Method and device for predicting the attention of works

Abstract

The present disclosure relates to a method and device for predicting the attention of a work, the method comprising: obtaining the previous attention data of the work and the attribute data affecting the attention of the work; inputting the previous attention data and the attribute data of the work as features into a first prediction model to obtain a first attention prediction value for the work; inputting the first attention prediction value into a second prediction model to obtain a second attention prediction value for the work as an attention prediction for the work. The prediction method according to the present disclosure can improve the prediction accuracy of the attention of the work.

Description

Technical Field

[0001] The present disclosure relates to the field of computer technology, and more specifically, to a method and device for training an attention prediction model and a method and device for predicting the attention of a work using the model. Background Art

[0002] After a multimedia work (e.g., short video content, audio content) is uploaded by the author to the work playback platform, users of the platform can watch or listen to the work. If a user becomes interested in the author of the work after watching the work and follows the author, the author's attention will increase. For example, the attention of a work can usually be evaluated by the number of times the work is viewed and the number of times the author is followed, that is, the attention of a work = the number of times the work is followed / the number of times the work is viewed.

[0003] For the prediction of the attention of a work, the attention data of the work in all previous time periods and some attribute data of the work can usually be used to predict the attention of the current period. If the prediction result is inaccurate, it is often processed from the model parameter adjustment or data distribution level, and the prediction result is uncontrollable and costly. Summary of the invention

[0004] The present disclosure provides a method and device for predicting the attention of a work, so as to at least solve the problem of low prediction accuracy in the above-mentioned related technologies.

[0005] According to a first aspect of an embodiment of the present disclosure, a training method for an attention prediction model is provided, comprising: obtaining previous attention data of a plurality of works and attribute data affecting the attention of the works; inputting the previous attention data and attribute data of the plurality of works as features into a first prediction model to obtain a first predicted attention value for each work, and adjusting the parameters of the first prediction model based on the first predicted attention value and the real attention value of the plurality of works; forming a sample pair with the real attention value and the first predicted attention value of each work in the plurality of works, dividing the plurality of sample pairs corresponding to the plurality of works into M groups in a predetermined order, and respectively determining the mean of the real attention values ​​and the mean of the first predicted attention values ​​of all sample pairs in each of the M groups as the second real attention value and the second predicted attention value of the new M sample data; inputting the second predicted attention value of each sample data of the M sample data as features into a second prediction model to obtain a third predicted attention value for each sample data of the M sample data, and adjusting the value of M and the parameters of the second attention model based on the third predicted attention value and the second real attention value of each sample data of the M sample data.

[0006] According to the first aspect of an embodiment of the present disclosure, dividing the multiple sample pairs into M groups in a predetermined order includes: arranging the multiple samples in ascending or descending order according to the real attention values ​​or the first predicted attention values ​​of the multiple sample pairs; and evenly dividing the arranged multiple samples into the M groups.

[0007] According to the first aspect of the embodiment of the present disclosure, the first prediction model is a linear model or a nonlinear model having predetermined weights for previous attention data and attribute data, wherein adjusting the parameters of the first prediction model includes: calculating an evaluation index based on the first predicted attention values ​​and actual attention values ​​of the multiple works; and adjusting the predetermined weights of the first prediction model according to the evaluation index.

[0008] According to the first aspect of the embodiments of the present disclosure, the second prediction model is a linear model or a nonlinear model having a predetermined weight for the second predicted attention value, wherein adjusting the numerical value of M and the parameters of the second prediction model includes: calculating an evaluation index based on the third predicted attention value and the second true attention value of the M sample data; and adjusting the number of M and the predetermined weight of the second prediction model according to the evaluation index.

[0009] According to a first aspect of an embodiment of the present disclosure, the evaluation index is a root mean square error (RMSE) or an R-square value.

[0010] According to a first aspect of an embodiment of the present disclosure, the attribute data includes at least one of the following information about the work: whether there is a title, whether there is subtitles, whether there is a tag, and whether there is background music.

[0011] According to a second aspect of an embodiment of the present disclosure, there is provided a training device for an attention prediction model, comprising: a data acquisition unit, configured to acquire previous attention data of a plurality of works and attribute data affecting the attention of the works; a first prediction unit, configured to input the previous attention data and attribute data of the plurality of works as features into a first prediction model to obtain a first predicted attention value of each work, and adjust the parameters of the first prediction model based on the first predicted attention values ​​and the true attention values ​​of the plurality of works; a sample generation unit, configured to form a sample pair with the true attention value and the first predicted attention value of each work in the plurality of works, and to input the sample pairs in a predetermined order; The multiple sample pairs corresponding to the multiple works are divided into M groups, and the mean of the real attention values ​​and the mean of the first predicted attention values ​​of all sample pairs in each of the M groups are respectively determined as the second real attention value and the second predicted attention value of the new M sample data; the second prediction unit is configured to input the second predicted attention value of each sample data of the M sample data as a feature into the second prediction model to obtain the third predicted attention value of each sample data of the M sample data, and adjust the value of M and the parameters of the second attention model based on the third predicted attention value and the second real attention value of each sample data of the M sample data.

[0012] According to the second aspect of the embodiment of the present disclosure, sample generation is configured as: arranging the multiple samples in ascending or descending order according to the real attention values ​​or the first predicted attention values ​​of the multiple sample pairs; and evenly dividing the arranged multiple samples into the M groups.

[0013] According to the second aspect of the embodiment of the present disclosure, the first prediction model is a linear model or a nonlinear model having predetermined weights for previous attention data and attribute data, wherein the first prediction unit is configured to: calculate a predetermined evaluation index based on the first predicted attention values ​​and the actual attention values ​​of the multiple works; and adjust the predetermined weight of the first prediction model according to the predetermined evaluation index.

[0014] According to the second aspect of an embodiment of the present disclosure, the second prediction model is a linear model or a nonlinear model having a predetermined weight for the second predicted attention value, wherein the second prediction unit is configured to: calculate a predetermined evaluation index based on the third predicted attention value and the second true attention value of the M sample data; and adjust the number of M and the predetermined weight of the second prediction model according to the predetermined evaluation index.

[0015] According to a second aspect of an embodiment of the present disclosure, the evaluation index is a root mean square error (RMSE) or an R-square value.

[0016] According to a second aspect of an embodiment of the present disclosure, the attribute data includes at least one of the following information about the work: whether there is a title, whether there is subtitles, whether there is a tag, and whether there is background music.

[0017] According to a third aspect of an embodiment of the present disclosure, there is provided a method for predicting attention, comprising: obtaining previous attention data of a work and attribute data affecting the attention of the work; inputting the previous attention data and attribute data of the work as features into a first prediction model to obtain a first attention prediction value for the work; inputting the first attention prediction value into a second prediction model to obtain a second attention prediction value for the work as an attention prediction for the work, wherein the first prediction model and the second prediction model are trained according to the training method described above.

[0018] According to a third aspect of an embodiment of the present disclosure, in response to not obtaining previous attention data and / or attribute data of the work, a default identification value for indicating execution of other operations is returned.

[0019] According to a third aspect of an embodiment of the present disclosure, the attribute data includes at least one of the following information about the work: whether there is a title, whether there is subtitles, whether there is a tag, and whether there is background music.

[0020] According to a fourth aspect of an embodiment of the present disclosure, there is provided an attention prediction device, comprising: a data acquisition unit, configured to acquire previous attention data of a work and attribute data affecting the attention of the work; a first prediction unit, configured to input the previous attention data and attribute data of the work as features into a first prediction model to obtain a first attention prediction value for the work; a second prediction unit, configured to input the first attention prediction value into a second prediction model to obtain a second attention prediction value for the work as an attention prediction for the work, wherein the first prediction model and the second prediction model are trained according to the training method as described above.

[0021] According to a fourth aspect of an embodiment of the present disclosure, the first prediction unit is configured to: in response to the data acquisition unit not acquiring previous attention data and / or attribute data of the work, output a default identification value for indicating execution of other operations.

[0022] According to a fourth aspect of an embodiment of the present disclosure, the attribute data includes at least one of the following information about the work: whether there is a title, whether there is subtitles, whether there are tags, and whether there is background music.

[0023] According to the fifth aspect of an embodiment of the present disclosure, an electronic device is provided, comprising: a processor; a memory for storing instructions executable by the processor; wherein the processor is configured to execute the instructions to implement the training method of the attention prediction model and the attention prediction method according to the present disclosure.

[0024] According to the sixth aspect of an embodiment of the present disclosure, a computer-readable storage medium is provided. When the instructions in the computer-readable storage medium are executed by a processor of an electronic device, the electronic device is enabled to execute the training method of the attention prediction model and the attention prediction method described in the present disclosure.

[0025] According to a seventh aspect of an embodiment of the present disclosure, a computer program product is provided, comprising a computer program, which, when executed by a processor, implements the training method of the attention prediction model and the attention prediction method according to the present disclosure.

[0026] The technical solution provided by the embodiments of the present disclosure brings at least the following beneficial effects:

[0027] According to the historical attention data of the work, the prediction data is corrected through multiple prediction models to improve the prediction accuracy, so as to obtain more accurate attention prediction values.

[0028] It is to be understood that the foregoing general description and the following detailed description are exemplary and explanatory only and are not restrictive of the present disclosure. BRIEF DESCRIPTION OF THE DRAWINGS

[0029] The drawings herein are incorporated into and constitute a part of the specification, illustrate embodiments consistent with the present disclosure, and together with the description are used to explain the principles of the present disclosure, and do not constitute improper limitations on the present disclosure.

[0030] Figure 1 is an overall schematic diagram showing a method for predicting attention according to an exemplary embodiment of the present disclosure;

[0031] Figure 2 is a flowchart illustrating a method for training an attention prediction model according to an exemplary embodiment of the present disclosure;

[0032] Figure 3 is a block diagram showing a training device for an attention prediction model according to an exemplary embodiment of the present disclosure;

[0033] Figure 4 is a flowchart illustrating a method for predicting attention according to an exemplary embodiment of the present disclosure;

[0034] Figure 5 is a block diagram showing an attention prediction device according to an exemplary embodiment of the present disclosure;

[0035] Figure 6 is a schematic diagram illustrating an electronic device according to an exemplary embodiment of the present disclosure.

[0036] Figure 7 is a schematic diagram illustrating another electronic device according to an exemplary embodiment of the present disclosure. DETAILED DESCRIPTION

[0037] In order to enable ordinary persons in the art to better understand the technical solutions of the present disclosure, the technical solutions in the embodiments of the present disclosure will be clearly and completely described below in conjunction with the accompanying drawings.

[0038] It should be noted that the terms "first", "second", etc. in the specification and claims of the present disclosure and the above-mentioned drawings are used to distinguish similar objects, and are not necessarily used to describe a specific order or sequence. It should be understood that the data used in this way can be interchanged where appropriate, so that the embodiments of the present disclosure described herein can be implemented in an order other than those illustrated or described herein. The implementation methods described in the following examples do not represent all implementation methods consistent with the present disclosure. Instead, they are merely examples of devices and methods consistent with some aspects of the present disclosure as detailed in the attached claims.

[0039] It should be noted that the phrase "at least one of the items" in the present disclosure includes three types of parallel situations: "any one of the items", "a combination of any number of the items", and "all of the items". For example, "including at least one of A and B" includes the following three parallel situations: (1) including A; (2) including B; (3) including A and B. Another example is "executing at least one of step 1 and step 2" which means the following three parallel situations: (1) executing step 1; (2) executing step 2; (3) executing step 1 and step 2.

[0040] Because there is usually a serious imbalance in the number of views and followers of a work or a live broadcast room, there will be uneven data distribution and serious distrust when calculating the attention. For example, within 1 day, work A is viewed twice, resulting in one attention behavior, and work B is viewed 200,000 times, resulting in 100,000 attention behaviors. If the calculation method of attention count / view count is directly used, then the attention of work A and work B will both be 0.5. However, common sense tells us that the attention of work A is obviously distrustful. In addition, there is a head effect in the number of attentions and views of works on the platform. The number of attentions and views of the head works are often much higher than those of the tail works. These will increase the difficulty of predicting attention.

[0041] In the related art, for attention prediction, the conventional method is generally to first sample sparse samples, perform Bayesian smoothing on the attention data, and then establish a regression or classification task to predict the attention data. The embodiment of the present disclosure adds a bucketing multi-model integrated prediction calibration step after the conventional regression modeling task on the attention data, and the predicted value and the true value are combined into sample pairs, and then bucketed after sorting in descending order according to the predicted value, and the mean of the sample predicted value and the true value in the bucket is calculated to form a new training sample set, and then a linear or nonlinear model is used for regression prediction.

[0042] Figure 1 is an overall schematic diagram showing an attention degree prediction method according to an exemplary embodiment of the present disclosure.

[0043] like Figure 1 As shown, the attention prediction method according to an exemplary embodiment of the present disclosure can be divided into two stages, namely, an offline model training stage and an online model use stage.

[0044] In the training phase, a statistical period (e.g., N days) can be pre-set, and the previous attention data of the work (e.g., the attention data of the work in the previous N-1 days) and the attribute characteristics corresponding to the work can be counted. For example, the previous attention data and attribute data can be obtained from the work history attention data table and the work attribute table stored in the database.

[0045] Then, the attention data and attribute data of the works in the previous N-1 days are used as sample features, and the attention data of the works on the Nth day are used as sample labels. The training samples are input into model M1, and model M1 is trained. The work attention data predicted by M1 and the real attention data form sample pairs and perform calibration. Specifically, the sample pairs can be sorted and divided into X buckets according to the predicted values ​​output by M1. The number of <predicted value, real value> sample pairs contained in each bucket is Y. The mean of the predicted value and the real value of the Y sample pairs in each bucket is calculated to form new X mean sample data, and then the model M2 is used to train the new sample pair set.

[0046] During the online use phase, the trained models M1 and M2 are loaded online. Whenever a request for a work's attention prediction is received, the work's historical attention data and attributes can be obtained from the database as features based on the work's ID, and the model M1 is used to perform predictions on the work's features (e.g., the attention and attribute data of the previous N-1 days) to obtain predicted values, and the predicted values ​​are input into the model M2 to obtain new predicted values, and the new predicted values ​​are returned as the final predicted values ​​of the work's attention. Here, if the historical attention data is not obtained, the default prediction result can be directly returned (for example, it can be a result predicted by other prediction methods or a prompt that a prediction cannot be made).

[0047] The above training phase and online phase may be implemented in a predetermined form (eg, software application) such as in a server and a client.

[0048] The following will refer to Figure 2 To illustrate the training method of the attention prediction model according to an exemplary embodiment of the present disclosure.

[0049] like Figure 2 As shown, first, in step S210, the previous attention data of multiple works and the attribute data affecting the attention of the works are obtained. Here, the previous attention data can be the cumulative attention data of the works before the current period. Assuming that the total statistical period of the work is N days and the statistical interval is 1 day, the attention data of the previous N-1 days can be used as the previous attention data of the work to predict the attention of the Nth day. For example, the number of clicks on the work and the number of times the author of the work has been followed in the previous N-1 days can be counted as the previous attention data. According to an exemplary embodiment of the present disclosure, the time of the statistical period usually needs to be greater than 15 days.

[0050] Attribute data can be any data expressed in numerical form that is related to the attributes that affect the attention of the work. For example, attributes that can be expressed in binary form (1 / 0) include: whether there is a title, whether there are letters, whether there are labels, whether there is background music. In the above attribute data, 1 can be used to represent "yes" and "0" can be used to represent no. It should be understood that in addition to binary attributes, other possible multi-valued attributes can also be used, such as evaluation scores of 1 to 5, etc., which are not limited here.

[0051] Next, in step S220, the previous attention data and attribute data of multiple works are input as features into the first prediction model to obtain a first predicted attention value for each work, and the parameters of the first prediction model are adjusted based on the first predicted attention values ​​and actual attention values ​​of the multiple works.

[0052] According to an exemplary embodiment of the present disclosure, the previous attention data and attribute data may be processed to improve the usability of the data. The data processing may include, for example, at least one of the following operations:

[0053] 1) Outlier filtering: including removing null values ​​and meaningless values, to prevent the model from learning abnormal weights based on abnormal data;

[0054] 2) Normalization: Unify the dimensions of different feature values ​​to avoid the large impact of features of different dimensions on model learning;

[0055] 3) Discretization: Discretize some features (e.g., work release time) in segments to enhance the model’s memory.

[0056] It should be understood that data processing is not limited to the above examples, and those skilled in the art may add other data processing processes as needed. After processing the data, the processed data is input as features into the first prediction model to obtain the first predicted attention value of each work, and then the parameters of the first prediction model can be adjusted based on the first predicted attention value and the actual attention value of the work.

[0057] According to an exemplary embodiment of the present disclosure, the first prediction model may be a linear model or a nonlinear model with a predetermined weight. According to an exemplary embodiment of the present disclosure, the linear model may be, for example, a linear regression model, and the nonlinear model may include, for example, a logarithmic regression model, an exponential regression model, and the like.

[0058] For example, assuming that the feature values ​​of the processed data are x1, x2, and x3, and the output predicted attention value is y, then the first prediction model can be expressed as: y = w1*x1+w2*x2+w3*x3, where w1, w2, and w3 are the weights of the linear model. Alternatively, the first prediction model can be expressed as y = w4*x1+w5*x2 2 +w6*x3 3 , where w4, w5, and w6 are weights of the nonlinear model. It should be understood that the first prediction model can also adopt other prediction models, such as prediction models based on artificial intelligence neural networks, and the present disclosure is not limited to this. After the predicted attention value of the work is obtained through the first prediction model, the evaluation index can be calculated based on the first predicted attention value and the real attention value based on multiple works, and the weight of the first prediction model can be adjusted according to the evaluation index. According to an exemplary embodiment of the present disclosure, evaluation indicators such as the root mean square error (RMSE) or the R-square value can be used as evaluation indicators. That is, the weights of the above linear model or nonlinear model can be adjusted according to the RMSE value or the R-square value between the first predicted attention value and the real attention value of the work, so as to optimize the prediction effect of the first prediction model. The trained first prediction model can be saved and used in subsequent online prediction processes.

[0059] Next, in step S230, the real attention value and the first predicted attention value of each of the multiple works are combined into sample pairs, and the multiple sample pairs corresponding to the multiple works are divided into M groups in a predetermined order, and the mean of the real attention values ​​and the mean of the first predicted attention values ​​of all sample pairs in each of the M groups are determined respectively as the second real attention value and the second predicted attention value of the new M sample data.

[0060] According to an exemplary embodiment of the present disclosure, in step S230, the multiple samples may be arranged in ascending or descending order according to the real attention values ​​or the first predicted attention values ​​of the multiple sample pairs, and the arranged multiple samples may be evenly divided into the M groups.

[0061] For example, the first attention prediction value y_pred and the real attention value y_real of the work output by the first prediction model can be used to form a sample pair. According to the size of the first attention prediction value y_pred or the real attention value y_real, the sample pairs of all works are arranged in ascending or descending order, and then the sample pairs are divided into M buckets, each of which contains Y sample pairs.<y_pred,y_real> For example, assuming the number of works is 10,000, all works are divided into 100 buckets, and each bucket contains 100 sample pairs.

[0062] After bucketing, all sample pairs in the bucket can be traversed, and the average of the first predicted attention value and the real attention value of Y sample pairs in each bucket is calculated as the second predicted attention value and the second real attention value mentioned above, which are expressed as<y_pred_mean,y_real_mean> , thus obtaining a total of M new samples.

[0063] Next, in step S240, the second predicted attention value of each sample data of the M sample data is input as a feature into the second prediction model to obtain the third predicted attention value of each sample data of the M sample data, and the value of M and the parameters of the second attention model are adjusted based on the third predicted attention value and the second true attention value of each sample data of the M sample data.

[0064] According to an exemplary embodiment of the present disclosure, the second prediction model may be a linear model or a nonlinear model having a predetermined weight for the second predicted attention value, wherein adjusting the value of M and the parameters of the second prediction model includes: calculating an evaluation index based on the third predicted attention value and the second true attention value of the M sample data, and adjusting the number of M and the predetermined weight of the second prediction model according to the evaluation index. That is, the second prediction model may be a model of the same type as the first prediction model, or a model of a different type. The evaluation index may be, for example, the root mean square error (RMSE) or the R-square value.

[0065] By adjusting the value of M (i.e., the number of buckets) and the parameters of the second prediction model (e.g., the weight of the second prediction model), the prediction effect of the second prediction model can be optimized. That is, after adjusting the number of M, step S230 can be performed again to regenerate M sample data and S240 can be performed again until the evaluation index of the second prediction model is optimized.

[0066] Through the above prediction model training method, based on the attention data after the first prediction, a more accurate prediction model can be obtained by correcting the deviation of the previously obtained prediction data through bucketing, averaging and a second prediction, thereby obtaining a more accurate attention prediction in the subsequent online use process.

[0067] Figure 3 FIG. 4 is a diagram showing a training device for an attention prediction model according to an exemplary embodiment of the present disclosure. Figure 3 As shown, the training device 300 of the attention prediction model according to an exemplary embodiment of the present disclosure may include a data acquisition unit 310, a first prediction unit 320, a sample generation unit 330 and a second prediction unit 340.

[0068] The data acquisition unit 310 is configured to acquire the previous attention data of a plurality of works and the attribute data affecting the attention of the works. According to an exemplary embodiment of the present disclosure, the previous attention data may be the cumulative attention data of the works before the current period, and the attribute data may be data related to the attributes affecting the attention of the works expressed in numerical form. According to an exemplary embodiment of the present disclosure, the attribute data includes at least one of the following information about the works: whether there is a title, whether there is a subtitle, whether there is a tag, and whether there is background music.

[0069] The first prediction unit 320 is configured to input the previous attention data and attribute data of the multiple works as features into the first prediction model to obtain a first predicted attention value for each work, and adjust the parameters of the first prediction model based on the first predicted attention values ​​and the actual attention values ​​of the multiple works. According to an exemplary embodiment of the present disclosure, the first prediction unit 320 may process the previous attention data and attribute data to improve the availability of the data. According to an exemplary embodiment of the present disclosure, the first prediction model may be a linear model or a nonlinear model with predetermined weights. According to an exemplary embodiment of the present disclosure, the linear model may be, for example, a linear regression model, and the nonlinear model may include, for example, a logarithmic regression model, an exponential regression model, and the like.

[0070] The sample generation unit 330 is configured to form sample pairs with the real attention value and the first predicted attention value of each of the multiple works, divide the multiple sample pairs corresponding to the multiple works into M groups in a predetermined order, and respectively determine the mean of the real attention values ​​and the mean of the first predicted attention values ​​of all sample pairs in each of the M groups as the second real attention value and the second predicted attention value of the new M sample data.

[0071] The second prediction unit 340 is configured to input the second predicted attention value of each sample data of the M sample data as a feature into the second prediction model to obtain the third predicted attention value of each sample data of the M sample data, and adjust the value of M and the parameters of the second attention model based on the third predicted attention value and the second true attention value of each sample data of the M sample data.

[0072] According to an exemplary embodiment of the present disclosure, the second prediction model is a linear model or a nonlinear model with a predetermined weight for the second predicted attention value, wherein the second prediction unit 340 is configured to: calculate a predetermined evaluation index based on the third predicted attention value and the second real attention value of the M sample data, and adjust the number of M and the predetermined weight of the second prediction model according to the predetermined evaluation index. The evaluation index may be a root mean square error RMSE or an R-square value.

[0073] After adjusting the number of M, the sample generation unit 330 can re-group the sample pairs of all works and calculate the average of the predicted attention values ​​and the actual attention values ​​of the sample pairs in the group, thereby generating M new sample data. The second prediction unit 340 can re-predict based on the new sample data and continue to adjust the parameters of the second prediction model and the number of M until the evaluation index of the second prediction model reaches the optimal level.

[0074] The following will refer to Figure 4 A method for predicting the degree of attention of a work according to an exemplary embodiment of the present disclosure is described below.

[0075] As mentioned above, according to Figure 2 and Figure 3 After the training method and device obtain the first prediction model and the second prediction model through training, the first prediction model and the second prediction model can be used to predict the attention of the work.

[0076] like Figure 4 As shown, in step S410, the previous attention data of the work and the attribute data affecting the attention of the work are obtained. Figure 2 Examples of previous attention data and attribute data of works are described in detail and will not be repeated here.

[0077] Next, in step S420, the previous attention data and attribute data of the work are input as features into a first prediction model to obtain a first attention prediction value for the work.

[0078] Then, in step S430, the first attention prediction value is input into the second prediction model to obtain a second attention prediction value for the work as an attention prediction for the work.

[0079] Here, the first prediction model and the second prediction model are based on Figure 2 The training method is used to train the obtained data.

[0080] According to an exemplary embodiment of the present disclosure, in response to not obtaining the previous attention data and / or attribute data of the work in step S410, a default identification value is returned to indicate that other operations are performed without executing subsequent steps S420 and S430. For example, it may be returned to indicate that a prediction algorithm of other related technologies is used to predict the attention of the work, or it may be prompted that the attention of the work cannot be predicted.

[0081] Through the above prediction method, the prediction accuracy can be improved by correcting the prediction data based on the historical attention data of the work through multiple prediction models, so as to obtain a more accurate attention prediction value.

[0082] Figure 5 is a block diagram illustrating an attention prediction apparatus according to an exemplary embodiment of the present disclosure.

[0083] like Figure 5 As shown, the attention prediction device 500 according to an exemplary embodiment of the present disclosure may include a data acquisition unit 510 , a first prediction unit 520 , and a second prediction unit 530 .

[0084] The data acquisition unit 510 is configured to acquire previous attention data of a work and attribute data affecting the attention of the work.

[0085] The first prediction unit 520 is configured to input the previous attention data and attribute data of the work as features into a first prediction model to obtain a first attention prediction value for the work.

[0086] The second prediction unit 530 is configured to input the first attention prediction value into the second prediction model to obtain a second attention prediction value about the work as an attention prediction for the work.

[0087] Here, the first prediction model and the second prediction model are based on Figure 2 In addition, reference has been made to Figure 2 Examples of previous attention data and attribute data of works will not be repeated here.

[0088] In addition, according to an exemplary embodiment of the present disclosure, the first prediction unit 520 is further configured to: in response to the data acquisition unit not acquiring previous attention data and / or attribute data of the work, output a default identification value for indicating execution of other operations.

[0089] Figure 6The electronic device 600 is a block diagram of an electronic device for predicting attention. The electronic device 600 may be, for example, a smart phone, a tablet computer, an MP4 (Moving Picture Experts Group Audio Layer IV) player, a laptop computer, or a desktop computer. The electronic device 600 may also be called a user device, a portable terminal, a laptop terminal, a desktop terminal, or other names.

[0090] Typically, the electronic device 600 includes a processor 601 and a memory 602 .

[0091] The processor 601 may include one or more processing cores, such as a 4-core processor, an 8-core processor, etc. The processor 601 may be implemented in at least one hardware form of DSP (Digital Signal Processing), FPGA (Field Programmable Gate Array), and PLA (Programmable Logic Array). The processor 601 may also include a main processor and a coprocessor. The main processor is a processor for processing data in an awake state, also known as a CPU (Central Processing Unit); the coprocessor is a low-power processor for processing data in a standby state. In some embodiments, the processor 601 may be integrated with a GPU (Graphics Processing Unit), which is responsible for rendering and drawing the content to be displayed on the display screen. In some embodiments, the processor 601 may also include an AI (Artificial Intelligence) processor, which is used to process computing operations related to machine learning.

[0092] The memory 602 may include one or more computer-readable storage media, which may be non-transitory. The memory 602 may also include a high-speed random access memory and a non-volatile memory, such as one or more disk storage devices and flash memory storage devices. In some embodiments, the non-transitory computer-readable storage medium in the memory 602 is used to store at least one instruction, which is used to be executed by the processor 601 to implement the present disclosure. Figure 2 and Figure 4 The method embodiments shown provide an attention prediction model training method and / or an attention prediction method.

[0093] In some embodiments, the electronic device 600 may further optionally include: a peripheral device interface 603 and at least one peripheral device. The processor 601, the memory 602 and the peripheral device interface 603 may be connected via a bus or a signal line. Each peripheral device may be connected to the peripheral device interface 603 via a bus, a signal line or a circuit board. Specifically, the peripheral device includes: at least one of a radio frequency circuit 604, a touch display screen 605, a camera 606, an audio circuit 607, a positioning component 608 and a power supply 609.

[0094] The peripheral device interface 603 may be used to connect at least one peripheral device related to I / O (Input / Output) to the processor 601 and the memory 602. In some embodiments, the processor 601, the memory 602, and the peripheral device interface 603 are integrated on the same chip or circuit board; in some other embodiments, any one or two of the processor 601, the memory 602, and the peripheral device interface 603 may be implemented on a separate chip or circuit board, which is not limited in this embodiment.

[0095] The radio frequency circuit 604 is used to receive and transmit RF (Radio Frequency) signals, also known as electromagnetic signals. The radio frequency circuit 604 communicates with the communication network and other communication devices through electromagnetic signals. The radio frequency circuit 604 converts the electrical signal into an electromagnetic signal for transmission, or converts the received electromagnetic signal into an electrical signal. Optionally, the radio frequency circuit 604 includes: an antenna system, an RF transceiver, one or more amplifiers, a tuner, an oscillator, a digital signal processor, a codec chipset, a user identity module card, and the like. The radio frequency circuit 604 can communicate with other terminals through at least one wireless communication protocol. The wireless communication protocol includes, but is not limited to: a metropolitan area network, various generations of mobile communication networks (2G, 3G, 4G and 5G), a wireless local area network and / or a WiFi (Wireless Fidelity) network. In some embodiments, the radio frequency circuit 604 may also include circuits related to NFC (Near Field Communication), which is not limited in the present disclosure.

[0096] The display screen 605 is used to display the UI (User Interface). The UI may include graphics, text, icons, videos and any combination thereof. When the display screen 605 is a touch display screen, the display screen 605 also has the ability to collect touch signals on the surface or above the surface of the display screen 605. The touch signal can be input to the processor 601 as a control signal for processing. At this time, the display screen 605 can also be used to provide virtual buttons and / or virtual keyboards, also known as soft buttons and / or soft keyboards. In some embodiments, the display screen 605 can be one, set on the front panel of the electronic device 600; in other embodiments, the display screen 605 can be at least two, respectively set on different surfaces of the terminal 600 or in a folding design; in some further embodiments, the display screen 605 can be a flexible display screen, set on the curved surface or folding surface of the terminal 600. Even, the display screen 605 can also be set to a non-rectangular irregular shape, that is, a special-shaped screen. The display screen 605 can be made of materials such as LCD (Liquid Crystal Display), OLED (Organic Light-Emitting Diode) and the like.

[0097] The camera assembly 606 is used to capture images or videos. Optionally, the camera assembly 606 includes a front camera and a rear camera. Typically, the front camera is arranged on the front panel of the terminal, and the rear camera is arranged on the back of the terminal. In some embodiments, there are at least two rear cameras, which are any one of a main camera, a depth of field camera, a wide-angle camera, and a telephoto camera, so as to realize the fusion of the main camera and the depth of field camera to realize the background blur function, the fusion of the main camera and the wide-angle camera to realize the panoramic shooting and VR (Virtual Reality) shooting function or other fusion shooting functions. In some embodiments, the camera assembly 606 may also include a flash. The flash can be a monochrome temperature flash or a dual-color temperature flash. A dual-color temperature flash refers to a combination of a warm light flash and a cold light flash, which can be used for light compensation at different color temperatures.

[0098] The audio circuit 607 may include a microphone and a speaker. The microphone is used to collect sound waves from the user and the environment, and convert the sound waves into electrical signals and input them into the processor 601 for processing, or input them into the radio frequency circuit 604 to achieve voice communication. For the purpose of stereo acquisition or noise reduction, there may be multiple microphones, which are respectively arranged at different parts of the terminal 600. The microphone may also be an array microphone or an omnidirectional acquisition microphone. The speaker is used to convert the electrical signal from the processor 601 or the radio frequency circuit 604 into sound waves. The speaker may be a traditional film speaker or a piezoelectric ceramic speaker. When the speaker is a piezoelectric ceramic speaker, it can not only convert the electrical signal into sound waves audible to humans, but also convert the electrical signal into sound waves inaudible to humans for purposes such as ranging. In some embodiments, the audio circuit 607 may also include a headphone jack.

[0099] Positioning component 608 is used to locate the current geographic location of electronic device 600 to implement navigation or LBS (Location Based Service). Positioning component 608 can be a positioning component based on the US GPS (Global Positioning System), China's Beidou system, Russia's Grenas system or the European Union's Galileo system.

[0100] The power supply 609 is used to power various components in the electronic device 600. The power supply 609 can be an alternating current, a direct current, a disposable battery, or a rechargeable battery. When the power supply 609 includes a rechargeable battery, the rechargeable battery can support wired charging or wireless charging. The rechargeable battery can also be used to support fast charging technology.

[0101] In some embodiments, the electronic device 600 further includes one or more sensors 610 , including but not limited to: an acceleration sensor 611 , a gyroscope sensor 612 , a pressure sensor 613 , a fingerprint sensor 614 , an optical sensor 615 , and a proximity sensor 616 .

[0102] The acceleration sensor 611 can detect the magnitude of acceleration on the three coordinate axes of the coordinate system established by the terminal 600. For example, the acceleration sensor 611 can be used to detect the components of gravity acceleration on the three coordinate axes. The processor 601 can control the touch display screen 605 to display the user interface in a horizontal view or a vertical view according to the gravity acceleration signal collected by the acceleration sensor 611. The acceleration sensor 611 can also be used for collecting game or user motion data.

[0103] The gyro sensor 612 can detect the body direction and rotation angle of the terminal 600, and the gyro sensor 612 can cooperate with the acceleration sensor 611 to collect the user's 3D actions on the terminal 600. The processor 601 can implement the following functions based on the data collected by the gyro sensor 612: motion sensing (such as changing the UI according to the user's tilt operation), image stabilization during shooting, game control, and inertial navigation.

[0104] The pressure sensor 613 can be set on the side frame of the terminal 600 and / or the lower layer of the touch display screen 605. When the pressure sensor 613 is set on the side frame of the terminal 600, it can detect the user's holding signal of the terminal 600, and the processor 601 performs left and right hand recognition or shortcut operation according to the holding signal collected by the pressure sensor 613. When the pressure sensor 613 is set on the lower layer of the touch display screen 605, the processor 601 controls the operability controls on the UI according to the user's pressure operation on the touch display screen 605. The operability controls include at least one of a button control, a scroll bar control, an icon control, and a menu control.

[0105] The fingerprint sensor 614 is used to collect the user's fingerprint, and the processor 601 identifies the user's identity based on the fingerprint collected by the fingerprint sensor 614, or the fingerprint sensor 614 identifies the user's identity based on the collected fingerprint. When the user's identity is identified as a trusted identity, the processor 601 authorizes the user to perform relevant sensitive operations, which include unlocking the screen, viewing encrypted information, downloading software, paying, and changing settings. The fingerprint sensor 614 can be set on the front, back, or side of the electronic device 600. When a physical button or a manufacturer logo is set on the electronic device 600, the fingerprint sensor 614 can be integrated with the physical button or the manufacturer logo.

[0106] The optical sensor 615 is used to collect the ambient light intensity. In one embodiment, the processor 601 can control the display brightness of the touch display screen 605 according to the ambient light intensity collected by the optical sensor 615. Specifically, when the ambient light intensity is high, the display brightness of the touch display screen 605 is increased; when the ambient light intensity is low, the display brightness of the touch display screen 605 is reduced. In another embodiment, the processor 601 can also dynamically adjust the shooting parameters of the camera assembly 606 according to the ambient light intensity collected by the optical sensor 615.

[0107] The proximity sensor 616, also called a distance sensor, is usually arranged on the front panel of the electronic device 600. The proximity sensor 616 is used to collect the distance between the user and the front of the electronic device 600. In one embodiment, when the proximity sensor 616 detects that the distance between the user and the front of the terminal 600 is gradually decreasing, the processor 601 controls the touch display screen 605 to switch from the screen-on state to the screen-off state; when the proximity sensor 616 detects that the distance between the user and the front of the electronic device 600 is gradually increasing, the processor 601 controls the touch display screen 605 to switch from the screen-off state to the screen-on state.

[0108] Those skilled in the art will understand that Figure 6 The structure shown in the figure does not constitute a limitation on the electronic device 600, and may include more or less components than those shown in the figure, or combine some components, or adopt a different component arrangement.

[0109] Figure 7 FIG. 7 is a block diagram of another electronic device 700. For example, the electronic device 700 may be provided as a server. Figure 7 , the electronic device 700 includes one or more processing processors 710 and a memory 720. The memory 720 may include one or more programs for executing the above attention prediction model training method and / or attention prediction method. The electronic device 700 may also include a power supply component 730 configured to perform power management of the electronic device 700, a wired or wireless network interface 740 configured to connect the electronic device 700 to the network, and an input / output (I / O) interface 750. The electronic device 700 can operate based on an operating system stored in the memory 720, such as Windows Server TM, Mac OS X TM, Unix TM, Linux TM, FreeBSD TM or the like.

[0110] According to an embodiment of the present disclosure, a computer-readable storage medium storing instructions may also be provided, wherein when the instructions are executed by at least one processor, the at least one processor is prompted to execute the attention prediction model training method and / or the attention prediction method according to the present disclosure. Examples of computer-readable storage media here include: read-only memory (ROM), random access programmable read-only memory (PROM), electrically erasable programmable read-only memory (EEPROM), random access memory (RAM), dynamic random access memory (DRAM), static random access memory (SRAM), flash memory, non-volatile memory, CD-ROM, CD-R, CD+R, CD-RW, CD+RW, DVD-ROM, DVD-R, DVD+R, DVD-RW, DVD+RW, DVD-RAM, BD-ROM, BD-R, BD-R LTH, BD-RE, Blu-ray or optical disk storage, hard disk drive (HDD), solid state drive (SSD), card storage (such as, multimedia card, secure digital (SD) card or extreme digital (XD) card), magnetic tape, floppy disk, magneto-optical data storage device, optical data storage device, hard disk, solid state disk and any other device, any other device is configured to store computer programs and any associated data, data files and data structures in a non-transitory manner and provide the computer programs and any associated data, data files and data structures to a processor or computer so that the processor or computer can execute the computer program. The computer program in the above-mentioned computer-readable storage medium can be run in an environment deployed in a computer device such as a client, a host, an agent device, a server, etc. In addition, in one example, the computer program and any associated data, data files and data structures are distributed on a networked computer system, so that the computer program and any associated data, data files and data structures are stored, accessed and executed in a distributed manner by one or more processors or computers.

[0111] According to an embodiment of the present disclosure, a computer program product may also be provided, and the instructions in the computer program product may be executed by a processor of a computer device to complete the above-mentioned attention prediction model training method and / or attention prediction method.

[0112] According to the attention prediction model training method and attention prediction method disclosed in the present invention, it is possible to improve the prediction accuracy by correcting the prediction data based on the historical attention data of the work, thereby obtaining a more accurate attention prediction.

[0113] Those skilled in the art will readily appreciate other embodiments of the present disclosure after considering the specification and practicing the invention disclosed herein. This application is intended to cover any variations, uses, or adaptations of the present disclosure that follow the general principles of the present disclosure and include common knowledge or customary techniques in the art that are not disclosed in the present disclosure. The specification and examples are intended to be exemplary only, and the true scope and spirit of the present disclosure are indicated by the following claims.

[0114] It should be understood that the present disclosure is not limited to the exact structures that have been described above and shown in the drawings, and that various modifications and changes may be made without departing from the scope thereof. The scope of the present disclosure is limited only by the appended claims.

Claims

1. A training method for an attention prediction model, characterized in that: include: Obtaining previous attention data of multiple works and attribute data affecting the attention of the works; Inputting the previous attention data and attribute data of the plurality of works as features into a first prediction model to obtain a first predicted attention value of each work, and adjusting parameters of the first prediction model based on the first predicted attention values ​​and the actual attention values ​​of the plurality of works; The real attention value and the first predicted attention value of each work in the multiple works form a sample pair, and the multiple sample pairs corresponding to the multiple works are divided into M groups in a predetermined order, and the mean of the real attention values ​​and the mean of the first predicted attention values ​​of all sample pairs in each of the M groups are determined respectively as the second real attention value and the second predicted attention value of the new M sample data; The second predicted attention value of each sample data of the M sample data is input as a feature into the second prediction model to obtain the third predicted attention value of each sample data of the M sample data, and the value of M and the parameters of the second attention model are adjusted based on the third predicted attention value and the second true attention value of each sample data of the M sample data.

2. The training method according to claim 1, characterized in that: Dividing the plurality of sample pairs into M groups in a predetermined order comprises: Arrange the multiple samples in ascending or descending order according to the real attention values ​​or the first predicted attention values ​​of the multiple sample pairs; The arranged samples are evenly divided into the M groups.

3. The training method according to claim 1, characterized in that: The first prediction model is a linear model or a nonlinear model having predetermined weights for previous attention data and attribute data. Wherein, adjusting the parameters of the first prediction model includes: Calculating an evaluation index based on the first predicted attention values ​​and the real attention values ​​of the plurality of works; The predetermined weight of the first prediction model is adjusted according to the evaluation index.

4. The training method according to claim 1, characterized in that: The second prediction model is a linear model or a nonlinear model having a predetermined weight for the second prediction attention value, The adjustment of the value of M and the parameters of the second prediction model includes: Calculate the evaluation index based on the third predicted attention value and the second real attention value of the M sample data; The number of M and the predetermined weight of the second prediction model are adjusted according to the evaluation index.

5. The training method according to claim 3 or 4, characterized in that: The evaluation index is the root mean square error RMSE or the R-square value.

6. The training method according to claim 1, characterized in that: The attribute data includes at least one of the following information about the work: Is there a title, is there subtitles, is there a label, is there background music.

7. A training device for an attention prediction model, characterized in that: include: A data acquisition unit, configured to acquire previous attention data of a plurality of works and attribute data affecting the attention of the works; A first prediction unit is configured to input the previous attention data and attribute data of the plurality of works as features into a first prediction model to obtain a first predicted attention value of each work, and adjust parameters of the first prediction model based on the first predicted attention values ​​and actual attention values ​​of the plurality of works; a sample generating unit configured to form a sample pair from the real attention value and the first predicted attention value of each of the multiple works, to divide the multiple sample pairs corresponding to the multiple works into M groups in a predetermined order, and to respectively determine the mean of the real attention values ​​and the mean of the first predicted attention values ​​of all sample pairs in each of the M groups as the second real attention value and the second predicted attention value of the new M sample data; The second prediction unit is configured to input the second predicted attention value of each sample data of the M sample data as a feature into the second prediction model to obtain the third predicted attention value of each sample data of the M sample data, and adjust the value of M and the parameters of the second attention model based on the third predicted attention value and the second true attention value of each sample data of the M sample data.

8. The training device according to claim 7, characterized in that The sample build is configured as: Arrange the multiple samples in ascending or descending order according to the real attention values ​​or the first predicted attention values ​​of the multiple sample pairs; The arranged samples are evenly divided into the M groups.

9. The training device according to claim 7, characterized in that The first prediction model is a linear model or a nonlinear model having predetermined weights for previous attention data and attribute data. Wherein, the first prediction unit is configured as: Calculating a predetermined evaluation index based on the first predicted attention values ​​and the real attention values ​​of the plurality of works; The predetermined weight of the first prediction model is adjusted according to the predetermined evaluation index.

10. The training device according to claim 7, characterized in that The second prediction model is a linear model or a nonlinear model having a predetermined weight for the second prediction attention value, Wherein, the second prediction unit is configured as: Calculate a predetermined evaluation index based on the third predicted attention value and the second real attention value of the M sample data; The number of M and the predetermined weight of the second prediction model are adjusted according to the predetermined evaluation index.

11. The training device according to claim 9 or 10, characterized in that The evaluation index is the root mean square error RMSE or the R-square value.

12. The training device according to claim 7, characterized in that The attribute data includes at least one of the following information about the work: Is there a title, is there subtitles, is there a label, is there background music.

13. A method for predicting attention, characterized in that: include: Obtaining previous attention data of the work and attribute data affecting the attention of the work; Inputting the previous attention data and attribute data of the work as features into a first prediction model to obtain a first attention prediction value for the work; Inputting the first attention prediction value into the second prediction model to obtain a second attention prediction value for the work as an attention prediction for the work, The first prediction model and the second prediction model are trained according to the training method of any one of claims 1-4.

14. The method according to claim 13, characterized in that In response to not obtaining previous attention data and / or attribute data of the work, a default identification value for indicating execution of other operations is returned.

15. The method according to claim 13, characterized in that The attribute data includes at least one of the following information about the work: Is there a title, is there subtitles, is there a label, is there background music.

16. An attention prediction device, characterized in that: include: A data acquisition unit, configured to acquire previous attention data of the work and attribute data affecting the attention of the work; A first prediction unit is configured to input the previous attention data and attribute data of the work as features into a first prediction model to obtain a first attention prediction value for the work; The second prediction unit is configured to input the first attention prediction value into a second prediction model to obtain a second attention prediction value for the work as an attention prediction for the work, The first prediction model and the second prediction model are trained according to the training method of any one of claims 1-4.

17. The device according to claim 16, characterized in that The first prediction unit is configured to: in response to the data acquisition unit not acquiring previous attention data and / or attribute data of the work, output a default identification value for indicating execution of other operations.

18. The device according to claim 16, characterized in that The attribute data includes at least one of the following information about the work: Is there a title, is there subtitles, is there a label, is there background music.

19. An electronic device, characterized in that: include: processor; a memory for storing instructions executable by the processor; Wherein, the processor is configured to execute the instructions to implement the attention prediction model training method as described in any one of claims 1-6 and the attention prediction method as described in any one of claims 13 to 15.

20. A computer-readable storage medium, characterized in that: When the instructions in the computer-readable storage medium are executed by a processor of an electronic device, the electronic device is enabled to execute the attention prediction model training method as described in any one of claims 1-6 and the attention prediction method as described in any one of claims 13 to 15.

21. A computer program product, comprising a computer program, characterized in that When the computer program is executed by a processor, it implements the attention prediction model training method as described in any one of claims 1 to 6 and the attention prediction method as described in any one of claims 13 to 15.

Images