Data processing method and device, equipment and storage medium

Abstract

The invention discloses a data processing method and device, equipment and a storage medium, and relates to the technical field of network televisions and communication. The data processing method comprises the following steps: acquiring equipment identification information of first network television equipment; acquiring first image data through a soft probe of a set top box corresponding to the first network television equipment based on the equipment identification information; inputting the first image data into a target feature extraction model to obtain an image feature corresponding to the first image data, the target feature extraction model being a model obtained by training according to second image data corresponding to a second network television device, the second network television device is a network television device obtained by screening the network television devices according to the network television information corresponding to the network television devices in the network television device set; and performing similarity matching on the image feature corresponding to the first image data and each candidate recommendation video in the candidate recommendation video set to generate a first to-be-recommended video.

Description

Technical Field

[0001] This application belongs to the technical fields of Internet TV and communication technology, and particularly relates to a data processing method, apparatus, device, and storage medium. Background Art

[0002] With the popularization of Internet TV, more and more users abandon traditional digital TVs and choose Internet TVs with stronger content selectivity. For the vast amount of content on Internet TVs, how to perform user preference analysis has become a hot topic at present.

[0003] In the current methods for video content preference analysis, the analysis is mainly based on the rating feedback of users for each video.

[0004] However, the existing methods rely too much on user autonomy. Not all users are willing to give rating feedback, and there are certain limitations in the scope of application. At the same time, due to the large number of users in the existing methods, a large amount of data often needs to be collected, resulting in a large data collection pressure. Summary of the Invention

[0005] Embodiments of this application provide a data processing method, apparatus, device, and storage medium, which can reduce the dependence on user autonomy and relieve the pressure of data collection at the same time.

[0006] On the one hand, an embodiment of this application provides a data processing method, which includes:

[0007] Obtain the device identification information of the first Internet TV device, where the device identification information is used to identify the set-top box corresponding to the first Internet TV device;

[0008] Based on the device identification information, collect first image data through the soft probe of the set-top box corresponding to the first Internet TV device. The first image data includes images corresponding to the historical played videos and the currently played videos of the first Internet TV device;

[0009] Input the first image data into a target feature extraction model to obtain image features corresponding to the first image data. The target feature extraction model is a model trained according to the second image data corresponding to the second Internet TV device. The second Internet TV device is an Internet TV device screened from each Internet TV device in the Internet TV device set according to the Internet TV information corresponding to each Internet TV device;

[0010] Perform similarity matching between the image features corresponding to the first image data and each candidate recommended video in the candidate recommended video set to generate a first to-be-recommended video. The first to-be-recommended video is the first preset number of candidate recommended videos selected from the candidate recommended video set in the order of similarity matching from high to low.

[0011] On the one hand, an embodiment of the present application provides a data processing device, which includes:

[0012] An information acquisition module, configured to acquire device identification information of a first Internet TV device, where the device identification information is used to identify a set-top box corresponding to the first Internet TV device;

[0013] A data collection module, configured to collect first image data through a soft probe of the set-top box corresponding to the first Internet TV device based on the device identification information, where the first image data includes images corresponding to the historical playback video and the current playback video of the first Internet TV device;

[0014] A feature extraction module, configured to input the first image data into a target feature extraction model to obtain image features corresponding to the first image data, where the target feature extraction model is a model trained according to second image data corresponding to a second Internet TV device, and the second Internet TV device is an Internet TV device screened from each Internet TV device in the Internet TV device set according to the Internet TV information corresponding to each Internet TV device;

[0015] A similarity matching module, configured to perform similarity matching between the image features corresponding to the first image data and each candidate recommended video in a candidate recommended video set, and generate a first to-be-recommended video, where the first to-be-recommended video is the first preset number of candidate recommended videos selected from the candidate recommended video set in descending order of similarity matching.

[0016] On the one hand, an embodiment of the present application provides an electronic device, which includes: a memory and a program or instruction stored on the memory and executable on a processor, and when the program or instruction is executed by the processor, it implements the data processing method provided in any one of the above embodiments of the present application.

[0017] On the one hand, an embodiment of the present application provides a readable storage medium, on which a program or instruction is stored, and when the program or instruction is executed by a processor, it implements the data processing method provided in any one of the above embodiments of the present application.

[0018] On the one hand, an embodiment of the present application provides a computer program product, and when the instructions in the computer program product are executed by a processor of an electronic device, the electronic device is caused to execute the data processing method provided in any one of the above embodiments of the present application.

[0019] In the data processing method provided by the embodiments of the present application, the device identification information of the first Internet TV device is obtained. Then, based on the device identification information, the first image data is collected through the soft probe of the set-top box corresponding to the first Internet TV device. Then, based on the image features corresponding to the first image data, the first video to be recommended can be determined. In this way, by automatically collecting the first image data, obtaining the corresponding image features according to the first image data, and directly determining the video recommendation result according to the image features, it is not necessary to wait for the user to give a rating feedback, reducing the dependence on the user's autonomy and overcoming the limitation of the applicable scope caused by the user's unwillingness to give a rating feedback. At the same time, in the embodiments of the present application, each Internet TV device in the set of Internet TV devices is screened according to the Internet TV information corresponding to each Internet TV device to obtain the second Internet TV device, and a target feature extraction model is trained based on the second image data corresponding to the second Internet TV device. In this way, it is possible to preferentially collect the image data of key Internet TV devices, reduce the collection of image data of invalid Internet TV devices, and relieve the pressure of data collection. Description of the Drawings

[0020] In order to more clearly illustrate the technical solutions in the embodiments of the present application, the following will briefly introduce the drawings required to be used in the embodiments of the present application. For those of ordinary skill in the art, without creative efforts, other drawings can also be obtained according to these drawings.

[0021] Figure 1 It is a flowchart of an embodiment of a data processing method provided by the present application;

[0022] Figure 2 It is a flowchart of another embodiment of a data processing method provided by the present application;

[0023] Figure 3 It is a schematic structural diagram of an embodiment of a data processing device provided by the present application;

[0024] Figure 4 It is a flowchart of another embodiment of a data processing device provided by the present application;

[0025] Figure 5 It is a schematic structural diagram of an embodiment of a data processing device provided by the present application. Detailed Embodiments

[0026] The features and exemplary embodiments of various aspects of the present application will be described in detail below. To make the objectives, technical solutions, and advantages of the present application clearer and more understandable, the present application will be further described in detail below in conjunction with the accompanying drawings and specific embodiments. It should be understood that the specific embodiments described herein are only intended to explain the present application, rather than limiting the present application. For those skilled in the art, the present application can be implemented without some of these specific details. The following description of the embodiments is only intended to provide a better understanding of the present application by showing examples of the present application.

[0027] It should be noted that, in this document, relational terms such as "first" and "second" are only used to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply any actual relationship or order between these entities or operations. Moreover, the term "comprising", "including" or any other variant thereof is intended to cover non-exclusive inclusion, so that a process, method, article or device comprising a series of elements not only includes those elements, but also includes other elements not expressly listed, or elements inherent to such process, method, article or device. Without further limitation, an element defined by the statement "comprising..." does not exclude the presence of additional identical elements in the process, method, article or device comprising the said element.

[0028] In the technical solution of the present application, the acquisition, storage, use, processing, etc. of data all comply with the relevant provisions of national laws and regulations.

[0029] In the related art, the current method for analyzing video content preferences mainly analyzes based on the rating feedback of each video by users. However, this method relies too much on user autonomy. Not all users are willing to give rating feedback, and there are certain limitations in the scope of application. At the same time, due to the large number of users, a relatively large amount of data often needs to be collected, resulting in a relatively large data collection pressure.

[0030] The purpose of this application is to provide a data processing method, apparatus, device, and storage medium. In the data processing method provided by the embodiments of this application, the device identification information of the first network television device is obtained, and then based on the device identification information, the first image data is collected through the soft probe of the set-top box corresponding to the first network television device. Then, based on the image features corresponding to the first image data, the first video to be recommended can be determined. In this way, by automatically collecting the first image data, obtaining the corresponding image features according to the first image data, and directly determining the video recommendation result according to the image features, it is not necessary to wait for the user to give a rating feedback, reducing the dependence on the user's autonomy and overcoming the limitation of the applicable range caused by the user's unwillingness to give a rating feedback. At the same time, in the embodiments of this application, each network television device in the network television device set is screened according to the network television information corresponding to each network television device to obtain the second network television device, and the target feature extraction model is trained based on the second image data corresponding to the second network television device. In this way, it is possible to preferentially collect the image data of key network television devices, reduce the collection of image data of invalid network television devices, and relieve the pressure of data collection.

[0031] The following introduces the specific embodiments of the data processing method, apparatus, device, and storage medium provided by the embodiments of this application. First, the data processing method will be introduced below.

[0032] Figure 1 A flowchart of a data processing method is provided. This data processing method is applied to a data analysis system, and this method may include the following steps S101 to S104.

[0033] S101, obtain the device identification information of the first network television device, where the device identification information is used to identify the set-top box corresponding to the first network television device.

[0034] In this embodiment, the first network television device is a television device that needs to perform video preference analysis through the data analysis system.

[0035] The device identification information is used to determine the information of the set-top box corresponding to the first network television device.

[0036] For example, the data analysis system first determines the first network television device that needs to perform video preference analysis, then obtains the device identification information of the first network television device, and determines the set-top box corresponding to the first network television device according to the device identification information.

[0037] S102, based on the device identification information, collect the first image data through the soft probe of the set-top box corresponding to the first network television device, where the first image data includes images corresponding to the historical played videos and the currently played videos of the first network television device.

[0038] In this embodiment, the soft probe of the set-top box is used to collect the image data of the corresponding first network television device during the user's use of the set-top box.

[0039] The first image data includes any image data in the historical playback video of the first network television device and any image data in the current playback video of the first network television device.

[0040] For example, the data analysis system determines the set-top box corresponding to the first network television device according to the device identification information, then configures a video capture task in the web page, and issues the video capture task to the soft probe of the set-top box corresponding to the first network television device through netty.

[0041] Then, the soft probe of the set-top box intercepts the image data in the historical playback video and the current playback video according to the content of the video capture task according to the preset capture configuration information, so as to generate the first image data.

[0042] The first image data will be stored in the local disk in the form of a log through the dp server, logstash will collect the data from the local disk into the kafka sequence, and finally the spark program will read and parse the data and store it in the local database.

[0043] S103, input the first image data into the target feature extraction model to obtain the image features corresponding to the first image data. The target feature extraction model is a model trained according to the second image data corresponding to the second network television device. The second network television device is a network television device obtained by screening each network television device according to the network television information corresponding to each network television device in the network television device set.

[0044] In this embodiment, the target feature extraction model is a model trained according to the second image data corresponding to the second network television device.

[0045] The second network television device is a network television device obtained by screening each network television device according to the network television information corresponding to each network television device in the network television device set.

[0046] For example, the data analysis system obtains the network television information corresponding to each network television device in the network television device set, and then screens each network television device according to the network television information corresponding to each network television device to determine that the network television device of important value is the second network television device.

[0047] Obtain the second image data corresponding to the second network television device, and train the feature extraction model through the second image data, so as to obtain the target feature extraction model.

[0048] Then, input the first image data corresponding to the first network TV device into the target feature extraction model, and the image features corresponding to the first image data can be obtained.

[0049] S104. Perform similarity matching between the image features corresponding to the first image data and each candidate recommended video in the candidate recommended video set to generate a first recommended video to be recommended. The first recommended video to be recommended is the first preset number of candidate recommended videos selected from the candidate recommended video set in descending order of similarity matching.

[0050] In this embodiment, the first recommended video to be recommended is the first preset number of candidate recommended videos selected from the candidate recommended video set in descending order of similarity matching.

[0051] For example, after the data analysis system obtains the image features corresponding to the first image data, it performs similarity matching between the image features corresponding to the first image data and each candidate recommended video in the candidate recommended video set, and then sequentially selects the first preset number of candidate recommended videos in descending order of similarity matching as the first recommended video to be recommended.

[0052] Through this embodiment, obtain the device identification information of the first network TV device, then based on the device identification information, collect the first image data through the soft probe of the set-top box corresponding to the first network TV device, and then based on the image features corresponding to the first image data, the first recommended video to be recommended can be determined. In this way, by automatically collecting the first image data, obtaining the corresponding image features according to the first image data, and directly determining the video recommendation result according to the image features, there is no need to wait for the user to give a rating feedback, reducing the dependence on the user's autonomy and overcoming the limitation of the applicable range caused by the user's unwillingness to give a rating feedback. At the same time, in the embodiment of the present application, each network TV device in the network TV device set is screened according to the network TV information corresponding to each network TV device to obtain a second network TV device, and the target feature extraction model is trained based on the second image data corresponding to the second network TV device. In this way, it is possible to preferentially collect the image data of key network TV devices, reduce the collection of image data of invalid network TV devices, and relieve the pressure of data collection.

[0053] As an optional embodiment, as Figure 2 shown, the data processing method may further include the following steps S201 to S206.

[0054] S201. Screen each network TV device based on the network TV information corresponding to each network TV device in the network TV device set to determine a second network TV device.

[0055] In this embodiment, the second network television device is a network television device screened according to the network television information corresponding to each network television device in the network television device set.

[0056] For example, the data analysis system obtains the network television information corresponding to each network television device in the network television device set, and then screens each network television device according to the network television information corresponding to each network television device to determine that the relatively important network television device is the second network television device.

[0057] S202. Obtain second image data corresponding to the second network television device, where the second image data includes an image sample and a corresponding image label.

[0058] In this embodiment, the second image data is any image data in the historical playback video of the second network television device and any image data in the current playback video of the second network television device, and the second image data includes an image sample and a corresponding image label.

[0059] For example, the data analysis system determines the set-top box corresponding to the second network television device, and then obtains the image samples in the historical playback video and the current playback video through the soft probe of the set-top box corresponding to the second network television device, and labels the categories of each image sample to generate corresponding image labels.

[0060] S203. Input the second image data into the feature extraction model to obtain predicted image features corresponding to the second image data.

[0061] In this embodiment, the predicted image features are the image features of the second image data obtained according to the feature extraction model.

[0062] For example, input the second image data corresponding to the second network television device into the feature extraction model to obtain the predicted image features corresponding to the second image data.

[0063] S204. Based on the image label corresponding to the second image data, determine the category feature corresponding to the second image data in the image feature set, where the image feature set includes category features corresponding to at least one image label.

[0064] In this embodiment, the image feature set includes category features corresponding to various image labels, and the category feature is a feature representing the category attribute determined by processing the image features of the image data of the same category.

[0065] Specifically, the category feature corresponding to the image label is determined by the following formula 1:

[0066]

[0067] Among them, c k represents the category feature corresponding to the k-th type of image label, and h k represents the set of image data of the k-th type of image label. |h k | represents the number of image data included in the set of image data corresponding to the k-th type of image label. u i represents the predicted image feature corresponding to the i-th image data in the set of image data.

[0068] After determining the category features corresponding to various types of image labels through the above formula 1, an image feature set is generated according to the category features corresponding to various types of image labels.

[0069] Based on the image label corresponding to the second image data, the data analysis system can determine the category feature corresponding to the second image data in the image feature set.

[0070] S205. Based on the predicted image feature corresponding to the second image data and the category feature corresponding to the second image data, determine the category loss function.

[0071] In this embodiment, the category loss function is used to characterize the feature distance between the predicted image features corresponding to the image data of the same type of image label.

[0072] Specifically, the category loss function is determined through the following formula 2:

[0073]

[0074] Among them, L q represents the category loss function corresponding to the second image data, u q represents the predicted image feature corresponding to the second image data, c + represents the category feature corresponding to the image label of the second image data, and c k represents the category feature corresponding to the k-th type of image label, and τ is the temperature hyperparameter.

[0075] For example, based on the predicted image feature corresponding to the second image data and the category feature corresponding to the second image data, the data analysis system can determine the category loss function corresponding to the second image data.

[0076] S206. Based on the category loss function, perform iterative training on the feature extraction model until the target feature extraction model is obtained.

[0077] In this embodiment, the data analysis system iteratively trains the feature extraction model through a category loss function, uses ResNet-50 as the neural network structure, and at the same time uses the Adam optimizer to optimize the network. The learning rate is set to 3.5e-4, and the learning rate is decreased by 10 times every 20 epochs. After training for a total of 50 epochs, the target feature extraction model can be obtained.

[0078] Through this embodiment, based on the predicted image features corresponding to the second image data and the category features corresponding to the second image data, the category loss function is determined, and then the feature extraction model is iteratively trained through the category loss function, so that the target feature extraction model can be obtained, which helps to extract more discriminative image features and improves the accuracy of image feature extraction.

[0079] As an alternative embodiment, S201 may specifically include:

[0080] Obtain the network TV information of the target quantity of each network TV device in the network TV device set;

[0081] Based on the network TV information of the target quantity of each network TV device, determine the network TV level of each network TV device;

[0082] Based on the network TV levels of each network TV device, determine the network TV devices that meet the preset level conditions as the second network TV devices.

[0083] In this embodiment, the network TV information is used to characterize the usage of the network TV device, and the network TV level is used to characterize the importance of the network TV device.

[0084] For example, the data analysis system collects the network TV information of the target quantity of each network TV device based on the preset acquisition configuration conditions according to each network TV device included in the network TV device set.

[0085] Then, according to the network TV information of the target quantity of each network TV device, determine the network TV level corresponding to each network TV device. Exemplarily, the network TV level may include an important value level, an important development level, an important maintenance level, an important retention level, a general value level, a general development level, a general maintenance level, and a general retention level.

[0086] Finally, according to the network TV levels of each network TV device, determine the network TV devices that meet the preset level conditions as the second network TV devices. Exemplarily, the network TV devices with the network TV levels of important value level, important development level, important maintenance level, and important retention level can be determined as the second network TV devices.

[0087] In this embodiment, the IPTV levels of each IPTV device are determined according to the IPTV information, and then, based on the IPTV levels of each IPTV device, the IPTV devices that meet the preset level conditions are determined as the second IPTV devices. By preferentially collecting image data for relatively important IPTV devices and excluding some invalid IPTV devices, not only is the data collection pressure relieved, but also the waste of server resources is avoided.

[0088] As an alternative embodiment, based on the IPTV information of the target quantity of each IPTV device, determining the IPTV levels of each IPTV device may specifically include:

[0089] Determining whether the IPTV information of the IPTV device meets the target threshold condition;

[0090] Processing the IPTV information that meets the target threshold condition to determine the playback feature information of the IPTV device;

[0091] Based on the playback center information and the playback feature information, determining the IPTV levels of the IPTV devices, where the playback center information is information generated by aggregating the IPTV information that meets the target threshold condition through the target distance algorithm.

[0092] In this embodiment, the playback feature information is feature information characterizing the importance of the IPTV device determined based on the IPTV information of the IPTV device that meets the target threshold condition. The playback center information is used to characterize the central value of each IPTV information.

[0093] For example, after the data analysis system collects the IPTV information of the IPTV device, it cleans the dirty data in the IPTV information, where the dirty data includes blank data, abnormal data, and special character data.

[0094] Then, a target threshold interval [T, S] is set, and then the IPTV information of the IPTV device is compared with the threshold T and the threshold S respectively, so as to determine whether the IPTV information of the IPTV device is within the target threshold interval. When the IPTV information is within the target threshold interval, it is determined that the target threshold condition is met.

[0095] Then, according to the IPTV information that meets the target threshold condition, the average value is calculated, and the average value is determined as the playback feature information of the IPTV device.

[0096] Then, the playback center information is generated by aggregating the IPTV information that meets the target threshold condition through the target distance algorithm, where the target distance algorithm may be the Euclidean distance algorithm. Specifically, by continuously iterating through the following formula 3, the playback center information can be determined:

[0097]

[0098] Among them, dist(x i , x j ) represents the playback center information, and x iu , x ju represent any two Internet TV information that meet the target threshold conditions.

[0099] Finally, compare the playback feature information with the playback center information to determine the Internet TV level of the Internet TV device.

[0100] Through this embodiment, according to the Internet TV information that meets the target threshold conditions, the playback feature information of the Internet TV device can be determined, and then according to the playback feature information of the Internet TV device, the Internet TV level of the Internet TV device can be determined, which helps to determine important Internet TV devices and relieve the data collection pressure.

[0101] As an alternative embodiment, obtaining the second image data corresponding to the second Internet TV device may specifically include:

[0102] Collecting image samples of the second Internet TV device through the soft probe of the set-top box corresponding to the second Internet TV device;

[0103] Clustering the image samples of the second Internet TV device through a clustering algorithm to generate image labels corresponding to the image samples;

[0104] Generating the second image data corresponding to the second Internet TV device based on the image samples and the image labels corresponding to the image samples.

[0105] In this embodiment, the data analysis system first intercepts the image data in the historical playback video and the current playback video of the second Internet TV device through the soft probe of the set-top box corresponding to the second Internet TV device according to the preset acquisition configuration information, so as to obtain a target preset number of image samples.

[0106] Then, cluster the image samples of the second Internet TV device through the DBSCAN clustering algorithm, and set the image samples belonging to the same class to the same class label, so as to generate image labels corresponding to each image sample.

[0107] Finally, generate the second image data corresponding to the second Internet TV device based on the image samples and the image labels corresponding to the image samples.

[0108] Through this embodiment, clustering the image samples of the second Internet TV device through a clustering algorithm to generate image labels corresponding to the image samples, and directly generating image labels corresponding to the image samples through the clustering method, which helps to overcome the drawbacks of manual annotation in traditional methods.

[0109] As an alternative embodiment, the data processing method may specifically further include:

[0110] Generating class update features for each image label based on the momentum update parameter and the class features corresponding to each image label;

[0111] Updating the image feature set based on the class update features of each image label.

[0112] In this embodiment, the class update features for each image label are generated by the following formula 4:

[0113] c m ← mc m +(1 - m).c m- Formula 4

[0114] where m represents the momentum update parameter, c m represents the class update feature corresponding to the m-th class of image labels, and c m- represents the class feature at the previous moment corresponding to the m-th class of image labels.

[0115] Specifically, the data analysis system generates the class update features for each image label based on the momentum update parameter, and then updates the image feature set according to the class update features of each image label.

[0116] Through this embodiment, updating the image feature set based on the momentum update parameter helps to update the class features in the image feature set in real time during the training of the feature extraction model, so as to adapt to the requirements of the feature extraction model training and improve the accuracy of the feature extraction model training.

[0117] As an alternative embodiment, the network TV information includes at least one of play time information, play amount information, and play frequency information.

[0118] In this embodiment, when the network TV information includes play time information, play amount information, and play frequency information, target threshold intervals for the play time information, play amount information, and play frequency information are respectively set, and then the play time information, play amount information, and play frequency information of the network TV device are respectively compared with their corresponding thresholds, so as to determine whether the network TV information of the network TV device is within the target threshold intervals. When the play time information, play amount information, and play frequency information are all within the corresponding target threshold intervals, it is determined that the target threshold condition is satisfied.

[0119] Then, according to the play time information, play amount information, and play frequency information that satisfy the target threshold condition, the corresponding average values are calculated.

[0120] Then, compare the average values of the play time information, play amount information, and play frequency information with the corresponding play center information. If it is less than the corresponding play center information, set the corresponding index information to 1; if it is not less than the corresponding play center information, set the corresponding index information to 0.

[0121] Finally, set the network TV level of the network TV device with the index information of the play time information, play amount information, and play frequency information all being 1 to the important value level; set the network TV level of the network TV device with the index information of the play time information and play amount information being 1 and the index information of the play amount frequency being 0 to the important development level; set the network TV level of the network TV device with the index information of the play amount information and play frequency information being 1 and the index information of the play time information being 0 to the important retention level; set the network TV level of the network TV device with the index information of the play amount information being 1 and the index information of the play time information and play frequency information being 0 to the important retention level.

[0122] Set the network TV level of the network TV device with the index information of the play amount information being 0 and the index information of the play time information and play frequency information being 1 to the general value level; set the network TV level of the network TV device with the index information of the play time information being 1 and the index information of the play amount information and play frequency information being 0 to the general development level; set the network TV level of the network TV device with the index information of the play frequency information being 1 and the index information of the play amount information and play time information being 0 to the general retention level; set the network TV level of the network TV device with the index information of the play time information, play amount information, and play frequency information all being 0 to the general retention level.

[0123] Through this embodiment, determine the network TV levels of each network TV device according to the play time information, play amount information, and play frequency information. By preferentially collecting image data of relatively important network TV devices and excluding some invalid network TV devices, not only the data collection pressure is relieved, but also the waste of server resources is avoided.

[0124] As an alternative embodiment, the data processing method may further specifically include:

[0125] In response to the user's operation of selecting a target analysis type from the candidate analysis types, determine the type attributes corresponding to the target analysis type of each first image data of the first network TV device, and the candidate analysis types at least include a content preference analysis type and a person preference analysis type;

[0126] Record the occurrence frequencies of the type attributes corresponding to each first image data and the target analysis type;

[0127] Based on the occurrence frequencies of the type attributes, determine a second video to be recommended. The second video to be recommended is a video that meets the target type attribute among the first videos to be recommended, and the target type attribute is the second preset number of type attributes selected in the order of decreasing occurrence frequencies of the type attributes.

[0128] In this embodiment, the candidate analysis types include a content preference analysis type and a person preference analysis type.

[0129] The content preference analysis type is used to represent the analysis of the specific categories of video content. Exemplarily, the type attributes corresponding to the content preference analysis type include eight type attributes: movie, TV drama, music, variety show, anime, children's, sports, and game.

[0130] The person preference analysis type is used to represent the analysis of the people in the video content. Exemplarily, the type attributes corresponding to the person preference analysis type include the names that appear in the video content.

[0131] For example, the data analysis system provides a user query interface. The user checks the target analysis type in the candidate analysis types of the user query interface, and then determines the type attributes corresponding to each first image data of the first Internet TV device and the target analysis type according to the target analysis type selected by the user.

[0132] Then, count the occurrence frequencies of each type attribute, select the second preset number of type attributes in the order of decreasing occurrence frequencies as the target type attributes, and then determine the videos that meet the target type attributes in the first videos to be recommended as the second videos to be recommended.

[0133] Through this embodiment, according to the target analysis type selected by the user, a more accurate second video to be recommended is selected from the first videos to be recommended, which can accurately locate the preference type of the first Internet TV device and facilitate formulating a customized marketing plan for the first Internet TV device.

[0134] Based on the data processing method. Correspondingly, the present application also provides a specific embodiment of a data processing device.

[0135] As Figure 3 shown, the data processing device provided by the embodiment of the present application includes an information acquisition module 310, a data acquisition module 320, a feature extraction module 330, and a similarity matching module 340.

[0136] The information acquisition module 310 is used to acquire the device identification information of the first Internet TV device, and the device identification information is used to identify the set-top box corresponding to the first Internet TV device.

[0137] A data acquisition module 320, configured to collect first image data through a soft probe of a set-top box corresponding to a first Internet TV device based on device identification information, where the first image data includes images corresponding to the historical playback video and the current playback video of the first Internet TV device.

[0138] A feature extraction module 330, configured to input the first image data into a target feature extraction model to obtain image features corresponding to the first image data, where the target feature extraction model is a model trained according to second image data corresponding to a second Internet TV device, and the second Internet TV device is an Internet TV device obtained by screening each Internet TV device according to the Internet TV information corresponding to each Internet TV device in the Internet TV device set.

[0139] A similarity matching module 340, configured to perform similarity matching between the image features corresponding to the first image data and each candidate recommended video in a candidate recommended video set to generate a first to-be-recommended video, where the first to-be-recommended video is the first preset number of candidate recommended videos selected from the candidate recommended video set in descending order of similarity matching.

[0140] Through this embodiment, the device identification information of the first Internet TV device is obtained, and then based on the device identification information, the first image data is collected through the soft probe of the set-top box corresponding to the first Internet TV device, and then based on the image features corresponding to the first image data, the first to-be-recommended video can be determined. In this way, by automatically collecting the first image data, obtaining the corresponding image features according to the first image data, and directly determining the video recommendation result according to the image features, it is not necessary to wait for the user to give a rating feedback, reducing the dependence on the user's autonomy and overcoming the limitation of the applicable range caused by the user's unwillingness to give a rating feedback. At the same time, in the embodiment of the present application, each Internet TV device in the Internet TV device set is screened according to the Internet TV information corresponding to each Internet TV device to obtain a second Internet TV device, and a target feature extraction model is trained based on the second image data corresponding to the second Internet TV device. In this way, it is possible to preferentially collect the image data of key Internet TV devices and reduce the collection of image data of invalid Internet TV devices, alleviating the pressure of data collection.

[0141] As an alternative embodiment, as Figure 4 shown, the data processing device may specifically further include:

[0142] A device screening module 410, configured to screen each Internet TV device based on the Internet TV information corresponding to each Internet TV device in the Internet TV device set to determine a second Internet TV device;

[0143] A data acquisition module 420, configured to acquire second image data corresponding to a second network television device, where the second image data includes an image sample and a corresponding image label;

[0144] A data input module 430, configured to input the second image data into a feature extraction model to obtain predicted image features corresponding to the second image data;

[0145] A feature determination module 440, configured to determine category features corresponding to the second image data from an image feature set based on the image label corresponding to the second image data, where the image feature set includes category features corresponding to at least one image label;

[0146] A loss determination module 450, configured to determine a category loss function based on the predicted image features corresponding to the second image data and the category features corresponding to the second image data;

[0147] A model training module 460, configured to iteratively train the feature extraction model based on the category loss function until a target feature extraction model is obtained.

[0148] As an optional embodiment, the device screening module 410 specifically includes the following units:

[0149] An information acquisition unit, configured to acquire network television information of a target number of each network television device in a network television device set;

[0150] A level determination unit, configured to determine the network television level of each network television device based on the network television information of a target number of each network television device;

[0151] A device determination unit, configured to determine, based on the network television levels of each network television device, the network television devices that meet a preset level condition as the second network television devices.

[0152] As an optional embodiment, the level determination unit specifically includes the following subunits:

[0153] A condition judgment subunit, configured to judge whether the network television information of a network television device meets a target threshold condition;

[0154] An information determination subunit, configured to process the network television information that meets the target threshold condition to determine the playback feature information of the network television device;

[0155] A level determination subunit, configured to determine the network television level of the network television device based on the playback center information and the playback feature information, where the playback center information is information generated by aggregating data of the network television information that meets the target threshold condition through a target distance algorithm.

[0156] As an alternative embodiment, the data acquisition module 420 specifically includes the following units:

[0157] A sample acquisition unit, configured to acquire an image sample of the second network television device through a soft probe of the set-top box corresponding to the second network television device;

[0158] A sample clustering unit, configured to cluster the image samples of the second network television device through a clustering algorithm to generate image labels corresponding to the image samples;

[0159] A data generation unit, configured to generate second image data corresponding to the second network television device based on the image samples and the image labels corresponding to the image samples.

[0160] As an alternative embodiment, the data processing device may specifically further include:

[0161] A feature update module, configured to generate class update features of each image label based on momentum update parameters and class features corresponding to each image label;

[0162] A feature set update module, configured to update the image feature set based on the class update features of each image label.

[0163] As an alternative embodiment, the data processing device may specifically further include:

[0164] A type determination module, configured to determine type attributes corresponding to each first image data of the first network television device and a target analysis type in response to an operation of a user selecting the target analysis type from candidate analysis types, where the candidate analysis types at least include a content preference analysis type and a person preference analysis type;

[0165] A frequency recording module, configured to record the occurrence frequencies of the type attributes corresponding to each first image data and the target analysis type;

[0166] A video recommendation module, configured to determine a second video to be recommended based on the occurrence frequencies of the type attributes, where the second video to be recommended is a video that meets the target type attributes among the first videos to be recommended, and the target type attributes are the second preset number of type attributes selected in descending order of the occurrence frequencies of the type attributes.

[0167] Figure 5 FIG. shows a schematic hardware structure diagram of a data processing device provided in an embodiment of the present application.

[0168] The data processing device may include a processor 501 and a memory 502 storing computer program instructions.

[0169] Specifically, the above-mentioned processor 501 may include a central processing unit (CPU), or an application specific integrated circuit (ASIC), or may be configured as one or more integrated circuits implementing the embodiments of the present application.

[0170] The memory may include a read-only memory (ROM), a random access memory (RAM), a magnetic disk storage media device, an optical storage media device, a flash memory device, an electrical, optical, or other physical / tangible memory storage device. Thus, generally, the memory includes one or more tangible (non-transitory) computer-readable storage media (e.g., memory devices) encoded with software including computer-executable instructions, and when the software is executed (e.g., by one or more processors), it is operable to perform the operations described with reference to the method according to an aspect of the present disclosure.

[0171] The processor 501 reads and executes the computer program instructions stored in the memory 502 to implement any of the data processing methods in the above embodiments.

[0172] In one example, the data processing device may further include a communication interface 503 and a bus 510. As shown, Figure 5 the processor 501, the memory 502, and the communication interface 503 are connected via the bus 510 and complete communication with each other.

[0173] The communication interface 503 is mainly used to implement communication between various modules, devices, units, and / or devices in the embodiments of the present application.

[0174] The bus 510 includes hardware, software, or both, coupling the components of the data processing device to each other. By way of example and not limitation, the bus may include an Accelerated Graphics Port (AGP) or other graphics bus, an Extended Industry Standard Architecture (EISA) bus, a Front Side Bus (FSB), a HyperTransport (HT) interconnect, an Industry Standard Architecture (ISA) bus, an InfiniBand interconnect, a Low Pin Count (LPC) bus, a memory bus, a MicroChannel Architecture (MCA) bus, a Peripheral Component Interconnect (PCI) bus, a PCI-Express (PCI-X) bus, a Serial Advanced Technology Attachment (SATA) bus, a Video Electronics Standards Association Local (VLB) bus, or other suitable bus or a combination of two or more of these. In suitable cases, the bus 510 may include one or more buses. Although the embodiments of the present application describe and illustrate specific buses, the present application contemplates any suitable bus or interconnect.

[0175] In addition, in combination with the data processing method in the above embodiments, an embodiment of the present application can be implemented by providing a computer storage medium. Computer program instructions are stored on the computer storage medium; when the computer program instructions are executed by a processor, any one of the data processing methods in the above embodiments is implemented.

[0176] It should be clear that the present application is not limited to the specific configurations and processes described above and shown in the figures. For the sake of brevity, detailed descriptions of known methods are omitted here. In the above embodiments, several specific steps are described and shown as examples. However, the method process of the present application is not limited to the specific steps described and shown. Those skilled in the art can make various changes, modifications, and additions, or change the order between steps after understanding the spirit of the present application.

[0177] The functional blocks shown in the above structure block diagrams can be implemented as hardware, software, firmware, or a combination thereof. When implemented in hardware, it can be, for example, an electronic circuit, an application-specific integrated circuit (ASIC), appropriate firmware, a plug-in, a functional card, and so on. When implemented in software, the elements of the present application are programs or code segments used to perform the required tasks. The program or code segment can be stored in a machine-readable medium or transmitted via a data signal carried in a carrier wave on a transmission medium or a communication link. "Machine-readable medium" can include any medium capable of storing or transmitting information. Examples of machine-readable media include electronic circuits, semiconductor memory devices, ROM, flash memory, erasable ROM (EROM), floppy disks, CD-ROMs, optical discs, hard disks, fiber optic media, radio frequency (RF) links, and so on. The code segment can be downloaded via a computer network such as the Internet, an intranet, and so on.

[0178] It should also be noted that the exemplary embodiments mentioned in the present application describe some methods or systems based on a series of steps or devices. However, the present application is not limited to the order of the above steps, that is, the steps can be executed in the order mentioned in the embodiments, or different from the order in the embodiments, or several steps can be executed simultaneously.

[0179] As described above with reference to the flowcharts and / or block diagrams of methods, apparatuses (systems), and computer program products according to embodiments of the present disclosure. It should be understood that each block in the flowchart and / or block diagram, and the combination of blocks in the flowchart and / or block diagram, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, or other programmable data processing device to generate a machine, such that the instructions executed by the processor of the computer or other programmable data processing device enable the implementation of the functions / actions specified in one or more blocks of the flowchart and / or block diagram. Such a processor can be, but is not limited to, a general-purpose processor, a special-purpose processor, a special application processor, or a field programmable logic circuit. It is also understood that each block in the block diagram and / or flowchart, and the combination of blocks in the block diagram and / or flowchart, can also be implemented by dedicated hardware that performs the specified functions or actions, or can be implemented by a combination of dedicated hardware and computer instructions.

[0180] As described above, the foregoing are only specific embodiments of the present application. Those skilled in the art can clearly understand that for the convenience and brevity of description, the specific working processes of the systems, modules, and units described above can refer to the corresponding processes in the foregoing method embodiments and will not be elaborated herein. It should be understood that the protection scope of the present application is not limited thereto. Any person skilled in the art can easily think of various equivalent modifications or substitutions within the technical scope disclosed in the present application, and these modifications or substitutions should all be covered by the protection scope of the present application.

Claims

1. A data processing method, characterized in that, The method includes: Obtaining device identification information of a first network television device, where the device identification information is used to identify a set-top box corresponding to the first network television device; Based on the device identification information, collecting first image data through a soft probe of the set-top box corresponding to the first network television device, where the first image data includes images corresponding to historical playback videos and current playback videos of the first network television device; Inputting the first image data into a target feature extraction model to obtain image features corresponding to the first image data, where the target feature extraction model is a model trained according to second image data of a second network television device, and the second network television device is a network television device obtained by screening each network television device according to network television information corresponding to each network television device in a network television device set; Performing similarity matching between the image features corresponding to the first image data and each candidate recommended video in a candidate recommended video set to generate a first to-be-recommended video, where the first to-be-recommended video is the first preset number of candidate recommended videos selected in descending order of similarity matching in the candidate recommended video set.

2. The method according to claim 1, wherein The method further includes: Screening each network television device according to network television information corresponding to each network television device in a network television device set to determine a second network television device; Obtaining second image data of the second network television device, where the second image data includes image samples and corresponding image labels; Inputting the second image data into a feature extraction model to obtain predicted image features corresponding to the second image data; Based on the image labels corresponding to the second image data, determining class features corresponding to the second image data in an image feature set, where the image feature set includes class features corresponding to at least one image label; Based on the predicted image features corresponding to the second image data and the class features corresponding to the second image data, determining a class loss function; Based on the class loss function, performing iterative training on the feature extraction model until a target feature extraction model is obtained.

3. The method according to claim 2, wherein The screening each network television device according to network television information corresponding to each network television device in a network television device set to determine a second network television device includes: Obtaining a target number of network television information of each network television device in a network television device set; Based on the target number of network television information of each network television device, determining the network television level of each network television device; Based on the network television levels of each network television device, determining the network television devices that meet the preset level condition as second network television devices.

4. The method according to claim 3, characterized in that, The determining the network television level of each network television device based on the target number of network television information of each network television device includes: Judging whether each network television information of a network television device meets a target threshold condition; Processing the network television information that meets the target threshold condition to determine playback feature information of the network television device; Based on the play center information and the play feature information, determine the network TV level of the network TV device, where the play center information is information generated by aggregating data of each network TV information that meets the target threshold condition through a target distance algorithm.

5. The method according to any one of claims 2-4, characterized in that, The obtaining of the second image data corresponding to the second network TV device includes: Collecting an image sample of the second network TV device through a soft probe of the set-top box corresponding to the second network TV device; Clustering the image samples of the second network TV device through a clustering algorithm to generate image labels corresponding to the image samples; Generating the second image data corresponding to the second network TV device based on the image samples and the image labels corresponding to the image samples.

6. The method according to any one of claims 2-4, characterized in that The method further includes: Generating category update features of each image label based on the momentum update parameter and the category features corresponding to each image label; Updating the image feature set based on the category update features of each image label.

7. The method according to any one of claims 1-4, characterized in that, The network TV information includes at least one of play time information, play amount information, and play frequency information.

8. The method according to any one of claims 1 to 4, characterized in that, The method further includes: In response to an operation in which a user selects a target analysis type from candidate analysis types, determining type attributes corresponding to the target analysis type for each of the first image data of the first network TV device, where the candidate analysis types at least include a content preference analysis type and a person preference analysis type; Recording the occurrence frequency of the type attributes corresponding to each of the first image data and the target analysis type; Determining a second video to be recommended based on the occurrence frequency of the type attributes, where the second video to be recommended is a video that meets the target type attributes among the first videos to be recommended, and the target type attributes are the second preset number of type attributes selected in descending order of the occurrence frequency of the type attributes.

9. A data processing device, characterized in that, The apparatus includes: An information acquisition module, configured to acquire device identification information of a first network TV device, where the device identification information is used to identify a set-top box corresponding to the first network TV device; A data collection module, configured to collect first image data through a soft probe of the set-top box corresponding to the first network TV device based on the device identification information, where the first image data includes images corresponding to historical play videos and current play videos of the first network TV device; A feature extraction module, configured to input the first image data into a target feature extraction model to obtain image features corresponding to the first image data, where the target feature extraction model is a model trained according to second image data corresponding to a second network TV device, and the second network TV device is a network TV device selected from each network TV device in a network TV device set according to network TV information corresponding to each network TV device; A similarity matching module, configured to perform similarity matching between the image features corresponding to the first image data and each candidate recommended video in a candidate recommended video set, and generate a first to-be-recommended video, where the first to-be-recommended video is the first preset number of candidate recommended videos selected from the candidate recommended video set in descending order of similarity matching.

10. An electronic device, characterized in that, The device includes: a processor and a memory storing computer program instructions; When the processor executes the computer program instructions, the data processing method according to any one of claims 1-8 is implemented.

11. A computer-readable storage medium, characterized in that, Computer program instructions are stored on the computer-readable storage medium, and when the computer program instructions are executed by a processor, the data processing method according to any one of claims 1-8 is implemented.

Images