CN111385659A - Video recommendation method, device, equipment and storage medium - Google Patents

Video recommendation method, device, equipment and storage medium Download PDF

Info

Publication number
CN111385659A
CN111385659A CN201811640356.6A CN201811640356A CN111385659A CN 111385659 A CN111385659 A CN 111385659A CN 201811640356 A CN201811640356 A CN 201811640356A CN 111385659 A CN111385659 A CN 111385659A
Authority
CN
China
Prior art keywords
video
historical
recommended
classification
recommendation model
Prior art date
Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
Granted
Application number
CN201811640356.6A
Other languages
Chinese (zh)
Other versions
CN111385659B (en
Inventor
刘运
刘文奇
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
Bigo Technology Pte Ltd
Original Assignee
Guangzhou Baiguoyuan Information Technology Co Ltd
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by Guangzhou Baiguoyuan Information Technology Co Ltd filed Critical Guangzhou Baiguoyuan Information Technology Co Ltd
Priority to CN201811640356.6A priority Critical patent/CN111385659B/en
Priority to PCT/CN2019/124582 priority patent/WO2020135054A1/en
Publication of CN111385659A publication Critical patent/CN111385659A/en
Application granted granted Critical
Publication of CN111385659B publication Critical patent/CN111385659B/en
Active legal-status Critical Current
Anticipated expiration legal-status Critical

Links

Images

Classifications

    • H—ELECTRICITY
    • H04—ELECTRIC COMMUNICATION TECHNIQUE
    • H04N—PICTORIAL COMMUNICATION, e.g. TELEVISION
    • H04N21/00—Selective content distribution, e.g. interactive television or video on demand [VOD]
    • H04N21/40—Client devices specifically adapted for the reception of or interaction with content, e.g. set-top-box [STB]; Operations thereof
    • H04N21/45—Management operations performed by the client for facilitating the reception of or the interaction with the content or administrating data related to the end-user or to the client device itself, e.g. learning user preferences for recommending movies, resolving scheduling conflicts
    • H04N21/466—Learning process for intelligent management, e.g. learning user preferences for recommending movies
    • H04N21/4668—Learning process for intelligent management, e.g. learning user preferences for recommending movies for recommending content, e.g. movies
    • H—ELECTRICITY
    • H04—ELECTRIC COMMUNICATION TECHNIQUE
    • H04N—PICTORIAL COMMUNICATION, e.g. TELEVISION
    • H04N21/00—Selective content distribution, e.g. interactive television or video on demand [VOD]
    • H04N21/20—Servers specifically adapted for the distribution of content, e.g. VOD servers; Operations thereof
    • H04N21/25—Management operations performed by the server for facilitating the content distribution or administrating data related to end-users or client devices, e.g. end-user or client device authentication, learning user preferences for recommending movies
    • H—ELECTRICITY
    • H04—ELECTRIC COMMUNICATION TECHNIQUE
    • H04N—PICTORIAL COMMUNICATION, e.g. TELEVISION
    • H04N21/00—Selective content distribution, e.g. interactive television or video on demand [VOD]
    • H04N21/20—Servers specifically adapted for the distribution of content, e.g. VOD servers; Operations thereof
    • H04N21/25—Management operations performed by the server for facilitating the content distribution or administrating data related to end-users or client devices, e.g. end-user or client device authentication, learning user preferences for recommending movies
    • H04N21/251—Learning process for intelligent management, e.g. learning user preferences for recommending movies
    • H—ELECTRICITY
    • H04—ELECTRIC COMMUNICATION TECHNIQUE
    • H04N—PICTORIAL COMMUNICATION, e.g. TELEVISION
    • H04N21/00—Selective content distribution, e.g. interactive television or video on demand [VOD]
    • H04N21/40—Client devices specifically adapted for the reception of or interaction with content, e.g. set-top-box [STB]; Operations thereof
    • H04N21/45—Management operations performed by the client for facilitating the reception of or the interaction with the content or administrating data related to the end-user or to the client device itself, e.g. learning user preferences for recommending movies, resolving scheduling conflicts
    • H04N21/466—Learning process for intelligent management, e.g. learning user preferences for recommending movies
    • H—ELECTRICITY
    • H04—ELECTRIC COMMUNICATION TECHNIQUE
    • H04N—PICTORIAL COMMUNICATION, e.g. TELEVISION
    • H04N21/00—Selective content distribution, e.g. interactive television or video on demand [VOD]
    • H04N21/40—Client devices specifically adapted for the reception of or interaction with content, e.g. set-top-box [STB]; Operations thereof
    • H04N21/45—Management operations performed by the client for facilitating the reception of or the interaction with the content or administrating data related to the end-user or to the client device itself, e.g. learning user preferences for recommending movies, resolving scheduling conflicts
    • H04N21/466—Learning process for intelligent management, e.g. learning user preferences for recommending movies
    • H04N21/4662—Learning process for intelligent management, e.g. learning user preferences for recommending movies characterized by learning algorithms
    • H04N21/4665—Learning process for intelligent management, e.g. learning user preferences for recommending movies characterized by learning algorithms involving classification methods, e.g. Decision trees
    • H—ELECTRICITY
    • H04—ELECTRIC COMMUNICATION TECHNIQUE
    • H04N—PICTORIAL COMMUNICATION, e.g. TELEVISION
    • H04N21/00—Selective content distribution, e.g. interactive television or video on demand [VOD]
    • H04N21/40—Client devices specifically adapted for the reception of or interaction with content, e.g. set-top-box [STB]; Operations thereof
    • H04N21/45—Management operations performed by the client for facilitating the reception of or the interaction with the content or administrating data related to the end-user or to the client device itself, e.g. learning user preferences for recommending movies, resolving scheduling conflicts
    • H04N21/466—Learning process for intelligent management, e.g. learning user preferences for recommending movies
    • H04N21/4667—Processing of monitored end-user data, e.g. trend analysis based on the log file of viewer selections

Landscapes

  • Engineering & Computer Science (AREA)
  • Databases & Information Systems (AREA)
  • Multimedia (AREA)
  • Signal Processing (AREA)
  • Computer Vision & Pattern Recognition (AREA)
  • Computing Systems (AREA)
  • Information Retrieval, Db Structures And Fs Structures Therefor (AREA)
  • Two-Way Televisions, Distribution Of Moving Picture Or The Like (AREA)

Abstract

The invention discloses a video recommendation method, a video recommendation device, video recommendation equipment and a storage medium. The method comprises the following steps: acquiring a current video set; inputting the current video set and the historical video set into a historical latest video recommendation model to obtain the classification labels and classification scores of the videos in the current video set; and obtaining the recommended video under the classification label according to the classification scores of the videos under different classification labels in the current video set. Compared with the video recommendation model generated based on deep learning training in the prior art, the historical latest video recommendation model in the embodiment of the invention has a shorter training period, so that the historical video recommendation model can be well adapted to the adjustment of actual services, and meanwhile, the historical video set which can be used as the prior knowledge of the current video set is also used as an input variable, thereby improving the prediction precision of the historical latest video recommendation model on the current video set.

Description

Video recommendation method, device, equipment and storage medium
Technical Field
The embodiment of the invention relates to a computer vision technology, in particular to a video recommendation method, a video recommendation device, video recommendation equipment and a storage medium.
Background
In recent years, with the improvement of computer hardware performance and the appearance of large-scale image data, deep learning is widely applied in the field of computer vision. Video recommendation is an important research direction in the field of computer vision, and for video recommendation, deep learning is widely applied in the aspect of video recommendation, namely, video recommendation is performed by adopting a video recommendation model generated based on deep learning training. The video recommendation model needs to be adjusted correspondingly according to actual service needs.
In the process of implementing the invention, the inventor finds that at least the following problems exist in the prior art: the video recommendation model generated based on the deep learning training cannot be well adapted to the adjustment of the actual service because the training period of the video recommendation model is relatively long, and the adjustment periods of the actual service are relatively short, so that the prediction accuracy of the video recommendation model generated based on the deep learning training is not high, and a proper video cannot be obtained for recommendation. The adjustment of the actual service may refer to operation policy adjustment or change of the hotspot video, the operation policy adjustment may be understood as changing the recommended animation video into the recommended game video, and the change of the hotspot video may be understood as changing the hotspot video from C type to D type, and correspondingly, changing the recommended C type video into the recommended D type video.
Disclosure of Invention
The embodiment of the invention provides a video recommendation method, a video recommendation device, video recommendation equipment and a storage medium, which are used for improving the prediction precision of a video recommendation model.
In a first aspect, an embodiment of the present invention provides a video recommendation method, where the method includes:
acquiring a current video set;
inputting the current video set and the historical video set into a historical latest video recommendation model to obtain the classification labels and classification scores of the videos in the current video set;
and obtaining the recommended video under the classification label according to the classification scores of the videos under different classification labels in the current video set.
Further, the obtaining of the recommended video under the classification label according to the classification scores of the videos under different classification labels in the current video set includes:
obtaining a video to be recommended under a classification label according to the classification scores of videos under different classification labels in the current video set;
and if the classification label of the video to be recommended is consistent with the content of the video to be recommended, taking the video to be recommended as the recommended video under the classification label.
Further, the method further comprises:
and if the classification label of the video to be recommended is inconsistent with the content of the video to be recommended, adding the video to be recommended into the historical video set as an input variable of the historical latest recommendation model after a preset time point is reached, and updating the historical latest video recommendation model.
Further, the method further comprises:
acquiring a target video which meets a recommendation condition in the current video set, and sending the target video;
receiving a target video marked with a classification label;
and when a preset time point is reached, adding the labeled and classified target videos into the historical video set to serve as input variables of the historical latest video recommendation model, and updating the historical latest video recommendation model.
Further, the step of adding the target video labeled with the classification label to the historical video set to serve as an input variable of the historical latest recommendation model, and updating the historical latest recommendation model includes:
assigning different weight values to a first video in the target video, which is the same as the recommended video, and a second video in the target video, which is different from the recommended video;
and when a preset time point is reached, adding the weighted first video and the weighted second video into the historical video set to serve as input variables of the historical latest video recommendation model, and updating the historical latest video recommendation model.
Further, the method further comprises:
and taking a second video which is different from the recommended video in the target video as an additional recommended video.
Further, before the inputting the current video set and the historical video set into the historical latest video recommendation model, the method further includes:
acquiring an original historical video set;
and clustering according to the classification labels of the videos in the original historical video set to obtain a historical video set.
In a second aspect, an embodiment of the present invention further provides a video recommendation apparatus, where the apparatus includes:
the current video set acquisition module is used for acquiring a current video set;
the classification label and classification score acquisition module is used for inputting the current video set and the historical video set into a historical latest video recommendation model to obtain a classification label and a classification score of the video in the current video set;
and the recommended video determining module is used for obtaining the recommended video under the classification label according to the classification scores of the videos under different classification labels in the current video set.
Further, the recommended video determining module includes:
the video to be recommended acquiring unit is used for acquiring videos to be recommended under the classification labels according to the classification scores of the videos under the different classification labels in the current video set;
and the recommended video determining unit is used for taking the video to be recommended as the recommended video under the classification label if the classification label of the video to be recommended is consistent with the content of the video to be recommended.
Further, the apparatus further comprises:
and the first historical latest video recommendation model updating module is used for adding the video to be recommended into the historical video set as an input variable of the historical latest recommendation model after a preset time point is reached and updating the historical latest video recommendation model if the classification label of the video to be recommended is inconsistent with the content of the video to be recommended.
Further, the apparatus further comprises:
the target video determining module is used for acquiring a target video which meets the recommendation condition in the current video set and sending the target video;
the target video classification label marking module is used for receiving the target video with the classification label marked;
and the second historical latest video recommendation model updating module is used for updating the historical latest video recommendation model after the target video with the labels classified is added into the historical video set at a preset time point and is used as an input variable of the historical latest video recommendation model.
Further, the second historical up-to-date video recommendation model updating module includes:
a weight value assigning unit configured to assign different weight values to a first video in the target video that is the same as the recommended video and a second video in the target video that is different from the recommended video;
and the third history latest video recommendation model updating unit is used for updating the history latest video recommendation model by adding the weighted first video and the weighted second video into the history video set when a preset time point is reached and using the added weighted first video and the weighted second video as input variables of the history latest video recommendation model.
Further, the apparatus further comprises:
and the additional recommended video determining module is used for taking a second video which is different from the recommended video in the target video as an additional recommended video.
Further, the apparatus further comprises:
the original video set acquisition module is used for acquiring an original historical video set;
and the historical video set acquisition module is used for carrying out clustering processing according to the classification labels of the videos in the original historical video set to obtain a historical video set.
In a third aspect, an embodiment of the present invention further provides an apparatus, where the apparatus includes:
one or more processors;
a memory for storing one or more programs;
when executed by the one or more processors, cause the one or more processors to implement a method as described in the first aspect of embodiments of the invention.
In a fourth aspect, the present invention further provides a computer-readable storage medium, on which a computer program is stored, which when executed by a processor implements the method according to the first aspect of the present invention.
According to the embodiment of the invention, the classification label and the classification score of the video in the current video set are obtained by acquiring the current video set and inputting the current video set and the historical video set into the historical latest video recommendation model, and the recommended video under the classification label is obtained according to the classification scores of the videos under different classification labels in the current video set.
Drawings
Fig. 1 is a flowchart of a video recommendation method in an embodiment of the present invention;
fig. 2 is a schematic structural diagram of a video recommendation apparatus according to an embodiment of the present invention;
fig. 3 is a schematic structural diagram of an apparatus in an embodiment of the present invention.
Detailed Description
The present invention will be described in further detail with reference to the accompanying drawings and examples. It is to be understood that the specific embodiments described herein are merely illustrative of the invention and are not limiting of the invention. It should be further noted that, for the convenience of description, only some of the structures related to the present invention are shown in the drawings, not all of the structures.
Examples
Fig. 1 is a flowchart of a video recommendation method according to an embodiment of the present invention, where the present embodiment is applicable to a case of improving prediction accuracy of a video recommendation model, and the method may be executed by a video recommendation apparatus, where the apparatus may be implemented in a software and/or hardware manner, and the apparatus may be configured in a device, such as a computer or a mobile terminal. As shown in fig. 1, the method specifically includes the following steps:
and step 110, acquiring a current video set.
And 120, inputting the current video set and the historical video set into a historical latest video recommendation model to obtain the classification labels and the classification scores of the videos in the current video set.
In an embodiment of the present invention, the current video set may be a set composed of videos without classification tags, and more specifically, the current video set may be a set composed of videos uploaded on the current day without classification tags. The historical video set may be a set composed of videos with classification tags, that is, each video in the historical video set has its corresponding classification tag. The category label of the video may refer to a content category corresponding to the video content, that is, the category label of the video may refer to a content category to which the video belongs. Since the historical video set is a set composed of videos with classification labels, the historical video set and the current video set are input into the historical latest video recommendation model together, and the historical latest video recommendation model can be used as prior knowledge of the current video set, so that the prediction accuracy of the historical latest video recommendation model on the current video set is improved.
The historical up-to-date video recommendation model may be used to determine a corresponding classification label for each video in the current video set. In addition, the network structure of the historical latest video recommendation model may be a network structure of a deep learning algorithm, such as a network structure of a conventional convolutional neural network, or a network structure of various lightweight convolutional neural networks formed by improving the network structure of the conventional convolutional neural network. The historical latest video recommendation model can be a video recommendation model updated according to a preset period, wherein the updating is to calculate a loss function of the model through forward propagation according to the updated input variable, calculate a partial derivative of the loss function on the network parameters of the model, and update the network parameters of the model through backward gradient propagation until determining new network parameters. It can be understood that, since the historical latest video recommendation model may be a video recommendation model updated according to a preset update period, the historical latest video recommendation model may refer to a video recommendation model corresponding to a current update period. Based on the above, the source of the updated input variable may be the prediction result of the video recommendation model corresponding to the previous update period.
It should be noted that, the historical latest video recommendation model in the embodiment of the present invention is different from the video recommendation model generated based on deep learning training in the conventional technology, and the video recommendation model that can be applied to actual services in the conventional technology often needs to pass through the following three stages: the method comprises the following steps of a first stage, collecting and labeling data; the second stage, training the model; and a third stage, inputting data in the on-line actual service into the trained video recommendation model, evaluating the prediction precision of the model according to the prediction result of the video recommendation model, adjusting the input variable according to the evaluation result, and retraining the model until the evaluation result meets the application requirement of the actual service. It should be noted that the data in the first stage refers to offline data, and correspondingly, the data in the second stage is offline data as input variables to the model in the process of training the model. It should be further noted that, if the actual service changes, the three phases need to be executed again to obtain the video recommendation model corresponding to the actual service.
The basis for establishing the historical latest video recommendation model in the embodiment of the invention is a pre-training model, the pre-training model is a multi-mode model for learning an overlarge related data set, and the multi-mode model comprises an audio recognition model, an action recognition model, a picture classification model, an image segmentation model, a face recognition model and the like. On the basis of the pre-training model, the video recommendation model directly applied to the actual service can be obtained only by carrying out fine adjustment according to the actual service.
It can be understood that the video recommendation model which can be applied to actual services in the conventional technology cannot achieve a good effect in recommending hot videos. This is because the period of obtaining a video recommendation model that can be applied to actual services in the conventional technology is long, and the update period of the hot video is relatively short, which results in: firstly, when a certain hotspot video is required to be recommended according to a video recommendation model, the video recommendation model is still in a training stage, so that the hotspot video cannot be obtained according to the video recommendation model in the training stage; secondly, when the heat of the hot video is reduced, that is, the hot video does not need to be recommended according to the video recommendation model any more, the video recommendation model can be used for recommending the hot video after the three stages, and a large number of hot videos are recommended, and because the heat of the hot videos is already passed, if the video recommendation model recommends the hot videos, the original effect cannot be achieved, and meanwhile, the user's sense of discomfort can be caused. Based on the above, the video recommendation model that can be applied to the actual service in the conventional technology cannot achieve a good effect in recommending the hot videos. The historical latest video recommendation model provided by the embodiment of the invention can achieve a better effect in recommending the hot videos, and the period for obtaining the historical latest video recommendation model which can be applied to actual services is shorter and can be matched with the update period of the hot videos. Meanwhile, the historical latest video recommendation model is the video recommendation model updated according to the updating period, so that the historical latest video recommendation model can adapt to the change of the hot videos in real time and correspondingly change so as to recommend the hot videos.
Inputting a current video set and a historical video set into a historical latest video recommendation set to obtain original classification labels and original classification scores of all videos in the current video set, sequencing the original classification labels of the videos according to the original classification scores of all the videos, determining the classification labels of the videos according to sequencing results, and taking the original classification scores corresponding to the classification labels of the videos as the classification scores of all the videos. Wherein each raw classification label corresponds to a raw classification score.
It should be noted that the historical latest video recommendation model in the embodiment of the present invention is a multi-classification video recommendation model, that is, the number of original classification tags of each video in the current video set obtained based on the historical latest video recommendation model is two or more. Meanwhile, each original classification label corresponds to one original classification score, that is, the number of the original classification labels is the same as that of the original classification scores, and the difference is the original classification score corresponding to each original classification label.
Illustratively, if the number of the original classification tags recognizable by the history latest video recommendation model is three, and the three original classification tags are music, basketball and originals respectively, the number of the original classification tags of the video a in the current video set is three, and the three original classification tags are music, basketball and originals respectively, and the corresponding original classification scores thereof are 5, 4 and 1 respectively, and the number of the original classification tags of the video B is also three, and the three original classification tags are music, basketball and originals respectively, and the corresponding original classification scores thereof are 1, 2 and 6 respectively. For the video A, sequencing original classification labels of the video A in a descending manner according to the original classification score of the video A, and selecting the first two original classification labels as the classification labels of the video A, namely, using the original classification labels of the video A, namely music and basketball as the classification labels of the video A; and for the video B, sequencing the original classification labels of the video B in a descending manner according to the original classification score of the video B, and selecting the first two original classification labels as the classification labels of the video B, namely, taking the original classification label basketball and the original of the video B as the classification labels of the video B.
And step 130, obtaining the recommended video under the classification label according to the classification scores of the videos under different classification labels in the current video set.
In the embodiment of the invention, since the videos are recommended to the user according to the classification tags of the videos, when the recommended videos are determined, the recommended videos under each classification tag need to be determined. For each classification label, the videos can be sorted according to the classification score corresponding to each video in the current video set, and the recommended video under the classification label is determined according to the sorting result. More specifically, for each classification label, the videos may be sorted in a descending order according to the classification score corresponding to each video in the current video set, and the top M videos are selected as the recommended videos under the classification label. And each classification label performs the same operation until the recommended video under each classification label is determined. It should be noted that the numerical values of M corresponding to different classification tags may be the same or different, and may be determined specifically according to actual situations, and are not specifically limited herein.
Optionally, on the basis of the above technical solution, obtaining the recommended video under the classification label according to the classification scores under different classification labels in the current video set may specifically include: and obtaining the video to be recommended under the classification label according to the classification scores of the videos under different classification labels in the current video set. And if the classification label of the video to be recommended is consistent with the content of the video to be recommended, taking the video to be recommended as the recommended video under the classification label.
In the embodiment of the present invention, because the original classification score corresponding to the original classification label of each video in the current video set obtained based on the historical latest video recommendation model may not be correct, the determined classification label of the video may also be incorrect based on the original classification score of each video, and correspondingly, the classification score corresponding to the classification label of the video may also not be correct, and on this basis, for each classification label, the determined recommended video may also not be correct according to the classification score of the video.
To ensure that the determined recommended video is correct, the following method can be adopted: according to the classification scores of the videos under different classification labels in the current video set, obtaining a video to be recommended under the classification label, determining whether the classification label of the video to be recommended is consistent with the content of the video to be recommended, if the classification label of the video to be recommended is consistent with the content of the video to be recommended, it can be shown that the recommended video under the classification label obtained based on a historical latest video recommendation model is correct, and the video to be recommended can be used as the recommended video under the classification label.
It should be noted that, the above determining whether the classification label of the video to be recommended is consistent with the content of the video to be recommended may be performed by a user having a right to review the classification label of the video.
According to the technical scheme of the embodiment, the classification label and the classification score of the video in the current video set are obtained by acquiring the current video set and inputting the current video set and the historical video set into the historical latest video recommendation model, and the recommended video under the classification label is obtained according to the classification scores of the videos under different classification labels in the current video set.
Optionally, on the basis of the above technical solution, the method may further include: and if the classification label of the video to be recommended is inconsistent with the content of the video to be recommended, adding the video to be recommended into the historical video set after a preset time point is reached, using the video to be recommended as an input variable of the historical latest video recommendation model, and updating the historical latest video recommendation model.
In the embodiment of the present invention, if the classification label of the video to be recommended is inconsistent with the content corresponding to the content of the video to be recommended, it may be indicated that the recommended video under the classification label obtained based on the historical latest video recommendation model is incorrect, and meanwhile, since the historical latest video recommendation model is the video recommendation model corresponding to the current update cycle, in other words, the video recommendation model is the video recommendation model updated according to the update cycle, in order to update the video recommendation model when the preset time point is reached to obtain the historical latest video recommendation model corresponding to the next update cycle, so that the prediction accuracy of the historical latest video recommendation model corresponding to the next update cycle is improved, the video to be recommended whose classification label is inconsistent with the content of the video to be recommended may be added to the historical video set, the method is used as an input variable of the historical latest video recommendation model and participates in the process of updating the historical latest video recommendation model.
The reason why the prediction accuracy of the historical latest video recommendation model corresponding to the next update period can be improved is as follows: the classification label of the video to be recommended and the video to be recommended with inconsistent content of the video to be recommended are regarded as videos for which the classification label of the historical newest video recommendation model is not easy to accurately determine, and the classification label is added into the historical video set to be input into the historical newest video recommendation model as an input variable, so that the characteristics of the video to be recommended can be further learned in the updating process of the historical newest video recommendation model, the identification accuracy of the classification label of the historical newest video recommendation model corresponding to the next updating period is improved, and the prediction accuracy of the historical newest video recommendation model corresponding to the next updating period is improved on the basis.
It should be noted that, here, a video whose classification label of the video to be recommended is inconsistent with the content corresponding to the content of the video to be recommended may be referred to as a difficult negative sample, that is, the difficult negative sample refers to a video of the following category: a certain video in the current video set does not belong to a certain classification label, but is labeled with the video of the classification label by the historical latest video recommendation model with a higher classification score. Wherein, determining whether the labeled classification label of the video is correct can be performed by a user having a rechecking right for the classification label of the video.
Optionally, on the basis of the above technical solution, the method may further include: and acquiring a target video meeting the recommendation condition in the current video set, and sending the target video. And receiving the target video marked with the classification label. And when the preset time point is reached, adding the target video marked with the classification label into the historical video set to serve as an input variable of the historical latest video recommendation model, and updating the historical latest video recommendation model.
In the embodiment of the invention, while the classification labels of all videos in the current video set are obtained based on the historical latest video recommendation model, the target videos meeting the recommendation condition in the current video set can be obtained, the target videos are sent to users with classification label marking authority to label the classification labels, the target videos with the labeled classification labels are received, and after the preset time point is reached, the target videos with the labeled classification labels are added into the historical video set to serve as input variables of the historical latest video recommendation model and participate in the process of updating the historical latest video recommendation model.
Adding the target video recommendation model marked with the classification label into a historical video set as an input variable of the historical latest video recommendation model, wherein the reason for participating in the process of updating the historical latest video recommendation model is as follows: because the user with the classification label labeling authority labels the target videos meeting the recommendation condition in the current video set, the labeling result obtained by labeling the target videos with the classification labels in the above manner (i.e. the classification labeling of the target videos) is generally considered to be relatively correct, and therefore, the labeling result is added into the historical video set and is used as an input variable of the historical latest video recommendation model to participate in the process of updating the historical latest video recommendation model, so that the historical latest video recommendation model can further enhance learning of the characteristics of the target videos, and the prediction accuracy of the historical latest video recommendation model corresponding to the next update period is also improved.
It should be noted that, if the target video recommendation model labeled with the classification tag is added to the history video set and used as an input variable of the history latest video recommendation model to participate in the process of updating the history latest video recommendation model, the target video labeled with the classification tag in the update period which is the longest distance from the current update period needs to be deleted from the history video set.
It should be further noted that the recommendation condition may include at least one of a video play rate, a video praise rate, a video comment rate, a video forwarding rate, and a user collection rate, and correspondingly, if the recommendation condition is a video play rate, the target video meeting the recommendation condition refers to a video whose video play rate is greater than or equal to a play rate threshold; if the recommendation condition is the video point praise rate, the target video meeting the recommendation condition refers to the video with the video point praise rate being greater than or equal to the video with the praise rate threshold; if the recommendation condition is that the video comment rate is greater than or equal to the comment rate threshold, the target video meeting the recommendation condition refers to the video with the comment rate greater than or equal to the comment rate threshold; if the recommendation condition is a video forwarding rate, the target video meeting the recommendation condition refers to a video with the video forwarding rate being greater than or equal to a forwarding rate threshold value; and if the recommendation condition is the user collection rate, the target video meeting the recommendation condition refers to the video with the user collection rate being greater than or equal to the collection rate threshold value. Further, if the recommended conditions are two or more of the above, all of them need to be satisfied. Illustratively, if the recommendation condition is a video play rate and a video praise rate, the target video meeting the recommendation condition refers to a video with the video play rate being greater than or equal to a play rate threshold value and the target video refers to a video with the video praise rate being greater than or equal to a praise rate threshold value; if the recommendation condition is that the video playing rate is greater than or equal to the playing rate threshold, the video point praise rate is greater than or equal to the praise rate threshold, and the video forwarding rate is greater than the forwarding rate threshold, the target video meeting the recommendation condition refers to the video with the video playing rate greater than or equal to the playing rate threshold, the video point praise rate greater than or equal to the praise rate threshold, and the video forwarding rate greater than or equal to the forwarding rate threshold.
Optionally, on the basis of the above technical solution, adding the target video labeled with the classification label to the historical video set to serve as an input variable of the historical latest video recommendation model, and updating the historical latest video recommendation model, which may specifically include: and endowing different weight values for a first video which is the same as the recommended video in the target video and a second video which is not the same as the recommended video in the target video. And when the preset time point is reached, adding the weighted first video and the weighted second video into the historical video set to serve as input variables of the historical latest video recommendation model, and updating the historical latest video recommendation model.
In the embodiment of the invention, the target video is the video meeting the recommendation condition in the current video set, and the recommendation condition is the reflection of the real behavior of the user, so the target video is the video needing to be labeled with the classification label. Based on the above, if the prediction accuracy of the historical latest video recommendation model is high, the target video should be included in the recommended video. It can be understood that if the recommended video includes the target video, the original classification score corresponding to the target video is required to be higher, so that the target video may be determined as the recommended video, and the requirement that the prediction accuracy of the historical latest video recommendation model is higher can be achieved. From another perspective, if the recommended video does not include the target video, it can be shown that the original classification score of the target video cannot be accurately determined due to the low prediction accuracy of the historical latest video recommendation model, and the reason why the original classification score of the video cannot be accurately determined by the historical latest video recommendation model is that: the historical and latest video recommendation model cannot extract effective features of the video. In order to enable the subsequent historical latest video recommendation model to give correct original classification scores to target videos not belonging to recommended videos, the target videos not belonging to the recommended videos can be added into the historical video set to participate in the process of updating the historical latest video recommendation model, so that the historical latest video recommendation model can further enhance learning of the characteristics of the target videos.
Meanwhile, for a target video belonging to a recommended video, the original classification score of the target video can be accurately determined by a historical latest video recommendation model, in order to enable the latest video recommendation model to further enhance learning of the characteristics of the target video not belonging to the recommended video, different weight values can be given to the target video belonging to the recommended video and the target video not belonging to the recommended video, and for the weight values corresponding to the two, a smaller weight value can be set for the target video belonging to the recommended video, and a larger weight value can be set for the target video not belonging to the recommended video.
Based on the above, it can be understood that the target video belonging to the recommended video represents a video that appears in both the target video and the recommended video, and from the viewpoint of the target video, the target video belonging to the recommended video can be understood as a first video that is the same as the recommended video in the target video, the target video not belonging to the recommended video represents a video that does not appear in the recommended video, and from the viewpoint of the target video, the target video not belonging to the recommended video can be understood as a second video that is different from the recommended video in the target video. And assigning different weight values to the target video belonging to the recommended video and the target video not belonging to the recommended video, namely assigning different weight values to a first video in the target video, which is the same as the recommended video, and a second video in the target video, which is different from the recommended video.
And when the preset time point is reached, adding the weighted first video and the weighted second video into the historical video set to serve as input variables of the historical latest video recommendation model, and participating in the process of updating the historical latest video recommendation model.
The weighted value represents the weight of the loss function corresponding to the first video and the loss function corresponding to the second video in the loss function of the video recommendation model when the loss function of the video recommendation model is calculated in the process of updating the historical latest video recommendation model.
It should be noted that the first video in the target video, which is the same as the recommended video, may be referred to as a positive sample. It can be understood that, for the target video, the classification label corresponding to the positive sample is obtained by labeling the user with the classification label labeling authority.
A second video different from the recommended video in the target video may be called a hard sample, that is, the hard sample refers to a video of the following category: a certain video in the current video set should actually be a recommended video under a certain classification label, but when the video is processed by a historical latest video recommendation model, a lower original classification score is given to the video, so that the video cannot be used as the recommended video under a certain classification label because the original classification score of the video is lower.
Optionally, on the basis of the above technical solution, a second video different from the recommended video in the target video is used as an additional recommended video.
In the embodiment of the invention, since the second video does not belong to the recommended video and is the video meeting the recommendation condition in the current video set, the second video which is different from the recommended video in the target video can be used as the additional recommended video.
Optionally, on the basis of the above technical solution, before inputting the current video set and the historical video set into the historical latest video recommendation model, the method may further include: an original historical video set is obtained. And clustering according to the classification labels of the videos in the original historical video set to obtain a historical video set.
In the embodiment of the invention, in order to balance the number of videos under different classification labels in the historical video set, the classification labels of the videos in the original historical video set can be clustered, and a proper number of videos are selected from each classification label according to a clustering result.
The reason why the above operation is performed is that: if the classification labels of the videos in the original historical video set are not clustered, that is, the original historical video set is used as the historical video set, the number of videos under different classification labels in the historical video set may not be balanced, i.e., the number of videos appearing under one or more category labels is relatively high, while the number of videos appearing under other category labels is relatively low, and, at the same time, since the historical video set will participate in the process of updating the historical up-to-date video recommendation model, if the number of videos under different category labels in the historical video set is not balanced, the updated historical latest video recommendation model cannot accurately give the original classification scores of the videos with the smaller number of videos under the classification labels, and in short, the updated historical latest video recommendation model has low prediction accuracy. By clustering the original historical video set, the number of videos under different classification labels in the historical video set is balanced, and the prediction precision of the updated historical latest video recommendation model is improved.
It should be noted that, it can be understood that the historical video set corresponding to the next update period may include, in addition to various videos generated by determining classification tags of videos in the current video set corresponding to the current update period, that is, videos in which the classification tags of the videos to be recommended are inconsistent with the content of the videos to be recommended (i.e., difficult negative samples), first videos in the target videos that are the same as the recommended videos (i.e., positive samples), and second videos in the target videos that are not the same as the recommended videos (i.e., difficult positive samples), a first video in the target videos that is the same as the recommended videos (i.e., positive samples) and a second video in the target videos that is not the same as the recommended videos (i.e., difficult positive samples) corresponding to T update periods before the current update period. It will also be appreciated that the historical video set is continually updated over the update period.
It should be further noted that the number of the original classification tags and the number of the original classification tags in the historical latest video recommendation model corresponding to the next update period are determined by the historical video set corresponding to the current update period, and further determined by a first video identical to the recommended video in the target video and a second video different from the recommended video in the target video corresponding to the current update period, and a first video identical to the recommended video in the target video and a second video different from the recommended video in the target video corresponding to T update periods before the current update period. It is understood that, from the updating period, the original classification tags and the number of original classification tags of the historical latest video recommendation model are also continuously updated.
In addition, in the process of updating the historical latest video recommendation model corresponding to the current updating period to obtain the historical latest video recommendation model corresponding to the next updating period, different weight values can be set for a video (i.e., a hard negative sample) in which the classification label of the video to be recommended in the current video set corresponding to the current update period is inconsistent with the content of the video to be recommended, a first video (i.e., a positive sample) in the target video in the current video set corresponding to the current update period that is the same as the recommended video, a second video (i.e., a hard positive sample) in the target video in the current video set corresponding to the current update period that is different from the recommended video, a first video (i.e., a positive sample) in the target video corresponding to T update periods before the current update period that is the same as the recommended video, and a second video (i.e., a hard positive sample) in the target video corresponding to T update periods before the current update period that is different from the recommended video. The weight value here also represents the weight of the loss function corresponding to each part in the loss function of the video recommendation model when the loss function of the video recommendation model is calculated in the process of updating the historical latest video recommendation model.
It should be noted that, clustering processing is performed on the original historical video set, and only a first video that is the same as the recommended video in the target videos in the current video set corresponding to the current update period and a second video that is different from the recommended video in the target videos, and a first video that is the same as the recommended video in the target videos and a second video that is different from the recommended video in the target videos corresponding to T update periods before the current update period may be clustered.
In order to better understand the technical solutions provided by the embodiments of the present invention, the following description is made by specific examples, specifically:
predicting that the video about the splash section becomes a recent hotspot video if the current date is adjacent to the splash section, wherein the historical video set corresponding to the current update period comprises various videos generated by determining classification labels of the videos in the current video set corresponding to the last update period, namely, videos with the classification labels of the videos to be recommended inconsistent with the content of the videos to be recommended (namely, difficult negative samples), first videos identical with the recommended videos in the target videos (namely, positive samples) and second videos different from the recommended videos in the target videos (namely, difficult positive samples) corresponding to T update periods before the last update period, wherein the videos about the splash section become the recent hotspot video, therefore, a first video (i.e., a positive sample) in the target video and a second video (i.e., a hard sample) in the target video, which are the same as the recommended video, in the current video set corresponding to the last update period, and a first video (i.e., a positive sample) in the target video and a second video (i.e., a hard sample) in the target video, which are different from the recommended video, in the T update periods before the last update period will appear as videos labeled as splash-water sections by classification.
Inputting a current video set and a historical video set corresponding to a current updating period into a historical latest video recommendation model to obtain an original classification label and an original classification score of a video in the current video set, wherein the original classification score of the video with the original classification label of splash-water section is higher, so that the original classification label of splash-water section is called as a classification label of the video, further, a video to be recommended under the classification label is obtained, if the classification label of the video to be recommended is consistent with the content of the video to be recommended, the video to be recommended is taken as the recommended video under the classification label, namely, the video with the classification label of splash-water section becomes the recommended video.
Based on the above, the splash-water section-related video is considered as a recent hotspot video, the splash-water section-related video can be obtained based on the historical latest video recommendation model, and the splash-water section-related video is used as the recommendation video, so that the mining requirements of most hotspot videos can be met, and hotspots cannot be missed due to tedious model training in the traditional technology.
Fig. 2 is a schematic structural diagram of a video recommendation apparatus according to an embodiment of the present invention, where the present embodiment is applicable to a case of improving prediction accuracy of a video recommendation model, the apparatus may be implemented in a software and/or hardware manner, and the apparatus may be configured in a device, such as a computer or a mobile terminal. As shown in fig. 2, the apparatus specifically includes:
a current video set obtaining module 210, configured to obtain a current video set.
A classification label and classification score obtaining module 220, configured to input the current video set and the historical video set into the historical latest video recommendation model, so as to obtain a classification label and a classification score of a video in the current video set.
And the recommended video determining module 230 is configured to obtain a recommended video under the classification label according to the classification scores of videos under different classification labels in the current video set.
In the technical scheme of the embodiment, the classification labels and the classification scores of the videos in the current video set are obtained by acquiring the current video set and inputting the current video set and the historical video set into the historical latest video recommendation model, obtaining a recommended video under the classification label according to the classification scores of the videos under different classification labels in the current video set, by adopting the historical latest video recommendation model obtained on the basis of the pre-training model to perform video recommendation, since the historical latest video recommendation model has a shorter training period than the video recommendation model generated based on deep learning training in the conventional technology, therefore, the historical video recommendation model can be well adapted to adjustment of actual services, and meanwhile, a historical video set which can be used as prior knowledge of a current video set is also used as an input variable, so that the prediction precision of the historical latest video recommendation model on the current video set is improved.
Optionally, on the basis of the above technical solution, the recommended video determining module 230 may specifically include:
and the video to be recommended acquiring unit is used for acquiring the video to be recommended under the classification label according to the classification scores of the videos under different classification labels in the current video set.
And the recommended video determining unit is used for taking the video to be recommended as the recommended video under the classification label if the classification label of the video to be recommended is consistent with the content of the video to be recommended.
Optionally, on the basis of the above technical solution, the apparatus may further include:
and the first historical latest video recommendation model updating module is used for adding the video to be recommended into the historical video set as an input variable of the historical latest recommendation model after a preset time point is reached and updating the historical latest video recommendation model if the classification label of the video to be recommended is inconsistent with the content of the video to be recommended.
Optionally, on the basis of the above technical solution, the apparatus may further include:
and the target video determining module is used for acquiring the target video which meets the recommendation condition in the current video set and sending the target video.
And the target video classification label marking module is used for receiving the target video marked with the classification label.
And the second historical latest video recommendation model updating module is used for updating the historical latest video recommendation model after the target video with the labels classified is added into the historical video set at the preset time point and is used as an input variable of the historical latest video recommendation model.
Optionally, on the basis of the above technical solution, the second historical latest video recommendation model updating module may specifically include:
and the weight value endowing unit is used for endowing different weight values for a first video which is the same as the recommended video in the target video and a second video which is different from the recommended video in the target video.
And the third history latest video recommendation model updating unit is used for updating the history latest video recommendation model by adding the weighted first video and the weighted second video into the history video set when the preset time point is reached and using the added first video and the weighted second video as input variables of the history latest video recommendation model.
Optionally, on the basis of the above technical solution, the apparatus may further include:
and the additional recommended video determining module is used for taking a second video which is different from the recommended video in the target video as an additional recommended video.
Optionally, on the basis of the above technical solution, the apparatus may further include:
and the original video set acquisition module is used for acquiring an original historical video set.
And the historical video set acquisition module is used for carrying out clustering processing according to the classification labels of the videos in the original historical video set to obtain a historical video set.
The video recommendation device provided by the embodiment of the invention can execute the video recommendation method provided by any embodiment of the invention, and has corresponding functional modules and beneficial effects of the execution method.
Fig. 3 is a schematic structural diagram of an apparatus according to an embodiment of the present invention. Fig. 3 illustrates a block diagram of an exemplary device 312 suitable for use in implementing embodiments of the present invention. The device 312 shown in fig. 3 is only an example and should not bring any limitations to the function and scope of use of the embodiments of the present invention.
As shown in FIG. 3, device 312 is in the form of a general purpose computing device. The components of device 312 may include, but are not limited to: one or more processors 316, a system memory 328, and a bus 318 that couples the various system components including the system memory 328 and the processors 316.
Bus 318 represents one or more of any of several types of bus structures, including a memory bus or memory controller, a peripheral bus, an accelerated graphics port, and a processor or local bus using any of a variety of bus architectures. By way of example, such architectures include, but are not limited to, an Industry Standard Architecture (ISA) bus, a Micro Channel Architecture (MCA) bus, an enhanced ISA (ISA) bus, a Video Electronics Standards Association (VESA) local bus, and a Peripheral Component Interconnect (PCI) bus.
Device 312 typically includes a variety of computer system readable media. Such media may be any available media that is accessible by device 312 and includes both volatile and nonvolatile media, removable and non-removable media.
The system Memory 328 may include computer system readable media in the form of volatile Memory, such as Random Access Memory (RAM) 330 and/or cache Memory 332. The device 312 may further include other removable/non-removable, volatile/nonvolatile computer system storage media. By way of example only, storage system 334 may be used to read from and write to non-removable, nonvolatile magnetic media (not shown in FIG. 3, and commonly referred to as a "hard drive"). Although not shown in FIG. 3, a magnetic disk drive for reading from and writing to a removable, nonvolatile magnetic disk (e.g., a "floppy disk") and an optical disk drive for reading from or writing to a removable, nonvolatile optical disk (e.g., a Computer disk Read-Only Memory, CD-ROM), Digital Video disk (DVD-ROM), or other optical media) may be provided. In these cases, each drive may be connected to bus 318 by one or more data media interfaces. Memory 328 may include at least one program product having a set (e.g., at least one) of program modules that are configured to carry out the functions of embodiments of the invention.
A program/utility 340 having a set (at least one) of program modules 342 may be stored, for example, in memory 328, such program modules 342 including, but not limited to, an operating system, one or more application programs, other program modules, and program data, each of which examples or some combination thereof may comprise an implementation of a network environment. Program modules 342 generally perform the functions and/or methodologies of the described embodiments of the invention.
Device 312 may also communicate with one or more external devices 314 (e.g., keyboard, pointing device, display 324, etc.), with one or more devices that enable a user to interact with device 312, and/or with any devices (e.g., network card, modem, etc.) that enable device 312 to communicate with one or more other computing devices. Such communication may occur via input/output (I/O) interfaces 322. Further, the device 312 may also communicate with one or more networks (e.g., a Local Area Network (LAN), a Wide Area Network (WAN), and/or a public Network, such as the internet) via the Network adapter 320. As shown, a network adapter 320 communicates with the other modules of the device 312 via a bus 318. It should be appreciated that although not shown in FIG. 3, other hardware and/or software modules may be used in conjunction with device 312, including but not limited to: microcode, device drivers, Redundant processing units, external disk drive Arrays, disk array (RAID) systems, tape drives, and data backup storage systems, to name a few.
The processor 316 executes various functional applications and data processing by executing programs stored in the system memory 328, for example, implementing a video recommendation method provided by an embodiment of the present invention, the method including:
and acquiring a current video set.
And inputting the current video set and the historical video set into a historical latest video recommendation model to obtain the classification labels and the classification scores of the videos in the current video set.
And obtaining the recommended video under the classification label according to the classification scores of the videos under different classification labels in the current video set.
Of course, those skilled in the art will understand that the processor may also implement the technical solution of the video recommendation method applied to the device provided by any embodiment of the present invention. The hardware structure and the function of the device can be explained with reference to the contents of the embodiment.
An embodiment of the present invention further provides a computer-readable storage medium, on which a computer program is stored, where the computer program, when executed by a processor, implements a video recommendation method according to an embodiment of the present invention, where the method includes:
and acquiring a current video set.
And inputting the current video set and the historical video set into a historical latest video recommendation model to obtain the classification labels and the classification scores of the videos in the current video set.
And obtaining the recommended video under the classification label according to the classification scores of the videos under different classification labels in the current video set.
Computer storage media for embodiments of the invention may employ any combination of one or more computer-readable media. The computer readable medium may be a computer readable signal medium or a computer readable storage medium. A computer readable storage medium may be, for example, but not limited to, an electronic, magnetic, optical, electromagnetic, infrared, or semiconductor system, apparatus, or device, or any combination of the foregoing. More specific examples (a non-exhaustive list) of the computer readable storage medium would include the following: an electrical connection having one or more wires, a portable Computer diskette, a hard disk, a Random Access Memory (RAM), a Read-Only Memory (ROM), an erasable programmable Read-Only Memory (EPROM), a flash Memory, an optical fiber, a portable compact Disc Read-Only Memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination of the foregoing. In the context of this document, a computer readable storage medium may be any tangible medium that can contain, or store a program for use by or in connection with an instruction execution system, apparatus, or device.
A computer readable signal medium may include a propagated data signal with computer readable program code embodied therein, for example, in baseband or as part of a carrier wave. Such a propagated data signal may take many forms, including, but not limited to, electro-magnetic, optical, or any suitable combination thereof. A computer readable signal medium may also be any computer readable medium that is not a computer readable storage medium and that can communicate, propagate, or transport a program for use by or in connection with an instruction execution system, apparatus, or device.
Program code embodied on a computer readable medium may be transmitted using any appropriate medium, including but not limited to wireless, wireline, optical fiber cable, RF, etc., or any suitable combination of the foregoing.
Computer program code for carrying out operations for aspects of the present invention may be written in any combination of one or more programming languages, including an object oriented programming language such as Java, Smalltalk, C + + or the like and conventional procedural programming languages, such as the "C" programming language or similar programming languages. The program code may execute entirely on the user's computer, partly on the user's computer, as a stand-alone software package, partly on the user's computer and partly on a remote computer or entirely on the remote computer or server. In the case of a remote computer, the remote computer may be connected to the user's computer through any type of Network, Local Area Network (LAN) or Wide Area Network (WAN), or the connection may be made to an external computer (for example, through the Internet using an Internet service provider).
Of course, the computer-readable storage medium provided by the embodiments of the present invention has computer-executable instructions that are not limited to the method operations described above, and may also perform related operations in the video recommendation method of the device provided by any embodiment of the present invention. The description of the storage medium is explained with reference to the embodiments.
It is to be noted that the foregoing is only illustrative of the preferred embodiments of the present invention and the technical principles employed. It will be understood by those skilled in the art that the present invention is not limited to the particular embodiments described herein, but is capable of various obvious changes, rearrangements and substitutions as will now become apparent to those skilled in the art without departing from the scope of the invention. Therefore, although the present invention has been described in greater detail by the above embodiments, the present invention is not limited to the above embodiments, and may include other equivalent embodiments without departing from the spirit of the present invention, and the scope of the present invention is determined by the scope of the appended claims.

Claims (10)

1. A method for video recommendation, comprising:
acquiring a current video set;
inputting the current video set and the historical video set into a historical latest video recommendation model to obtain the classification labels and classification scores of the videos in the current video set;
and obtaining the recommended video under the classification label according to the classification scores of the videos under different classification labels in the current video set.
2. The method of claim 1, wherein obtaining the recommended video under the classification label according to the classification scores of the videos under different classification labels in the current video set comprises:
obtaining a video to be recommended under a classification label according to the classification scores of videos under different classification labels in the current video set;
and if the classification label of the video to be recommended is consistent with the content of the video to be recommended, taking the video to be recommended as the recommended video under the classification label.
3. The method of claim 2, further comprising:
and if the classification label of the video to be recommended is inconsistent with the content of the video to be recommended, adding the video to be recommended into the historical video set after a preset time point is reached, and then taking the video to be recommended as an input variable of the historical latest video recommendation model to update the historical latest video recommendation model.
4. The method of claim 1, further comprising:
acquiring a target video which meets a recommendation condition in the current video set, and sending the target video;
receiving a target video marked with a classification label;
and when a preset time point is reached, adding the target video labeled with the classification label into the historical video set to serve as an input variable of the historical latest video recommendation model, and updating the historical latest video recommendation model.
5. The method of claim 4, wherein the adding the target video labeled with the classification label to the historical video set as an input variable of the historical latest video recommendation model, and updating the historical latest video recommendation model comprises:
assigning different weight values to a first video in the target video, which is the same as the recommended video, and a second video in the target video, which is different from the recommended video;
and when a preset time point is reached, adding the weighted first video and the weighted second video into the historical video set to serve as input variables of the historical latest video recommendation model, and updating the historical latest video recommendation model.
6. The method of claim 5, further comprising:
and taking a second video which is different from the recommended video in the target video as an additional recommended video.
7. The method of claim 1, wherein before entering the current video set and the historical video set into a historical up-to-date video recommendation model, further comprising:
acquiring an original historical video set;
and clustering according to the classification labels of the videos in the original historical video set to obtain a historical video set.
8. A video recommendation apparatus, comprising:
the current video set acquisition module is used for acquiring a current video set;
the classification label and classification score acquisition module is used for inputting the current video set and the historical video set into a historical latest video recommendation model to obtain a classification label and a classification score of the video in the current video set;
and the recommended video determining module is used for obtaining the recommended video under the classification label according to the classification scores of the videos under different classification labels in the current video set.
9. An apparatus, comprising:
one or more processors;
a memory for storing one or more programs;
when executed by the one or more processors, cause the one or more processors to implement the method of any one of claims 1-7.
10. A computer-readable storage medium, on which a computer program is stored which, when being executed by a processor, carries out the method according to any one of claims 1 to 7.
CN201811640356.6A 2018-12-29 2018-12-29 Video recommendation method, device, equipment and storage medium Active CN111385659B (en)

Priority Applications (2)

Application Number Priority Date Filing Date Title
CN201811640356.6A CN111385659B (en) 2018-12-29 2018-12-29 Video recommendation method, device, equipment and storage medium
PCT/CN2019/124582 WO2020135054A1 (en) 2018-12-29 2019-12-11 Method, device and apparatus for video recommendation and storage medium

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN201811640356.6A CN111385659B (en) 2018-12-29 2018-12-29 Video recommendation method, device, equipment and storage medium

Publications (2)

Publication Number Publication Date
CN111385659A true CN111385659A (en) 2020-07-07
CN111385659B CN111385659B (en) 2021-08-17

Family

ID=71128373

Family Applications (1)

Application Number Title Priority Date Filing Date
CN201811640356.6A Active CN111385659B (en) 2018-12-29 2018-12-29 Video recommendation method, device, equipment and storage medium

Country Status (2)

Country Link
CN (1) CN111385659B (en)
WO (1) WO2020135054A1 (en)

Cited By (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN112016613A (en) * 2020-08-26 2020-12-01 广州市百果园信息技术有限公司 Training method and device for video content classification model, computer equipment and medium
CN112364202A (en) * 2020-11-06 2021-02-12 上海众源网络有限公司 Video recommendation method and device and electronic equipment
CN114401310A (en) * 2021-12-31 2022-04-26 苏州市瑞川尔自动化设备有限公司 Visual cloud service data optimization method and server
CN114401310B (en) * 2021-12-31 2024-05-31 苏州市瑞川尔自动化设备有限公司 Visual cloud service data optimization method and server

Citations (21)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN101112092A (en) * 2005-05-09 2008-01-23 皇家飞利浦电子股份有限公司 Auxiliary user control in recommending device
CN101834837A (en) * 2009-12-18 2010-09-15 北京邮电大学 On-line landscape video active information service system of scenic spots in tourist attraction based on bandwidth network
CN102651033A (en) * 2012-04-09 2012-08-29 百度在线网络技术（北京）有限公司 Method and device for recommending online resource
CN103838835A (en) * 2014-02-25 2014-06-04 中国科学院自动化研究所 Network sensitive video detection method
US20140215506A1 (en) * 2013-01-25 2014-07-31 Mobitv, Inc. Time context weighted content recommendation
CN105512331A (en) * 2015-12-28 2016-04-20 海信集团有限公司 Video recommending method and device
CN106294783A (en) * 2016-08-12 2017-01-04 乐视控股（北京）有限公司 A kind of video recommendation method and device
CN106407477A (en) * 2016-11-22 2017-02-15 深圳市互联在线云计算股份有限公司 Multidimensional interconnection recommendation method and system
CN106855876A (en) * 2015-12-08 2017-06-16 奥多比公司 The attribute weight of the recommendation based on media content
CN107194419A (en) * 2017-05-10 2017-09-22 百度在线网络技术（北京）有限公司 Video classification methods and device, computer equipment and computer-readable recording medium
CN107451148A (en) * 2016-05-31 2017-12-08 北京金山安全软件有限公司 Video classification method and device and electronic equipment
CN107515909A (en) * 2017-08-11 2017-12-26 深圳市耐飞科技有限公司 A kind of video recommendation method and system
CN107562875A (en) * 2017-08-31 2018-01-09 北京麒麟合盛网络技术有限公司 A kind of update method of model, apparatus and system
CN107563455A (en) * 2017-10-18 2018-01-09 百度在线网络技术（北京）有限公司 For obtaining the method and device of information
CN107743249A (en) * 2017-11-27 2018-02-27 四川长虹电器股份有限公司 A kind of CTR predictor methods based on Model Fusion
CN107846629A (en) * 2017-10-11 2018-03-27 五八有限公司 Recommend the method, apparatus and server of video to user
US9934515B1 (en) * 2013-12-20 2018-04-03 Google Llc Content recommendation system using a neural network language model
CN107967616A (en) * 2016-10-19 2018-04-27 三星Sds株式会社 Content recommendation method, apparatus and system
JP2018116515A (en) * 2017-01-19 2018-07-26 株式会社日立製作所 Control apparatus
CN108647293A (en) * 2018-05-07 2018-10-12 广州虎牙信息科技有限公司 Video recommendation method, device, storage medium and server
CN109104620A (en) * 2018-07-26 2018-12-28 腾讯科技（深圳）有限公司 A kind of short video recommendation method, device and readable medium

Family Cites Families (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
KR101169170B1 (en) * 2010-11-09 2012-07-30 경희대학교 산학협력단 Method for recommending content based on user preference with time flow
CN104065981A (en) * 2014-06-20 2014-09-24 海信集团有限公司 Method and device for recommending videos
CN108243357A (en) * 2018-01-25 2018-07-03 北京搜狐新媒体信息技术有限公司 A kind of video recommendation method and device
CN108228911A (en) * 2018-02-11 2018-06-29 北京搜狐新媒体信息技术有限公司 The computational methods and device of a kind of similar video

Patent Citations (21)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN101112092A (en) * 2005-05-09 2008-01-23 皇家飞利浦电子股份有限公司 Auxiliary user control in recommending device
CN101834837A (en) * 2009-12-18 2010-09-15 北京邮电大学 On-line landscape video active information service system of scenic spots in tourist attraction based on bandwidth network
CN102651033A (en) * 2012-04-09 2012-08-29 百度在线网络技术（北京）有限公司 Method and device for recommending online resource
US20140215506A1 (en) * 2013-01-25 2014-07-31 Mobitv, Inc. Time context weighted content recommendation
US9934515B1 (en) * 2013-12-20 2018-04-03 Google Llc Content recommendation system using a neural network language model
CN103838835A (en) * 2014-02-25 2014-06-04 中国科学院自动化研究所 Network sensitive video detection method
CN106855876A (en) * 2015-12-08 2017-06-16 奥多比公司 The attribute weight of the recommendation based on media content
CN105512331A (en) * 2015-12-28 2016-04-20 海信集团有限公司 Video recommending method and device
CN107451148A (en) * 2016-05-31 2017-12-08 北京金山安全软件有限公司 Video classification method and device and electronic equipment
CN106294783A (en) * 2016-08-12 2017-01-04 乐视控股（北京）有限公司 A kind of video recommendation method and device
CN107967616A (en) * 2016-10-19 2018-04-27 三星Sds株式会社 Content recommendation method, apparatus and system
CN106407477A (en) * 2016-11-22 2017-02-15 深圳市互联在线云计算股份有限公司 Multidimensional interconnection recommendation method and system
JP2018116515A (en) * 2017-01-19 2018-07-26 株式会社日立製作所 Control apparatus
CN107194419A (en) * 2017-05-10 2017-09-22 百度在线网络技术（北京）有限公司 Video classification methods and device, computer equipment and computer-readable recording medium
CN107515909A (en) * 2017-08-11 2017-12-26 深圳市耐飞科技有限公司 A kind of video recommendation method and system
CN107562875A (en) * 2017-08-31 2018-01-09 北京麒麟合盛网络技术有限公司 A kind of update method of model, apparatus and system
CN107846629A (en) * 2017-10-11 2018-03-27 五八有限公司 Recommend the method, apparatus and server of video to user
CN107563455A (en) * 2017-10-18 2018-01-09 百度在线网络技术（北京）有限公司 For obtaining the method and device of information
CN107743249A (en) * 2017-11-27 2018-02-27 四川长虹电器股份有限公司 A kind of CTR predictor methods based on Model Fusion
CN108647293A (en) * 2018-05-07 2018-10-12 广州虎牙信息科技有限公司 Video recommendation method, device, storage medium and server
CN109104620A (en) * 2018-07-26 2018-12-28 腾讯科技（深圳）有限公司 A kind of short video recommendation method, device and readable medium

Cited By (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN112016613A (en) * 2020-08-26 2020-12-01 广州市百果园信息技术有限公司 Training method and device for video content classification model, computer equipment and medium
CN112364202A (en) * 2020-11-06 2021-02-12 上海众源网络有限公司 Video recommendation method and device and electronic equipment
CN112364202B (en) * 2020-11-06 2023-11-14 上海众源网络有限公司 Video recommendation method and device and electronic equipment
CN114401310A (en) * 2021-12-31 2022-04-26 苏州市瑞川尔自动化设备有限公司 Visual cloud service data optimization method and server
CN114401310B (en) * 2021-12-31 2024-05-31 苏州市瑞川尔自动化设备有限公司 Visual cloud service data optimization method and server

Also Published As

Publication number Publication date
CN111385659B (en) 2021-08-17
WO2020135054A1 (en) 2020-07-02

Similar Documents

Publication Publication Date Title
US11645554B2 (en) Method and apparatus for recognizing a low-quality article based on artificial intelligence, device and medium
CN112560912B (en) Classification model training method and device, electronic equipment and storage medium
WO2020125445A1 (en) Classification model training method, classification method, device and medium
CN108319723B (en) Picture sharing method and device, terminal and storage medium
CN110751224B (en) Training method of video classification model, video classification method, device and equipment
CN111639710A (en) Image recognition model training method, device, equipment and storage medium
US10671895B2 (en) Automated selection of subjectively best image frames from burst captured image sequences
US11182447B2 (en) Customized display of emotionally filtered social media content
US20160063376A1 (en) Obtaining user traits
CN112559800B (en) Method, apparatus, electronic device, medium and product for processing video
CN110598157A (en) Target information identification method, device, equipment and storage medium
CN111738041A (en) Video segmentation method, device, equipment and medium
CN111708876A (en) Method and device for generating information
CN111597446B (en) Content pushing method and device based on artificial intelligence, server and storage medium
CN110543637A (en) Chinese word segmentation method and device
CN112507090A (en) Method, apparatus, device and storage medium for outputting information
CN111385659B (en) Video recommendation method, device, equipment and storage medium
CN112995690A (en) Live content item identification method and device, electronic equipment and readable storage medium
CN107944026A (en) A kind of method, apparatus, server and the storage medium of atlas personalized recommendation
CN109446324B (en) Sample data processing method and device, storage medium and electronic equipment
CN110826327A (en) Emotion analysis method and device, computer readable medium and electronic equipment
CN113010785A (en) User recommendation method and device
CN113239883A (en) Method and device for training classification model, electronic equipment and storage medium
US10540541B2 (en) Cognitive image detection and recognition
CN115017385A (en) Article searching method, device, equipment and storage medium

Legal Events

Date Code Title Description
PB01 Publication
PB01 Publication
SE01 Entry into force of request for substantive examination
SE01 Entry into force of request for substantive examination
GR01 Patent grant
GR01 Patent grant
TR01 Transfer of patent right
TR01 Transfer of patent right

Effective date of registration: 20220609

Address after: 31a, 15 / F, building 30, maple mall, bangrang Road, Brazil, Singapore

Patentee after: Baiguoyuan Technology (Singapore) Co.,Ltd.

Address before: 511400 floor 23-39, building B-1, Wanda Plaza North, Wanbo business district, 79 Wanbo 2nd Road, Nancun Town, Panyu District, Guangzhou City, Guangdong Province

Patentee before: GUANGZHOU BAIGUOYUAN INFORMATION TECHNOLOGY Co.,Ltd.