obtain the parameter determination model includes:
inputting the historical user characteristic data and the characteristic data of the historical display information in the training sample into the multitask neural network model to obtain estimated values of the multiple kinds of display characteristic information which are output by the multitask neural network model and correspond to the historical display information;
determining the target loss of the multitask neural network model according to the estimated value and the actual value of the various display characteristic information corresponding to the historical display information;
under the condition that the target loss is not less than a preset threshold value, updating model parameters of the multitask neural network model, and re-executing the step of inputting the historical user feature data and the feature data of the historical display information in the training sample into the multitask neural network model to obtain estimated values of the multiple kinds of display feature information corresponding to the historical display information and output by the multitask neural network model to the step of determining the target loss of the multitask neural network model according to the estimated values and the actual values of the multiple kinds of display feature information corresponding to the historical display information;
and under the condition that the target loss is smaller than the preset threshold value, determining the current multitask neural network model as the parameter determination model.
Optionally, the plurality of display characteristic information includes: click rate, conversion rate and unit price;
the determining the target loss of the multitask neural network model according to the estimated value and the actual value of the multiple kinds of display characteristic information corresponding to the historical display information comprises the following steps:
determining a cross entropy determined according to the estimated click rate value and the actual click rate value as a first loss;
determining a cross entropy determined from the estimated value of the conversion rate and the actual value of the conversion rate as a second loss;
determining a mean square error determined from the estimated value of the unit price and the actual value of the unit price as a third loss;
determining a weighted sum of the first loss, the second loss, and the third loss as the target loss.
Optionally, the plurality of sub-models share an embedding layer vector.
According to a second aspect of the present disclosure, there is provided an information display apparatus, the apparatus comprising:
the first acquisition module is used for responding to a page access request of a user and acquiring user characteristic data of the user;
the input module is used for inputting the user characteristic data and the characteristic data of the information to be displayed into a parameter determination model aiming at each information to be displayed of a page, and obtaining a plurality of display characteristic information corresponding to the information to be displayed output by the parameter determination model, wherein the page is provided with a plurality of information to be displayed, the display characteristic information is used for representing the interest degree of a user on the information to be displayed, the parameter determination model comprises a plurality of sub models, and each sub model is respectively used for determining one display characteristic information;
the determining module is used for determining the display sequence of the plurality of information to be displayed according to the plurality of display characteristic information corresponding to the information to be displayed;
and the display module is used for displaying the plurality of information to be displayed according to the display sequence.
Optionally, the determining module includes:
the first determining submodule is used for determining display parameters corresponding to the information to be displayed according to the various display characteristic information;
and the sequencing submodule is used for sequencing according to the display parameters corresponding to each piece of information to be displayed from large to small to obtain the display sequence of the plurality of pieces of information to be displayed.
Optionally, the plurality of display characteristic information includes: click rate, conversion rate, unit price.
Optionally, the apparatus further comprises:
the second acquisition module is used for acquiring a plurality of training samples, wherein the training samples comprise historical user characteristic data, characteristic data of historical display information and actual values of various display characteristic information corresponding to the historical display information;
and the training module is used for training the multitask neural network model according to the training sample so as to obtain the parameter determination model.
Optionally, the training module comprises:
the input submodule is used for inputting the historical user characteristic data and the characteristic data of the historical display information in the training sample into the multitask neural network model to obtain the estimated values of the various display characteristic information which are output by the multitask neural network model and correspond to the historical display information;
the second determining submodule is used for determining the target loss of the multitask neural network model according to the estimated value and the actual value of the various display characteristic information corresponding to the historical display information;
the updating module is used for updating model parameters of the multitask neural network model under the condition that the target loss is not less than a preset threshold value, triggering the input sub-module to input the historical user characteristic data and the characteristic data of the historical display information in the training sample into the multitask neural network model, and obtaining the estimated values of the multiple kinds of display characteristic information corresponding to the historical display information and output by the multitask neural network model and the second determining sub-module to determine the target loss of the multitask neural network model according to the estimated values and the actual values of the multiple kinds of display characteristic information corresponding to the historical display information; and under the condition that the target loss is smaller than the preset threshold value, determining the current multitask neural network model as the parameter determination model.
Optionally, the plurality of display characteristic information includes: click rate, conversion rate and unit price;
the second determination submodule includes:
a third determining submodule, configured to determine, as a first loss, a cross entropy determined according to the estimated click rate value and the actual click rate value;
a fourth determination submodule for determining a cross entropy determined from the estimated value of the conversion rate and the actual value of the conversion rate as a second loss;
a fifth determining submodule for determining a mean square error determined from the estimated value of the unit price and the actual value of the unit price as a third loss;
a sixth determining submodule configured to determine a weighted sum of the first loss, the second loss, and the third loss as the target loss.
Optionally, the plurality of sub-models share an embedding layer vector.
According to a third aspect of the present disclosure, there is provided a computer readable storage medium having stored thereon a computer program which, when executed by a processor, performs the steps of any of the methods of the first aspect described above.
According to a fourth aspect of the present disclosure, there is provided an electronic device comprising:
a memory having a computer program stored thereon;
a processor for executing the computer program in the memory to implement the steps of the method of any of the first aspects above.
In the technical scheme, the user characteristic data of the user is acquired in response to the page access request of the user; the user characteristic data and the characteristic data of the information to be displayed are input into the parameter determination model aiming at each information to be displayed of the page, so that various display characteristic information corresponding to the information to be displayed can be obtained, the display sequence of the information to be displayed can be determined according to the various display characteristic information corresponding to the information to be displayed, and the information to be displayed is displayed according to the display sequence. Therefore, according to the technical scheme, the model can be determined through the parameters, and the various display characteristic information can be output at the same time, namely, the factors influencing the selection of the user can be output at the same time, so that the accuracy of information display can be effectively ensured. Compared with the prior art that each factor is output through a plurality of models respectively, the method disclosed by the invention can ensure the matching degree among the plurality of display characteristic information, further improve the accuracy of the information, improve the efficiency of calculating and displaying the characteristic information, reduce the data calculation amount, save the occupation of calculation resources, improve the efficiency of information display and the response speed of page access requests, and improve the use experience of users.
Additional features and advantages of the disclosure will be set forth in the detailed description which follows.
Drawings
The accompanying drawings, which are included to provide a further understanding of the disclosure and are incorporated in and constitute a part of this specification, illustrate embodiments of the disclosure and together with the description serve to explain the disclosure without limiting the disclosure. In the drawings:
FIG. 1 is a flow chart of an information display method provided according to one embodiment of the present disclosure;
FIG. 2 is a flow diagram of an exemplary implementation for determining a display order of a plurality of information to be displayed according to a plurality of display characteristic information corresponding to the information to be displayed, provided according to an embodiment of the present disclosure;
FIG. 3 is a schematic diagram of a multitasking neural network model provided in accordance with one embodiment of the present disclosure;
FIG. 4 is a block diagram of an information display device provided in accordance with one embodiment of the present disclosure;
FIG. 5 is a block diagram illustrating an electronic device in accordance with an exemplary embodiment;
FIG. 6 is a block diagram illustrating an electronic device in accordance with an example embodiment.
Detailed Description
The following detailed description of specific embodiments of the present disclosure is provided in connection with the accompanying drawings. It should be understood that the detailed description and specific examples, while indicating the present disclosure, are given by way of illustration and explanation only, not limitation.
Fig. 1 is a flowchart of an information display method according to an embodiment of the present disclosure, where as shown in fig. 1, the method includes:
in step 11, user characteristic data of the user is acquired in response to a page access request of the user.
Wherein the user characteristic data may include, but is not limited to, one or more of: user identification, age, gender, height, weight, education level, occupation, etc. For example, the user characteristic data may be obtained from user registration information, and in response to the page access request, the user characteristic data may be obtained by querying the user registration information according to the user ID in the page access request.
In step 12, for each piece of information to be displayed of the page, inputting the user characteristic data and the characteristic data of the information to be displayed to a parameter determination model, and obtaining multiple pieces of display characteristic information corresponding to the information to be displayed output by the parameter determination model, where the page has multiple pieces of information to be displayed, the page is a page corresponding to the page access request, the display characteristic information is used to represent the interest degree of the user in the information to be displayed, the parameter determination model includes multiple submodels, and each submodel is used to determine one piece of the display characteristic information.
Wherein, the information to be displayed may be determined according to an actual usage platform, for example, in a website search application scenario, the information to be displayed may be website information, and the feature data of the information to be displayed may include, but is not limited to, one or more of the following: website identification, website type, website security level, website scale, and the like. As another example, in an outsourcing scenario, the information to be displayed may be merchant information, and the characteristic data of the information to be displayed may include, but is not limited to, one or more of the following: merchant identification, items contained by the merchant, merchant security level, merchant location, merchant sales volume, and the like. The foregoing is merely exemplary and is not intended to limit the present disclosure.
The multiple kinds of display characteristic information can be used for representing the interest degree of the user for the information to be displayed, and each kind of display characteristic information is a factor for selection. In the step, through the parameter determination model, various display characteristic information corresponding to the information to be displayed can be directly obtained according to the user characteristic data and the characteristic data of the information to be displayed, so that the adaptability among the various display characteristic information can be ensured, and the efficiency of determining the display characteristic information can also be ensured.
In step 13, a display order of the plurality of information to be displayed is determined according to the plurality of display characteristic information corresponding to the information to be displayed.
In step 14, a plurality of pieces of information to be displayed are displayed in the display order.
And responding to a page access request of a user, and displaying a plurality of pieces of information to be displayed. In the embodiment of the disclosure, a plurality of pieces of information to be displayed are displayed based on a plurality of pieces of display characteristic information corresponding to each piece of information to be displayed, that is, it can be ensured to display information that a user is interested in preferentially to a certain extent, so that the number of pieces of information viewed by the user can be effectively reduced, and the accuracy of the information provided for the user can be ensured.
In the technical scheme, the user characteristic data of the user is acquired in response to the page access request of the user; the user characteristic data and the characteristic data of the information to be displayed are input into the parameter determination model aiming at each information to be displayed of the page, so that various display characteristic information corresponding to the information to be displayed can be obtained, the display sequence of the information to be displayed can be determined according to the various display characteristic information corresponding to the information to be displayed, and the information to be displayed is displayed according to the display sequence. Therefore, according to the technical scheme, the model can be determined through the parameters, and the various display characteristic information can be output at the same time, namely, the factors influencing the selection of the user can be output at the same time, so that the accuracy of information display can be effectively ensured. Compared with the prior art that each factor is output through a plurality of models respectively, the method disclosed by the invention can ensure the matching degree among the plurality of display characteristic information, further improve the accuracy of the information, improve the efficiency of calculating and displaying the characteristic information, reduce the data calculation amount, save the occupation of calculation resources, improve the efficiency of information display and the response speed of page access requests, and improve the use experience of users.
Optionally, an exemplary implementation manner of determining the display order of the multiple pieces of information to be displayed according to the multiple pieces of display characteristic information corresponding to the information to be displayed in step 13 is as follows, as shown in fig. 2, the step may include:
in step 21, according to the various display characteristic information, the display parameters corresponding to the information to be displayed are determined.
As an example, the result of weighted summation of the plurality of kinds of display characteristic information may be determined as the display parameter. The weight corresponding to each piece of display characteristic information can be preset according to the actual use scene.
As another example, the plurality of display characteristic information may include: click rate, conversion rate, unit price. Accordingly, the product of multiple kinds of display characteristic information corresponding to the information to be displayed may be determined as the display parameter corresponding to the information to be displayed, for example:
S＝CTR*CVR*P；
wherein S is used for representing the display parameter;
the CTR (Click-Through-Rate) is used for representing the Click Rate and can be used for representing the probability of clicking on the displayed information by the user;
the cvr (conversion rate) is used for representing a conversion rate, and may represent a probability that a user invests clicked information, for example, in a scene recommended by a website, after the user clicks to enter the website, the user registers to become a member on the website;
p is a unit price representing the cost of the predicted investment of the displayed information by the user in a single click on the information, for example, in a site recommendation scenario, the user clicks into the site and becomes a member's predicted investment cost.
In the embodiment, the click, conversion and predicted investment of the user on the display information are considered, so that the display parameters can more comprehensively balance the user experience and the use experience of the information provider, and data support can be provided for improving the accuracy of the information to be displayed.
In the scenario of displaying the information of the take-away merchant, the merchant experience and the user experience may be effectively balanced by the above formula, and further, for the relationship between the merchant experience, the user experience and the display platform experience, the display parameter may be determined by the following formula:
S＝CTR*CVR*P+CTR*Bid；
wherein, the Bid can be used for the value that the user can obtain for the merchant information one-click display platform.
In step 22, the display parameters corresponding to each piece of information to be displayed are sorted from large to small to obtain a display order of the pieces of information to be displayed.
In the embodiment of the present disclosure, when the specific attribute value of each piece of display characteristic information is quantized, the quantization is performed in a manner of positively correlating with the user's interest level, for example, the larger the click rate is, the more the user is interested in. Therefore, when the display parameters are determined, the display parameters can be directly sorted from big to small, the determination mode of the display sequence of the information to be displayed can be simplified, and the information display efficiency can be further improved.
Therefore, by the technical scheme, the display sequence of each information to be displayed can be simply, conveniently and quickly determined, and the display sequence is made to be adaptive to the interest of the user, so that the use requirement of the user can be met when each information to be displayed is displayed, the time and the information viewing amount required by the user for selecting the information required by the user are reduced, the accuracy of information display is improved, and the use experience of the user is improved.
Alternatively, the parameter determination model may be obtained by:
obtaining a plurality of training samples, wherein the training samples comprise historical user characteristic data, characteristic data of historical display information and actual values of various display characteristic information corresponding to the historical display information.
Historical data of the user operating on the displayed information can be acquired from the information display platform or system, and the historical data is processed to obtain the training sample. For example, for display characteristic information whose value is binary, such as whether the user clicks the information, if the user clicks, the actual value of the display characteristic information may be represented as 1, and if the user does not click, the actual value of the display characteristic information may be represented as 0. And aiming at the display characteristic information with the continuity value, the actual value can be directly represented by the value. Through the processing, the representation of the actual values of various display characteristic information can be simplified, and the complexity of subsequent calculation is reduced conveniently.
And training the multitask neural network model according to the training sample to obtain a parameter determination model, wherein the multitask neural network model comprises a plurality of submodels, and each submodel is respectively used for determining display characteristic information.
The multitask neural network model is a neural network model based on multitask Learning (MTL). In the present disclosure, each display characteristic information corresponds to a sub-model, one sub-model corresponding to learning and target optimization of one sub-task in multi-task learning.
Therefore, through the technical scheme, the learning and optimization of various display characteristic information can be completed in the training process of a model, on one hand, the workload for training a plurality of models can be effectively reduced, on the other hand, the various display characteristic information may have an association relationship, such as the click rate and the conversion rate, and the situation that various display characteristic information optimized by a single target in the prior art is not matched can be avoided, so that the determined various display characteristic information is matched, the accuracy of the various display characteristic information is improved, and data support is provided for improving the accuracy of information display.
In addition, in the prior art, a single-target optimization is performed on each display characteristic information individual training model, for example, when a click rate model is trained, data used for training of the model is click data, that is, training is performed based on a sample clicked by a user; when the conversion rate model is trained, the data used for training the model is conversion data, that is, training is performed based on a sample that a user clicks and puts in. When the model is applied, the model is tested in a full sample space, but the training samples of each single-target model are smaller than the full samples in number, that is, the ratio of the information clicked and converted by the user to all the displayed information is smaller, so that the trained model has insufficient applicability and low accuracy when the model is predicted in the full sample space.
In the embodiment of the disclosure, when the parameter determination model is trained, the plurality of training samples include characteristic data of history display information, for example, a click rate sub-model, wherein the history display information clicked by a user may be used as a positive sample, and the history display information not clicked by the user may be used as a negative sample, so that the click rate sub-model may be trained in a full sample space, thereby effectively improving the accuracy of the parameter determination model, ensuring the distribution consistency of the training samples and the test samples, and improving the generalization of the parameter determination model.
Optionally, the plurality of sub-models share an embedding layer vector. Therefore, the multiple submodels can be trained on the basis of the same sample space, and matching and adaptability among the multiple submodels can be guaranteed. And aiming at the model trained based on the sparse training sample, each sub-model can be trained in the whole sample space through the multitask neural network model, so that the accuracy and effectiveness of the parameter determination model can be further improved, and support is provided for improving the accuracy of information display.
Optionally, an exemplary implementation of training the multitask neural network model according to the training sample to obtain the parameter determination model is as follows, and the step may include:
inputting the historical user characteristic data and the characteristic data of the historical display information in the training sample into the multitask neural network model to obtain the estimated values of the multiple kinds of display characteristic information which are output by the multitask neural network model and correspond to the historical display information.
The historical user characteristic data and the characteristic data of the historical display information can be connected to obtain an input vector, and the input vector is input to the multitask neural network model. For example, the discrete data and the continuous data of the feature data of the historical user feature data and the historical display information may be respectively subjected to vectorization processing, and then the obtained discrete data vector and the obtained continuous data vector are subjected to concatemate connection to obtain an input vector. The vectorization processing method for discrete data and continuous data is the prior art, and is not described herein again.
And then, determining the target loss of the multitask neural network model according to the estimated values and the actual values of the various display characteristic information corresponding to the historical display information.
As an example, the loss for each submodel in the multitasking neural network model may be determined separately, such that a target loss for the multitasking neural network model may be determined from the loss for each submodel.
Further, the plurality of display characteristic information includes: click rate, conversion rate and unit price;
an exemplary implementation manner of determining the target loss of the multitask neural network model according to the estimated value and the actual value of the multiple kinds of display characteristic information corresponding to the historical display information is as follows, and the step may include:
determining a cross entropy determined according to the estimated click rate value and the actual click rate value as a first loss;
determining a cross entropy determined from the estimated value of the conversion rate and the actual value of the conversion rate as a second loss;
determining a mean square error determined from the estimated value of the unit price and the actual value of the unit price as a third loss;
determining a weighted sum of the first loss, the second loss, and the third loss as the target loss.
In the embodiment of the present disclosure, for displayed information, such as merchant information displayed in a take-away scene, the corresponding user operation may be user click or not click, and for merchant information clicked by a user, the corresponding user operation may be order placing or not placing, that is, whether the merchant information clicked by the user is converted or not. Based on this, the click-through rate submodel and the conversion rate submodel may be classification models. In the disclosure, the loss of the classification model is calculated by adopting cross entropy, and the cross entropy can be used for representing the difference between the real probability distribution and the estimated probability distribution, so that the calculation of the loss of the sub model can improve the accuracy of the sub model. And through cross entropy calculation loss, the activation function can be reduced quickly when the error is large, the convergence efficiency of the submodel is improved, and the training efficiency of the submodel is further improved.
The unit price forming submodel may be a regression model, and for the unit price forming submodel, a mean-square error (MSE) calculated as an estimated value of the unit price and an actual value of the unit price forming submodel may be used to calculate a loss of the submodel to ensure accuracy and training efficiency of the submodel, for example, taking a multitask neural network model including the click rate submodel, the conversion rate submodel, and the unit price forming submodel as an example, a schematic diagram of the multitask neural network model is shown in fig. 3, where the click rate submodel a1, the conversion rate submodel a2, and the unit price forming submodel A3 share an Embedding layer vector Embedding. For example, as described above, the historical display data may be preprocessed, as shown in the input layer B1 in fig. 3, that is, when the training samples are input into the multitasking neural network model, the discrete data and the continuous data of the historical user feature data and the feature data of the historical display information may be vectorized through the input layer, and then the obtained discrete data vector and the obtained continuous data vector are configured to obtain the input vector. Each submodel may include a plurality of FC full-link layers, and then the obtained input vectors are input into the click rate submodel a1, the conversion rate submodel a2, and the unit cost submodel A3, so as to obtain an output result of each submodel as an output result of the multitask neural network model.
The weights of the first loss, the second loss and the third loss may be preset, or may be adjusted according to an actual usage scenario, which is not limited in this disclosure. The calculation method of the cross entropy and the mean square error is the prior art, and is not described herein again.
Therefore, the loss calculation mode corresponding to each submodel in the multitask neural network model can be set respectively, the prediction accuracy of each submodel and the training efficiency of each submodel can be effectively guaranteed, the overall loss of the multitask neural network model is determined based on the loss of each submodel, the accuracy of the determined loss of the multitask neural network model can be guaranteed, and accurate data support is provided for training of the multitask neural network model.
After the target loss is determined, under the condition that the target loss is not less than a preset threshold value, updating model parameters of the multitask neural network model, and re-executing the step of inputting the historical user characteristic data and the characteristic data of the historical display information in the training sample into the multitask neural network model to obtain estimated values of the multiple kinds of display characteristic information corresponding to the historical display information and output by the multitask neural network model to the step of determining the target loss of the multitask neural network model according to the estimated values and the actual values of the multiple kinds of display characteristic information corresponding to the historical display information;
and under the condition that the target loss is smaller than the preset threshold value, determining the current multitask neural network model as the parameter determination model.
The preset threshold may be set according to an actual usage scenario, which is not limited by the present disclosure. Under the condition that the target loss is not less than the preset threshold value, the precision of the multi-task neural network model is not up to the standard, and at the moment, the model parameters of the multi-task neural network model can be updated, namely the parameters of each sub-model in the multi-task neural network model are updated, so that the estimated value output by each sub-model is closer to the actual value of the sub-model. When the target loss is smaller than the preset threshold value, the precision of the multitask neural network model is shown to reach the standard, at the moment, the training of the multitask neural network model is finished, the trained multitask neural network model is used as a parameter determination model, and accurate display characteristic information can be obtained according to the user characteristic information and the characteristic data of the information to be displayed when an access request of a user is received.
According to the technical scheme, parameter adjustment of each submodel can be achieved based on the loss of the multitask neural network model, and the target loss is determined based on the loss of each submodel, so that the loss of other submodels can be comprehensively considered when the parameter of each submodel is adjusted, the adaptability of each submodel is guaranteed, and the accuracy of the output result of the multitask neural network model is improved.
The present disclosure also provides an information display apparatus, as shown in fig. 4, the apparatus 10 including:
a first obtaining module 100, configured to obtain user characteristic data of a user in response to a page access request of the user;
an input module 200, configured to input, for each piece of information to be displayed of a page, the user feature data and feature data of the piece of information to be displayed into a parameter determination model, and obtain multiple pieces of display feature information corresponding to the piece of information to be displayed output by the parameter determination model, where the page has multiple pieces of information to be displayed, the display feature information is used to represent a degree of interest of a user in the piece of information to be displayed, the parameter determination model includes multiple submodels, and each submodel is used to determine one piece of the display feature information;
a determining module 300, configured to determine a display order of the multiple pieces of information to be displayed according to multiple pieces of display characteristic information corresponding to the information to be displayed;
a display module 400, configured to display the multiple pieces of information to be displayed according to the display sequence.
Optionally, the determining module includes:
the first determining submodule is used for determining display parameters corresponding to the information to be displayed according to the various display characteristic information;
and the sequencing submodule is used for sequencing according to the display parameters corresponding to each piece of information to be displayed from large to small to obtain the display sequence of the plurality of pieces of information to be displayed.
Optionally, the plurality of display characteristic information includes: click rate, conversion rate, unit price.
Optionally, the apparatus further comprises:
the second acquisition module is used for acquiring a plurality of training samples, wherein the training samples comprise historical user characteristic data, characteristic data of historical display information and actual values of various display characteristic information corresponding to the historical display information;
and the training module is used for training the multitask neural network model according to the training sample so as to obtain the parameter determination model.
Optionally, the training module comprises:
the input submodule is used for inputting the historical user characteristic data and the characteristic data of the historical display information in the training sample into the multitask neural network model to obtain the estimated values of the various display characteristic information which are output by the multitask neural network model and correspond to the historical display information;
the second determining submodule is used for determining the target loss of the multitask neural network model according to the estimated value and the actual value of the various display characteristic information corresponding to the historical display information;
the updating module is used for updating model parameters of the multitask neural network model under the condition that the target loss is not less than a preset threshold value, triggering the input sub-module to input the historical user characteristic data and the characteristic data of the historical display information in the training sample into the multitask neural network model, and obtaining the estimated values of the multiple kinds of display characteristic information corresponding to the historical display information and output by the multitask neural network model and the second determining sub-module to determine the target loss of the multitask neural network model according to the estimated values and the actual values of the multiple kinds of display characteristic information corresponding to the historical display information; and under the condition that the target loss is smaller than the preset threshold value, determining the current multitask neural network model as the parameter determination model.
Optionally, the plurality of display characteristic information includes: click rate, conversion rate and unit price;
the second determination submodule includes:
a third determining submodule, configured to determine, as a first loss, a cross entropy determined according to the estimated click rate value and the actual click rate value;
a fourth determination submodule for determining a cross entropy determined from the estimated value of the conversion rate and the actual value of the conversion rate as a second loss;
a fifth determining submodule for determining a mean square error determined from the estimated value of the unit price and the actual value of the unit price as a third loss;
a sixth determining submodule configured to determine a weighted sum of the first loss, the second loss, and the third loss as the target loss.
Optionally, the plurality of sub-models share an embedding layer vector.
With regard to the apparatus in the above-described embodiment, the specific manner in which each module performs the operation has been described in detail in the embodiment related to the method, and will not be elaborated here.
Fig. 5 is a block diagram illustrating an electronic device 700 according to an example embodiment. As shown in fig. 5, the electronic device 700 may include: a processor 701 and a memory 702. The electronic device 700 may also include one or more of a multimedia component 703, an input/output (I/O) interface 704, and a communication component 705.
The processor 701 is configured to control the overall operation of the electronic device 700, so as to complete all or part of the steps in the information display method. The memory 702 is used to store various types of data to support operation at the electronic device 700, such as instructions for any application or method operating on the electronic device 700 and application-related data, such as contact data, transmitted and received messages, pictures, audio, video, and the like. The Memory 702 may be implemented by any type of volatile or non-volatile Memory device or combination thereof, such as Static Random Access Memory (SRAM), Electrically Erasable Programmable Read-Only Memory (EEPROM), Erasable Programmable Read-Only Memory (EPROM), Programmable Read-Only Memory (PROM), Read-Only Memory (ROM), magnetic Memory, flash Memory, magnetic disk, or optical disk. The multimedia components 703 may include screen and audio components. Wherein the screen may be, for example, a touch screen and the audio component is used for outputting and/or inputting audio signals. For example, the audio component may include a microphone for receiving external audio signals. The received audio signal may further be stored in the memory 702 or transmitted through the communication component 705. The audio assembly also includes at least one speaker for outputting audio signals. The I/O interface 704 provides an interface between the processor 701 and other interface modules, such as a keyboard, mouse, buttons, etc. These buttons may be virtual buttons or physical buttons. The communication component 705 is used for wired or wireless communication between the electronic device 700 and other devices. Wireless communication, such as Wi-Fi, bluetooth, Near Field Communication (NFC), 2G, 3G, 4G, NB-IOT, eMTC, or other 5G, etc., or a combination of one or more of them, which is not limited herein. The corresponding communication component 705 may thus include: Wi-Fi module, Bluetooth module, NFC module, etc.
In an exemplary embodiment, the electronic Device 700 may be implemented by one or more Application Specific Integrated Circuits (ASICs), Digital Signal Processors (DSPs), Digital Signal Processing Devices (DSPDs), Programmable Logic Devices (PLDs), Field Programmable Gate Arrays (FPGAs), controllers, microcontrollers, microprocessors, or other electronic components for performing the above-described information display method.
In another exemplary embodiment, there is also provided a computer readable storage medium including program instructions which, when executed by a processor, implement the steps of the information display method described above. For example, the computer readable storage medium may be the memory 702 described above including program instructions that are executable by the processor 701 of the electronic device 700 to perform the information display method described above.
Fig. 6 is a block diagram illustrating an electronic device 1900 according to an example embodiment. For example, the electronic device 1900 may be provided as a server. Referring to fig. 6, an electronic device 1900 includes a processor 1922, which may be one or more in number, and a memory 1932 for storing computer programs executable by the processor 1922. The computer program stored in memory 1932 may include one or more modules that each correspond to a set of instructions. Further, the processor 1922 may be configured to execute the computer program to perform the information display method described above.
Additionally, electronic device 1900 may also include a power component 1926 and a communication component 1950, the power component 1926 may be configured to perform power management of the electronic device 1900, and the communication component 1950 may be configured to enable communication, e.g., wired or wireless communication, of the electronic device 1900. In addition, the electronic device 1900 may also include input/output (I/O) interfaces 1958. The electronic device 1900 may operate based on an operating system, such as Windows Server, Mac OS XTM, UnixTM, Linux, etc., stored in memory 1932.
In another exemplary embodiment, there is also provided a computer readable storage medium including program instructions which, when executed by a processor, implement the steps of the information display method described above. For example, the computer readable storage medium may be the memory 1932 that includes program instructions executable by the processor 1922 of the electronic device 1900 to perform the information display method described above.
In another exemplary embodiment, a computer program product is also provided, which comprises a computer program executable by a programmable apparatus, the computer program having code portions for performing the above-mentioned information display method when executed by the programmable apparatus.
The preferred embodiments of the present disclosure are described in detail with reference to the accompanying drawings, however, the present disclosure is not limited to the specific details of the above embodiments, and various simple modifications may be made to the technical solution of the present disclosure within the technical idea of the present disclosure, and these simple modifications all belong to the protection scope of the present disclosure.
It should be noted that the various features described in the above embodiments may be combined in any suitable manner without departing from the scope of the invention. In order to avoid unnecessary repetition, various possible combinations will not be separately described in this disclosure.
In addition, any combination of various embodiments of the present disclosure may be made, and the same should be considered as the disclosure of the present disclosure, as long as it does not depart from the spirit of the present disclosure.
Claims (10)
1. An information display method, characterized in that the method comprises:
responding to a page access request of a user, and acquiring user characteristic data of the user;
for each piece of information to be displayed of a page, inputting the user characteristic data and the characteristic data of the information to be displayed into a parameter determination model, and obtaining multiple pieces of display characteristic information corresponding to the information to be displayed output by the parameter determination model, wherein the page is provided with multiple pieces of information to be displayed, the display characteristic information is used for representing the interest degree of a user on the information to be displayed, the parameter determination model comprises multiple sub models, and each sub model is respectively used for determining one piece of display characteristic information;
determining a display sequence of the plurality of information to be displayed according to a plurality of display characteristic information corresponding to the information to be displayed;
and displaying the plurality of information to be displayed according to the display sequence.
2. The method according to claim 1, wherein the determining the display order of the plurality of information to be displayed according to a plurality of display characteristic information corresponding to the information to be displayed comprises:
determining display parameters corresponding to the information to be displayed according to the various display characteristic information;
and sequencing the display parameters corresponding to each piece of information to be displayed from large to small to obtain the display sequence of the plurality of pieces of information to be displayed.
3. The method of claim 1, wherein the plurality of display characteristic information comprises: click rate, conversion rate, unit price.
4. The method of claim 1, wherein the parameter determination model is obtained by:
acquiring a plurality of training samples, wherein the training samples comprise historical user characteristic data, characteristic data of historical display information and actual values of various display characteristic information corresponding to the historical display information;
and training a multitask neural network model according to the training sample to obtain the parameter determination model.
5. The method of claim 4, wherein training a multitask neural network model to obtain the parameter determination model according to the training samples comprises:
inputting the historical user characteristic data and the characteristic data of the historical display information in the training sample into the multitask neural network model to obtain estimated values of the multiple kinds of display characteristic information which are output by the multitask neural network model and correspond to the historical display information;
determining the target loss of the multitask neural network model according to the estimated value and the actual value of the various display characteristic information corresponding to the historical display information;
under the condition that the target loss is not less than a preset threshold value, updating model parameters of the multitask neural network model, and re-executing the step of inputting the historical user feature data and the feature data of the historical display information in the training sample into the multitask neural network model to obtain estimated values of the multiple kinds of display feature information corresponding to the historical display information and output by the multitask neural network model to the step of determining the target loss of the multitask neural network model according to the estimated values and the actual values of the multiple kinds of display feature information corresponding to the historical display information;
and under the condition that the target loss is smaller than the preset threshold value, determining the current multitask neural network model as the parameter determination model.
6. The method of claim 5, wherein the plurality of display characteristic information comprises: click rate, conversion rate and unit price;
the determining the target loss of the multitask neural network model according to the estimated value and the actual value of the multiple kinds of display characteristic information corresponding to the historical display information comprises the following steps:
determining a cross entropy determined according to the estimated click rate value and the actual click rate value as a first loss;
determining a cross entropy determined from the estimated value of the conversion rate and the actual value of the conversion rate as a second loss;
determining a mean square error determined from the estimated value of the unit price and the actual value of the unit price as a third loss;
determining a weighted sum of the first loss, the second loss, and the third loss as the target loss.
7. The method of claim 1, wherein the plurality of submodels share an embedding layer vector.
8. An information display apparatus, characterized in that the apparatus comprises:
the first acquisition module is used for responding to a page access request of a user and acquiring user characteristic data of the user;
the input module is used for inputting the user characteristic data and the characteristic data of the information to be displayed into a parameter determination model aiming at each information to be displayed of a page, and obtaining a plurality of display characteristic information corresponding to the information to be displayed output by the parameter determination model, wherein the page is provided with a plurality of information to be displayed;
the determining module is used for determining the display sequence of the plurality of information to be displayed according to the plurality of display characteristic information corresponding to the information to be displayed;
and the display module is used for displaying the plurality of information to be displayed according to the display sequence.
9. A computer-readable storage medium, on which a computer program is stored which, when being executed by a processor, carries out the steps of the method according to any one of claims 1 to 7.
10. An electronic device, comprising:
a memory having a computer program stored thereon;
a processor for executing the computer program in the memory to carry out the steps of the method of any one of claims 1 to 7.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202010388476.2A CN111639257A (en) | 2020-05-09 | 2020-05-09 | Information display method, information display device, storage medium and electronic equipment |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202010388476.2A CN111639257A (en) | 2020-05-09 | 2020-05-09 | Information display method, information display device, storage medium and electronic equipment |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN111639257A true CN111639257A (en) | 2020-09-08 |
Family
ID=72333191
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202010388476.2A Withdrawn CN111639257A (en) | 2020-05-09 | 2020-05-09 | Information display method, information display device, storage medium and electronic equipment |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN111639257A (en) |
Citations (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN108920717A (en) * | 2018-07-27 | 2018-11-30 | 百度在线网络技术（北京）有限公司 | For showing the method and device of information |
| CN109684510A (en) * | 2018-10-31 | 2019-04-26 | 北京达佳互联信息技术有限公司 | Video sequencing method, device, electronic equipment and storage medium |
| CN110458649A (en) * | 2019-07-11 | 2019-11-15 | 北京三快在线科技有限公司 | Information recommendation method, device, electronic equipment and readable storage medium storing program for executing |
| CN110866199A (en) * | 2019-10-17 | 2020-03-06 | 北京三快在线科技有限公司 | Position determination method, device, electronic equipment and computer readable medium |
| CN111080338A (en) * | 2019-11-11 | 2020-04-28 | 中国建设银行股份有限公司 | User data processing method and device, electronic equipment and storage medium |
| CN111078998A (en) * | 2019-11-19 | 2020-04-28 | Oppo（重庆）智能科技有限公司 | Information retrieval method, information retrieval device, storage medium and server |
| CN111125521A (en) * | 2019-12-13 | 2020-05-08 | 上海喜马拉雅科技有限公司 | Information recommendation method, device, equipment and storage medium |
-
2020
- 2020-05-09 CN CN202010388476.2A patent/CN111639257A/en not_active Withdrawn
Patent Citations (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN108920717A (en) * | 2018-07-27 | 2018-11-30 | 百度在线网络技术（北京）有限公司 | For showing the method and device of information |
| CN109684510A (en) * | 2018-10-31 | 2019-04-26 | 北京达佳互联信息技术有限公司 | Video sequencing method, device, electronic equipment and storage medium |
| CN110458649A (en) * | 2019-07-11 | 2019-11-15 | 北京三快在线科技有限公司 | Information recommendation method, device, electronic equipment and readable storage medium storing program for executing |
| CN110866199A (en) * | 2019-10-17 | 2020-03-06 | 北京三快在线科技有限公司 | Position determination method, device, electronic equipment and computer readable medium |
| CN111080338A (en) * | 2019-11-11 | 2020-04-28 | 中国建设银行股份有限公司 | User data processing method and device, electronic equipment and storage medium |
| CN111078998A (en) * | 2019-11-19 | 2020-04-28 | Oppo（重庆）智能科技有限公司 | Information retrieval method, information retrieval device, storage medium and server |
| CN111125521A (en) * | 2019-12-13 | 2020-05-08 | 上海喜马拉雅科技有限公司 | Information recommendation method, device, equipment and storage medium |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| WO2023061087A1 (en) | Information recommendation method and apparatus, and electronic device, computer-readable storage medium and computer program product | |
| US10958748B2 (en) | Resource push method and apparatus | |
| WO2021047593A1 (en) | Method for training recommendation model, and method and apparatus for predicting selection probability | |
| CN110766142A (en) | Model generation method and device | |
| CN109345302A (en) | Machine learning model training method, device, storage medium and computer equipment | |
| CN112579909A (en) | Object recommendation method and device, computer equipment and medium | |
| JP7263463B2 (en) | Method, device, electronic device, storage medium and computer program for determining recommended model and determining product price | |
| CN113919923B (en) | Live broadcast recommendation model training method, live broadcast recommendation method and related equipment | |
| CN110910201B (en) | Information recommendation control method and device, computer equipment and storage medium | |
| CN110796513A (en) | Multitask learning method and device, electronic equipment and storage medium | |
| CN110766513A (en) | Information sorting method and device, electronic equipment and readable storage medium | |
| CN114240555A (en) | Click rate prediction model training method and device and click rate prediction method and device | |
| CN113761348A (en) | Information recommendation method and device, electronic equipment and storage medium | |
| CN112784998A (en) | Data processing method and device and computing equipment | |
| CN116684330A (en) | Traffic prediction method, device, equipment and storage medium based on artificial intelligence | |
| CN108667877B (en) | Method and device for determining recommendation information, computer equipment and storage medium | |
| CN114791982A (en) | Object recommendation method and device | |
| CN114360027A (en) | Training method and device for feature extraction network and electronic equipment | |
| CN113869596A (en) | Task prediction processing method, device, product and medium | |
| US12062060B2 (en) | Method and device for processing user interaction information | |
| CN112906896A (en) | Information processing method and device and computing equipment | |
| CN114511095A (en) | Data processing method and device, computing equipment and storage medium | |
| CN116956204A (en) | Network structure determining method, data predicting method and device of multi-task model | |
| CN111639257A (en) | Information display method, information display device, storage medium and electronic equipment | |
| CN115796984A (en) | Training method of item recommendation model, storage medium and related equipment |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| WW01 | Invention patent application withdrawn after publication | ||
| WW01 | Invention patent application withdrawn after publication |
Application publication date: 20200908 |