CN113676770B - Member rights prediction method, member rights prediction device, electronic equipment and storage medium - Google Patents
Member rights prediction method, member rights prediction device, electronic equipment and storage medium Download PDFInfo
- Publication number
- CN113676770B CN113676770B CN202110790348.5A CN202110790348A CN113676770B CN 113676770 B CN113676770 B CN 113676770B CN 202110790348 A CN202110790348 A CN 202110790348A CN 113676770 B CN113676770 B CN 113676770B
- Authority
- CN
- China
- Prior art keywords
- predicted
- video
- diversity
- sample
- membership
- 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.)
- Active
Links
- 238000000034 method Methods 0.000 title claims abstract description 68
- 238000006243 chemical reaction Methods 0.000 claims description 56
- 238000012549 training Methods 0.000 claims description 34
- 230000003068 static effect Effects 0.000 claims description 21
- 238000004891 communication Methods 0.000 claims description 19
- 230000008569 process Effects 0.000 claims description 19
- 238000012545 processing Methods 0.000 claims description 17
- 230000014759 maintenance of location Effects 0.000 claims description 10
- 230000009466 transformation Effects 0.000 claims description 10
- 238000004590 computer program Methods 0.000 claims description 8
- 238000004364 calculation method Methods 0.000 claims description 7
- 230000009191 jumping Effects 0.000 claims description 6
- 238000010238 partial least squares regression Methods 0.000 description 15
- 238000012417 linear regression Methods 0.000 description 7
- 230000002776 aggregation Effects 0.000 description 6
- 238000004220 aggregation Methods 0.000 description 6
- 230000001419 dependent effect Effects 0.000 description 5
- 238000010586 diagram Methods 0.000 description 5
- 230000009471 action Effects 0.000 description 3
- 230000000694 effects Effects 0.000 description 3
- 238000005516 engineering process Methods 0.000 description 2
- 230000006870 function Effects 0.000 description 2
- 230000006872 improvement Effects 0.000 description 2
- 238000011551 log transformation method Methods 0.000 description 2
- 238000007477 logistic regression Methods 0.000 description 2
- 230000007246 mechanism Effects 0.000 description 2
- 238000000513 principal component analysis Methods 0.000 description 2
- 238000000611 regression analysis Methods 0.000 description 2
- 238000012706 support-vector machine Methods 0.000 description 2
- 238000004422 calculation algorithm Methods 0.000 description 1
- 239000012141 concentrate Substances 0.000 description 1
- 238000013500 data storage Methods 0.000 description 1
- 238000011161 development Methods 0.000 description 1
- 238000002474 experimental method Methods 0.000 description 1
- 239000000835 fiber Substances 0.000 description 1
- 238000003064 k means clustering Methods 0.000 description 1
- 238000010801 machine learning Methods 0.000 description 1
- 239000000463 material Substances 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000003287 optical effect Effects 0.000 description 1
- 230000002093 peripheral effect Effects 0.000 description 1
- 230000002035 prolonged effect Effects 0.000 description 1
- 230000033764 rhythmic process Effects 0.000 description 1
- 239000004065 semiconductor Substances 0.000 description 1
- 239000007787 solid Substances 0.000 description 1
- 238000007619 statistical method Methods 0.000 description 1
- 230000001502 supplementing effect Effects 0.000 description 1
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
- H04N21/43—Processing of content or additional data, e.g. demultiplexing additional data from a digital video stream; Elementary client operations, e.g. monitoring of home network or synchronising decoder's clock; Client middleware
- H04N21/44—Processing of video elementary streams, e.g. splicing a video clip retrieved from local storage with an incoming video stream or rendering scenes according to encoded video stream scene graphs
- H04N21/44008—Processing of video elementary streams, e.g. splicing a video clip retrieved from local storage with an incoming video stream or rendering scenes according to encoded video stream scene graphs involving operations for analysing video streams, e.g. detecting features or characteristics in the video stream
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/21—Design or setup of recognition systems or techniques; Extraction of features in feature space; Blind source separation
- G06F18/214—Generating training patterns; Bootstrap methods, e.g. bagging or boosting
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/24—Classification techniques
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06Q10/00—Administration; Management
- G06Q10/04—Forecasting or optimisation specially adapted for administrative or management purposes, e.g. linear programming or "cutting stock problem"
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06Q30/00—Commerce
- G06Q30/02—Marketing; Price estimation or determination; Fundraising
- G06Q30/0207—Discounts or incentives, e.g. coupons or rebates
-
- 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
- H04N21/462—Content or additional data management, e.g. creating a master electronic program guide from data received from the Internet and a Head-end, controlling the complexity of a video stream by scaling the resolution or bit-rate based on the client capabilities
- H04N21/4627—Rights management associated to the content
-
- 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
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Business, Economics & Management (AREA)
- Databases & Information Systems (AREA)
- Data Mining & Analysis (AREA)
- Strategic Management (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Signal Processing (AREA)
- Multimedia (AREA)
- Economics (AREA)
- Development Economics (AREA)
- Game Theory and Decision Science (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Evolutionary Computation (AREA)
- Evolutionary Biology (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Human Resources & Organizations (AREA)
- Bioinformatics & Computational Biology (AREA)
- Entrepreneurship & Innovation (AREA)
- Marketing (AREA)
- Finance (AREA)
- Accounting & Taxation (AREA)
- Life Sciences & Earth Sciences (AREA)
- General Business, Economics & Management (AREA)
- General Engineering & Computer Science (AREA)
- Artificial Intelligence (AREA)
- Tourism & Hospitality (AREA)
- Quality & Reliability (AREA)
- Operations Research (AREA)
- Computer Security & Cryptography (AREA)
- Information Retrieval, Db Structures And Fs Structures Therefor (AREA)
Abstract
The embodiment of the invention provides a member rights and interests prediction method and device, electronic equipment and a storage medium. The member interest prediction method comprises the following steps: obtaining a pre-trained prediction model; acquiring characteristic data to be predicted of the video diversity to be predicted aiming at each video diversity to be predicted in a video series set to be predicted, and taking the characteristic data to be predicted as the input of the prediction model to obtain member rights and interests information of the video diversity to be predicted, which is output by the prediction model; and determining the membership right diversity in the video drama set to be predicted based on the membership right information of the video diversity to be predicted. In the embodiment of the invention, the diversity of the videos to be predicted in the video episode can be analyzed in a more targeted manner aiming at different video episodes, so that the more appropriate and more accurate membership interest diversity of the video episode can be determined.
Description
Technical Field
The present invention relates to the field of internet technologies, and in particular, to a member rights prediction method, an apparatus, an electronic device, and a storage medium.
Background
With the rapid development of internet technology, users increasingly rely on obtaining information through networks. In order to meet the requirement of users for watching videos, various video websites come along.
Users of video websites can be divided into affiliate users and non-affiliate users. Video episodes provided in the video website enter a member interest period after a hot broadcast period, and the video episodes in the member interest period are generally divided into a non-member interest set and a member interest set.
At present, for video episodes in a membership interest period, the same membership interest mechanism is usually adopted, and the fixed first few sets are set as non-membership interest sets and the later sets are set as membership interest sets. However, this arrangement is not accurate. For example, after non-member equity aggregation of some video episodes is finished, the user may not be immersed in the episodes yet, and the user starts to turn to the member equity aggregation at the moment and is easy to dissuade the user; for another example, after non-member equity aggregation of some video episodes, the user may have predicted a subsequent scenario, resulting in no open desire for the member to continue watching.
Therefore, the setting of the member interest set by the member interest mechanism in the prior art is inaccurate, and the member interest set suitable for the video episode can not be obtained.
Disclosure of Invention
Embodiments of the present invention provide a member interest prediction method, an apparatus, an electronic device, and a storage medium, so as to accurately predict member interest information of each video diversity, and further accurately determine member interest diversity in a video drama set based on the member interest information of the video diversity. The specific technical scheme is as follows:
in a first aspect of the present invention, there is provided a membership rights prediction method, including:
obtaining a pre-trained prediction model; the prediction model is obtained by training based on a plurality of sample data, wherein the sample data comprises sample characteristic data of sample video diversity and actual membership rights and interests information of the sample video diversity;
acquiring characteristic data to be predicted of the video diversity to be predicted aiming at each video diversity to be predicted in a video series set to be predicted, and taking the characteristic data to be predicted as the input of the prediction model to obtain member rights and interests information of the video diversity to be predicted, which is output by the prediction model;
and determining the membership right diversity in the video drama set to be predicted based on the membership right information of the video diversity to be predicted.
Optionally, the feature data to be predicted includes a static feature to be predicted, which is not related to the membership rights and interests, and a dynamic feature to be predicted, which is related to the membership rights and interests; the static feature to be predicted comprises at least one of the following: the video episode to be predicted comprises the independent broadcasting information of the video episode to be predicted, the search index of the video episode to be predicted and the score of the video episode to be predicted, wherein the independent broadcasting information is used for indicating whether the video episode to be predicted is independently broadcast or not; the dynamic features to be predicted comprise at least one of the following: the number of the non-member right and interest diversity in the video episode to be predicted, the number of the member right and interest diversity in the video episode to be predicted, the retention rate of the video diversity to be predicted, the average playing completion degree of the video diversity to be predicted, the playing proportion of the video diversity to be predicted exceeding the preset playing completion degree, and the jumping rate of the video diversity to be predicted within the preset time length.
Optionally, the membership right information is used to indicate a membership conversion rate when the video diversity to be predicted is used as a last set of non-membership right diversity; the determining the membership right diversity in the video drama set to be predicted based on the membership right information of the video diversity to be predicted comprises the following steps: and selecting the video diversity to be predicted of the next set of the video diversity to be predicted with the maximum membership conversion rate, and taking the selected video diversity to be predicted as the first set membership right diversity in the video drama set to be predicted.
Optionally, after obtaining the to-be-predicted feature data of the to-be-predicted video diversity, the method further includes: converting the characteristic data to be predicted to obtain newly added characteristic data to be predicted; the conversion processing comprises four arithmetic operations and/or logarithmic transformation; the using the feature data to be predicted as the input of the prediction model comprises: and taking the feature data to be predicted and the newly added feature data to be predicted as the input of the prediction model.
Optionally, after performing conversion processing on the feature data to be predicted to obtain new feature data to be predicted, the method further includes: selecting target feature data to be predicted corresponding to the target feature type from the feature data to be predicted and the newly added feature data to be predicted; the target feature type is obtained by calculation in the training process of the prediction model, and the correlation coefficient with the output of the prediction model is larger than a preset correlation threshold; the using the feature data to be predicted as the input of the prediction model comprises: and taking the target characteristic data to be predicted as the input of the prediction model.
Optionally, the predictive model is trained by: acquiring the sample data; training a preset model to be trained by using the sample data; the input of the model to be trained is the sample characteristic data, and the output is the predicted membership rights and interests information of the sample video diversity; and after training is judged to be completed based on the actual member rights and interests information and the predicted member rights and interests information, taking a trained model as the predicted model.
Optionally, after acquiring the sample data, the method further includes: converting the sample characteristic data to obtain newly added sample characteristic data; calculating a correlation coefficient between the characteristic type and the output of the prediction model aiming at each characteristic type in the sample characteristic data and the newly added sample characteristic data, and selecting the characteristic type with the correlation coefficient larger than a preset correlation threshold value as a target characteristic type; selecting target sample characteristic data corresponding to the target characteristic type from the sample characteristic data and the newly added sample characteristic data; and the input of the model to be trained is the target sample characteristic data.
Optionally, the predictive model is a partial least squares regression, PLS, model.
In a second aspect of the present invention, there is also provided a membership rights prediction apparatus including:
the first acquisition module is used for acquiring a pre-trained prediction model; the prediction model is obtained by training based on a plurality of sample data, wherein the sample data comprises sample characteristic data of sample video diversity and actual membership rights and interests information of the sample video diversity;
the second acquisition module is used for acquiring the characteristic data to be predicted of the video diversity to be predicted aiming at each video diversity to be predicted in the video diversity to be predicted;
the prediction module is used for taking the characteristic data to be predicted as the input of the prediction model to obtain the membership rights and interests information of the video diversity to be predicted, which is output by the prediction model;
and the determining module is used for determining the membership right diversity in the video drama set to be predicted based on the membership right information of the video diversity to be predicted.
Optionally, the feature data to be predicted includes a static feature to be predicted, which is not related to the membership rights and interests, and a dynamic feature to be predicted, which is related to the membership rights and interests; the static feature to be predicted comprises at least one of the following: the video episode to be predicted comprises the independent broadcasting information of the video episode to be predicted, the search index of the video episode to be predicted and the score of the video episode to be predicted, wherein the independent broadcasting information is used for indicating whether the video episode to be predicted is independently broadcast or not; the dynamic features to be predicted comprise at least one of the following: the number of the non-member interest diversity in the video episode to be predicted, the number of the member interest diversity in the video episode to be predicted, the retention rate of the video diversity to be predicted, the average playing completion degree of the video diversity to be predicted, the playing proportion of the video diversity to be predicted exceeding the preset playing completion degree, and the jumping rate of the video diversity to be predicted within the preset time length.
Optionally, the membership right information is used to indicate a membership conversion rate when the video diversity to be predicted is used as a last set of non-membership right diversity; the determining module is specifically configured to select a next set of to-be-predicted video diversity of the to-be-predicted video diversity with the largest member conversion rate, and use the selected to-be-predicted video diversity as a first set of member right diversity in the to-be-predicted video drama set.
Optionally, the apparatus further comprises: the first conversion module is used for carrying out conversion processing on the feature data to be predicted to obtain newly added feature data to be predicted; the conversion processing comprises four arithmetic operations and/or logarithmic transformation; the prediction module is specifically configured to use the feature data to be predicted and the newly added feature data to be predicted as inputs of the prediction model.
Optionally, the apparatus further comprises: the first selection module is used for selecting target feature data to be predicted corresponding to a target feature type from the feature data to be predicted and the newly added feature data to be predicted; the target feature type is obtained by calculation in the training process of the prediction model, and the correlation coefficient with the output of the prediction model is larger than a preset correlation threshold; the prediction module is specifically configured to use the target feature data to be predicted as an input of the prediction model.
Optionally, the prediction model is trained by: a third obtaining module, configured to obtain the sample data; the training module is used for training a preset model to be trained by utilizing the sample data; the input of the model to be trained is the sample characteristic data, and the output is the predicted membership rights and interests information of the sample video diversity; and the judging module is used for judging that training is finished based on the actual member rights and interests information and the predicted member rights and interests information, and then taking the trained model as the predicted model.
Optionally, the apparatus further comprises: the second conversion module is used for carrying out conversion processing on the sample characteristic data to obtain newly added sample characteristic data; the calculation module is used for calculating a correlation coefficient between the characteristic type and the output of the prediction model aiming at each characteristic type in the sample characteristic data and the newly added sample characteristic data, and selecting the characteristic type of which the correlation coefficient is greater than a preset correlation threshold value as a target characteristic type; the second selection module is used for selecting target sample characteristic data corresponding to the target characteristic type from the sample characteristic data and the newly added sample characteristic data; and the input of the model to be trained is the target sample characteristic data.
Optionally, the predictive model is a partial least squares regression, PLS, model.
In another aspect of the present invention, there is also provided an electronic device, including a processor, a communication interface, a memory and a communication bus, where the processor, the communication interface, and the memory complete communication with each other through the communication bus; a memory for storing a computer program; and a processor for implementing any of the above described methods of membership rights prediction when executing the program stored in the memory.
In yet another aspect of the present invention, there is also provided a computer-readable storage medium having stored therein instructions, which when run on a computer, cause the computer to implement any of the above described membership rights prediction methods.
In yet another aspect of the present invention, there is also provided a computer program product containing instructions which, when run on a computer, cause the computer to implement any of the above described membership rights prediction methods.
The member interest prediction method, the member interest prediction device, the electronic equipment and the storage medium provided by the embodiment of the invention are used for accurately predicting the member interest information of the video diversity to be predicted by utilizing a pre-trained prediction model according to the characteristic data to be predicted of the video diversity to be predicted aiming at each video diversity to be predicted in the video diversity to be predicted, and determining the member interest diversity in the video diversity to be predicted according to the member interest information of each video diversity to be predicted. Therefore, in the embodiment of the invention, the diversity of the videos to be predicted in the video episode can be analyzed in a more targeted manner aiming at different video episodes, so that the more appropriate and more accurate membership interest diversity of the video episode can be determined.
Drawings
In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings used in the description of the embodiments or the prior art will be briefly described below.
Fig. 1 is a flowchart illustrating steps of a membership rights prediction method according to an embodiment of the present invention.
Fig. 2 is a flowchart illustrating steps of a predictive model training method according to an embodiment of the present invention.
FIG. 3 is a flowchart illustrating steps of another method for membership rights prediction according to an embodiment of the present invention.
Fig. 4 is a flowchart illustrating steps of a method for predicting membership rights in an embodiment of the present invention.
Fig. 5 is a block diagram illustrating a membership rights prediction apparatus according to an embodiment of the present invention.
Fig. 6 is a block diagram illustrating another embodiment of an affiliate equity prediction apparatus according to the present invention.
Fig. 7 is a block diagram of an electronic device in an embodiment of the present invention.
Detailed Description
The technical solutions in the embodiments of the present invention will be described below with reference to the drawings in the embodiments of the present invention.
Some video episodes enter the member equity period after the hot broadcast period. For example, the hot-cast period may be after the video episode finishes for a preset duration, such as a preset duration of one week, two weeks, three weeks, and so on. For video episodes in the affiliate interest period, the non-affiliate interest set and the affiliate interest set are usually divided. Wherein, the non-member interest set is free to be watched by all users, and the member interest set is free to be watched by paid member users.
If the division is performed according to the form of fixed non-member interest sets, member interest diversity more suitable for the characteristics of the video episode per se cannot be obtained. Because the material, the number of episodes, the audience, the prior rhythm, and the like of different video episodes may be different, the dividing manner of different video episodes should be differentiated. Based on this, the embodiment of the invention provides a method for analyzing the diversity of each video in different video episodes more pertinently, so as to determine the proper and accurate membership interest diversity of the video episodes.
The member rights prediction method and device in the embodiment of the invention can be applied to a server side. The server may be a server corresponding to the video website.
Fig. 1 is a flowchart illustrating steps of a membership rights prediction method according to an embodiment of the present invention.
As shown in fig. 1, the membership rights prediction method may include the steps of:
In the embodiment of the invention, in order to predict the membership right information of the video diversity more quickly and accurately, a prediction model can be trained in advance. The prediction model can predict and obtain the membership rights and interests information of the video diversity based on the relevant characteristic data of the video diversity. Therefore, the input of the prediction model can be related characteristic data of video diversity, and the output can be membership interest information of the video diversity.
The specific process for training the prediction model will be described in detail in the following examples.
102, acquiring characteristic data to be predicted of the video diversity to be predicted aiming at each video diversity to be predicted in a video drama set to be predicted, and taking the characteristic data to be predicted as input of the prediction model to obtain member rights and interests information of the video diversity to be predicted, which is output by the prediction model.
The episode of the video to be predicted can be any episode containing at least two video diversities. For example, a tv show may be a video episode to be predicted, and each episode in the tv show may be a video diversity; one cartoon can be used as a video episode to be predicted, and each cartoon can be used as a video diversity; the first-level comprehensive program can be used as a video episode to be predicted, and each period in the program can be used as a video episode; a documentary may serve as an episode of video to be predicted, each session in the documentary may serve as a video episode, and so on.
The video diversity to be predicted may be at least two video diversities in a video burst to be predicted. For example, each video diversity in the video episode to be predicted may be regarded as a video diversity to be predicted. For another example, each of the remaining video diversity in the video episode to be predicted, except for the first at least one set and the last at least one set, may be regarded as a video diversity to be predicted, and so on.
And acquiring the characteristic data to be predicted of the video diversity to be predicted aiming at each video diversity to be predicted in the video series set to be predicted. The feature data to be predicted may include, but is not limited to, static features to be predicted that are not related to membership rights and dynamic features to be predicted that are related to membership rights.
The static characteristics to be predicted are not influenced by member rights, namely the static characteristics to be predicted of the video diversity to be predicted are not influenced no matter whether the video diversity to be predicted is the member rights diversity or not, and the static characteristics to be predicted of the video diversity to be predicted are the same under different member rights. Alternatively, although the static feature to be predicted is not related to the interest of the member, in selecting the static feature to be predicted, a feature that can affect the intention of the user to provision the member to some extent may be selected, and the difference in these features can attract the user to provision the member to some extent. Optionally, the static feature to be predicted of the video diversity to be predicted includes, but is not limited to, at least one of the following: the video episode to be predicted comprises the independent broadcasting information of the video episode to be predicted, the search index of the video episode to be predicted, the score of the video episode to be predicted, and the like. For example, if the video episode to be predicted is the video website broadcast alone, the possibility that the user opens the member is higher; the higher the search index of the video episode to be predicted is, the higher the possibility that the user opens the member is; the higher the score of the video episode to be predicted, the greater the likelihood that the user will turn on a member.
The independent broadcasting information is used for indicating whether the video is independently broadcast or not, namely whether the video episode to be predicted is independently broadcast by the video website or not. The search index may include, but is not limited to, at least one of: the search index of the video website, the search index of at least one search website, and the like. The score may include, but is not limited to, a score of at least one scoring platform, and the like.
The dynamic characteristics to be predicted are influenced by the membership rights, that is, whether the video diversity to be predicted is the membership rights diversity or not influences the dynamic characteristics to be predicted of the video diversity to be predicted, and the dynamic characteristics to be predicted of the video diversity to be predicted are different under different membership rights conditions. The dynamic characteristics to be predicted of the video diversity to be predicted include but are not limited to at least one of the following: the number of the non-member interest diversity in the video episode to be predicted, the number of the member interest diversity in the video episode to be predicted, the retention rate of the video diversity to be predicted, the average playing completion degree of the video diversity to be predicted, the playing proportion of the video diversity to be predicted exceeding the preset playing completion degree, the jumping-out rate of the video diversity to be predicted within the preset time length, and the like.
The number of the non-member interest diversity in the video episode to be predicted refers to the number of the non-member interest diversity in the video episode to be predicted when the video diversity to be predicted is taken as a first episode membership interest set. The number of the member interest diversity in the video episode to be predicted refers to the number of the member interest diversity in the video episode to be predicted when the video diversity to be predicted is taken as a first episode of member interest set. The retention rate of the to-be-predicted video diversity is the user proportion of the next diversity for continuously playing the to-be-predicted video diversity after the to-be-predicted video diversity is played. The average playing completion degree of the video diversity to be predicted refers to an average value of playing completion degrees of all users playing the video diversity to be predicted. The playing proportion of the video diversity to be predicted exceeding the preset playing completion degree refers to the proportion of users exceeding the preset playing completion degree in all the users playing the video diversity to be predicted. The jumping-out rate in the preset time length of the video diversity to be predicted refers to the proportion of users who jump out in the preset time length in all the users who play the video diversity to be predicted.
For the preset playing completion degree, any applicable numerical value can be selected according to actual experience. For example, considering that the playing completion of the video of interest by the user is high, the playing completion of the video of interest by most users may be set as the preset playing completion. For example, the preset playing completion is 75%, 80%, 85%, and so on. For the preset duration, any suitable value can be selected according to actual experience. For example, considering that the user may jump out for a video that is not of interest in the first few minutes, the playing time length when most of the users jump out for the video that is not of interest can be set as the preset time length. E.g., the predetermined time period is 4 minutes, 5 minutes, 6 minutes, etc.
And aiming at each video diversity to be predicted in the video series set to be predicted, taking the characteristic data to be predicted of the video diversity to be predicted as the input of the prediction model, and obtaining the membership rights and interests information of the video diversity to be predicted, which is output by the prediction model.
And 103, determining the membership right diversity in the video drama set to be predicted based on the membership right information of the video diversity to be predicted.
The membership right information of the video diversity to be predicted can indicate whether the video diversity to be predicted is suitable to be used as the membership right diversity. Therefore, based on the membership right information of each video diversity to be predicted, the membership right diversity in the video drama set to be predicted can be determined. Details will be discussed in the following examples.
The membership right information of the video diversity to be predicted may include, but is not limited to: the membership conversion rate when the video diversity to be predicted is used as the last set of non-membership right diversity, the retention rate when the video diversity to be predicted is used as the last set of non-membership right diversity, the playing amount when the video diversity to be predicted is used as the last set of non-membership right diversity, and the like.
In the embodiment of the invention, the diversity of the videos to be predicted in the video episode can be analyzed more pertinently aiming at different video episodes, so that the more appropriate and accurate member rights and interests diversity of the video episode can be determined, the member can be more powerfully promoted by a user, and the user activity can be effectively improved.
In the embodiment of the invention, any model structure can be selected as the prediction model. May include, but is not limited to: linear Regression (Linear Regression) model, logistic Regression (Logistic Regression) model, polynomial Regression (polymorphous Regression) model, stepwise Regression (Stepwise Regression) model, ridge Regression (Ridge Regression) model, SVM (Support Vector Machine) model, FM (Factorization Machine) model, etc.
The prediction model is used for predicting the membership rights and interests information of the video diversity based on the characteristic data of the video diversity, and reflects the relationship between the membership rights and interests information of the video diversity and the characteristic data of the video diversity. Because the linear regression model can effectively describe the relationship between the dependent variable and the independent variable, the linear regression model can be selected as the prediction model in the embodiment of the invention.
Fig. 2 is a flowchart illustrating steps of a predictive model training method according to an embodiment of the present invention.
As shown in fig. 2, the predictive model training method may include the following steps:
And acquiring the video episode in the member interest period from the video website as a sample video episode. For example, video episodes that end more than a predetermined number of days (e.g., 60 days, 90 days, etc.) are captured, and video episodes that are marketed or free during the member's equity period are filtered out.
Optionally, the last set of non-membership rights diversity in the sample video drama set is used as the sample video diversity. And acquiring the characteristic data of the sample video diversity and the actual membership rights and interests information of the sample video diversity aiming at each sample video diversity. And taking sample characteristic data of a sample video diversity and actual membership rights and interests information as sample data.
Similar to the video diversity to be predicted, the sample feature data of the sample video diversity may include, but is not limited to, a sample static feature unrelated to the membership interest and a sample dynamic feature related to the membership interest. Wherein the sample static characteristics of the sample video diversity include, but are not limited to, at least one of: the sample video episode (i.e., the sample video episode to which the sample video diversity belongs) may include, for example, a search index for the sample video episode, a score for the sample video episode, and so on. The sample dynamic characteristics of the sample video diversity include, but are not limited to, at least one of: the number of non-member interest diversity in the sample video episode (in case that the sample video diversity is used as the last non-member interest diversity), the number of member interest diversity in the sample video episode (in case that the sample video diversity is used as the last non-member interest diversity), the retention rate of the sample video diversity, the average playing completion degree of the sample video diversity, the playing proportion of the sample video diversity exceeding the preset playing completion degree, the jumping-out rate of the sample video diversity within the preset time length, and the like.
Optionally, after obtaining the sample feature data of the sample video diversity, the sample feature data may be subjected to data discretization, missing value padding, and the like. Wherein, the data discretization may refer to converting data in a non-numerical form into a numerical form. For example, non-numerical forms of "yes" and "no" of the exclusive broadcast information are converted into numerical forms of "0" and "1", and so on. Missing value padding may refer to supplementing missing sample feature data with a preset value. For example, if a sample video diversity loss score is found, the score of the sample video diversity may be supplemented with a preset experience score, such as a minimum score, an average score, etc., as the score of the sample video diversity.
In an alternative embodiment, the membership rights information for sample video diversity may be used to indicate membership conversion for sample video diversity as the last set of non-membership rights diversity. The member conversion rate is a ratio of the number of users who open members among users who play videos to the number of users who play videos. Therefore, the actual membership right information of the sample video diversity can be the ratio of the number of users who open members among the users who play the sample video diversity to the number of users who play the sample video diversity.
And 202, training a preset model to be trained by using the sample data.
The model to be trained refers to a model with a prediction function that has not been trained yet. In the process of training the model to be trained by using the sample data, the input of the model to be trained is sample characteristic data of the sample video diversity, and the output of the model to be trained is predicted membership rights and interests information of the sample video diversity.
In regression analysis, the influence of multiple independent variables on dependent variables is generally considered, and when multiple collinearity problems exist in the independent variables, the problems cannot be well solved by common multivariate linear regression. The PLS (Partial least squares regression) model is a multivariate statistical method for solving the problem of collinearity and researching the influence relationship when processing small samples. The PLS model integrates principal component analysis, canonical correlation, and multiple linear regression. Briefly, the principle of the PLS model can be understood as follows: the PLS model respectively concentrates a plurality of X and a plurality of Y into components (X corresponds to a principal component U, Y corresponds to a principal component V) by applying the principle of principal component analysis, and then the relation between X and U and the relation between Y and V can be analyzed by means of a typical correlation principle; and analyzing the relation of X to V by combining a multiple linear regression principle, thereby researching the relation of X to Y. In the regression analysis, if it is desired to study the influence relationship such that the normal distribution of the dependent variable is theoretically required and the sample size cannot be too small or too large, the PLS model can be used. In the embodiment of the present invention, the last non-membership rights diversity in the video episode is used as a sample, the sample size is small, but the sample size is not too small (e.g., about 200 samples), and the feature data of the samples have collinearity, so that the PLS model in the sklern machine learning package in Python can be used as the prediction model. The PLS model is suitable for prediction projects with small sample size, and has the advantages of interpretability, high algorithm speed and the like.
For example, the PLS model can be expressed as the following equation:
y＝B 0 +B 1 x 1 +B 2 x 2 +……+B k x k formula one
Wherein x is 1 、x 2 ……x k Is an argument which is an input, i.e. x 1 、x 2 ……x k The characteristic data are respectively. y is a dependent variable and the dependent variable is an output, namely y is predicted membership rights information. B is 0 、B 1 、B 2 ……B k Are the regression coefficients to be trained.
In an optional implementation manner, in order to increase the dimension of the sample feature data so as to more comprehensively and accurately characterize the features of the sample video diversity, after the sample feature data of the sample video diversity is acquired, the sample feature data of the sample video diversity is converted to obtain new sample feature data of the sample video diversity, and the sample feature data and the new sample feature data are used as the inputs of the model to be trained.
Optionally, four arithmetic operations and/or logarithmic transformation may be performed on the sample feature data of the sample video diversity, so as to obtain the newly added sample feature data of the sample video diversity.
In the four arithmetic processes, four arithmetic operations can be performed on at least two sample feature data to obtain a new added sample feature data. The four arithmetic operations may include, but are not limited to, at least one of the following: addition, subtraction, multiplication, division, exponentiation, rounding, and the like. Through four arithmetic operations, the features with collinearity can be associated. For example, for the single sample feature data of the search index and the single sample feature data of the score, the product of the search index and the two sample feature data of the score may more accurately indicate the interest degree of the user in the sample video episode, so the product of the search index and the two sample feature data of the score may be used as an additional sample feature data.
In the log transformation process, part or all of the sample feature data may be subjected to log transformation. Wherein the logarithmic transformation may be an ln transformation. Through logarithmic transformation, the sample characteristic data can be normally distributed as much as possible, so that the condition that a large amount of data is biased to one end is avoided.
In an alternative embodiment, considering that some feature types in the sample feature data and the newly added sample feature data may have low correlation with the output of the prediction model, the feature types with low correlation may contribute less to the process of model training. Therefore, part of the feature data can be further selected from the sample feature data and the newly added sample feature data to be used as the input of the model to be trained. It should be noted that, one sample feature data corresponds to one feature type, and one newly added sample data corresponds to one feature type. For example, the search index is a feature type, the score is a feature type, and so on.
Optionally, for each feature type in the sample feature data and the newly added sample feature data, a correlation coefficient between the feature type and the output of the prediction model is calculated, and a feature type with a correlation coefficient larger than a preset correlation threshold is selected as a target feature type. And selecting target sample characteristic data corresponding to the target characteristic type from the sample characteristic data and the newly added sample characteristic data, and taking the target sample characteristic data as the input of the model to be trained.
The correlation coefficient may be a pearson correlation coefficient, a spearman correlation coefficient, a kendall correlation coefficient, or the like. For example, a Pearson correlation coefficient refers to the product of the covariance between two variables divided by their respective standard deviations. The Pearson correlation coefficient can measure the degree of linear correlation, and the larger the correlation coefficient of the two is, the larger the degree of correlation of the two is. Any suitable value may be set for the correlation threshold according to practical experience, which is not limited in this embodiment of the present invention. For example, the correlation threshold may be 0.7, 0.8, etc.
And 203, after training is judged to be finished based on the actual member rights and interests information and the predicted member rights and interests information, taking a trained model as the predicted model.
In an alternative embodiment, the degree of fit (i.e., the coefficient of determination) R of the model may be used 2 And performing parameter adjustment training model for optimizing the target. Based on actual membership rights information of sample diversity and predicted membership rights information of sample diversity, fitting degree R of model can be calculated 2 。
Suppose y is the actual membership rights information of the sample diversity, and the average value of the actual membership rights information of all the sample diversity is
The predicted membership entitlement information for sample diversity is ≥ h>
Sum of squares
Regression sum of squares
Sum of squares of residuals
Then there are: SST = SSR + SSE
In the above formula, n represents the total number of sample data, and i represents the ith sample data.
R 2 Is 1. The closer the value of R2 is to 1, the better the fitting degree of the prediction model to the observed value is; conversely, a smaller value of R2 indicates a poorer degree of fitting of the prediction model to the observed value. Therefore, a fitting degree threshold can be preset, and R can be judged 2 Above a fitness threshold, training completion may be determined. Any suitable value for the fitness threshold may be set based on practical experience. For example, the fitting degree corresponding to most of the trained prediction models can be set as a fitting degree threshold, and so on.
After the prediction model is trained, member interest information of the video diversity to be predicted can be predicted according to the video diversity to be predicted in the episode to be predicted by utilizing the prediction model, so that the member interest diversity in the video episode to be predicted is determined.
FIG. 3 is a flowchart illustrating steps of another method for membership rights prediction according to an embodiment of the present invention.
As shown in fig. 3, the membership rights prediction method may include the steps of:
Optionally, four arithmetic operations and/or logarithmic transformation may be performed on the to-be-predicted feature data of the to-be-predicted video diversity, so as to obtain the newly-added to-be-predicted feature data of the to-be-predicted video diversity. For the specific processing procedure, reference may be made to the specific description in step 202, and details are not described herein in this embodiment of the present invention.
And 304, taking the feature data to be predicted and the newly added feature data to be predicted as the input of the prediction model to obtain the membership right information of the video diversity to be predicted, which is output by the prediction model.
And aiming at each video diversity to be predicted in the video series set to be predicted, taking the characteristic data to be predicted of the video diversity to be predicted and newly added characteristic data to be predicted as the input of the prediction model, and obtaining the membership rights and interests information of the video diversity to be predicted, which is output by the prediction model.
And 305, determining the membership right diversity in the video drama set to be predicted based on the membership right information of the video diversity to be predicted.
Optionally, the membership right information is used to indicate a membership conversion rate when the video diversity to be predicted is used as the last set of non-membership right diversity.
In an optional implementation manner, the process of determining the membership right diversity in the video episode to be predicted based on the membership right information of the video episode to be predicted may include: and selecting the video diversity to be predicted of the next set of the video diversity to be predicted with the maximum membership conversion rate, and taking the selected video diversity to be predicted as the first set membership right diversity in the video drama set to be predicted. That is, the diversity of the videos to be predicted of the next set of the diversity of the selected videos to be predicted is started, and the diversity of the videos to be predicted of the last set is ended, so that the diversity of the membership rights and interests in the video drama set to be predicted is used.
In the implementation, if only the video diversity to be predicted with the maximum member conversion rate is considered in the above manner, a situation may occur in which the video diversity to be predicted next to the video diversity to be predicted with the maximum member conversion rate is not suitable as the member interest diversity of the first set. For example, in a video episode to be predicted, the member conversion rate of the video diversity to be predicted in the 1 st set is 10%, the member conversion rate of the video diversity to be predicted in the 2 nd set is 20%, the member conversion rate of the video diversity to be predicted in the 3 rd set is 50%, the member conversion rate of the video diversity to be predicted in the 4 th set is 60%, the member conversion rate of the video diversity to be predicted in the 5 th set is 20%, the member conversion rate of the video diversity to be predicted in the 6 th set is 55%, the member conversion rate of the video diversity to be predicted in the 7 th set is 70%, the member conversion rate of the video diversity to be predicted in the 8 th set is 20%, and the member conversion rate of the video diversity to be predicted in the 9 th set, that is, the member conversion rate of the video diversity to be predicted in the last set is 10%. In this case, although the membership conversion rate of the 7 th group is the highest, if the 8 th group is regarded as the first group membership interest and interest group, the user may give up opening the membership because the user can only see two groups after purchasing the membership and feels worthless.
In an optional implementation manner, the process of determining the membership interest diversity in the video episode to be predicted based on the membership interest information of the video episode to be predicted may include: and performing aggregation operation on the member conversion rates to obtain at least two sets, selecting the set with the maximum member conversion rate sum, and taking the video diversity to be predicted of the next set of the video diversity to be predicted of the foremost set in the selected sets as the first set of member interest diversity in the video drama set to be predicted. That is, the diversity of the video to be predicted of the next set of the diversity of the video to be predicted of the foremost set in the selected set is started, and the diversity of the video to be predicted of the last set is ended, so that the diversity of the member rights and interests in the video drama set to be predicted is used; starting from the diversity of the video to be predicted in the first set and ending to the diversity of the video to be predicted in the foremost set in the selected set, and taking the diversity of the non-member rights in the series set of the video to be predicted.
For example, the member conversion rates of the 1 st to 9 th sets in the above-mentioned example are aggregated, and the 1 st, 2 nd, 5 th, 8 th and 9 th sets may be aggregated into a first set, and the 3 rd, 4 th, 6 th and 7 th sets may be aggregated into a second set. The sum of the membership conversion rates of the second set is the maximum, so that the set 4 next to the set 3 of the video to be predicted in the first set in the second set can be used as the first set membership interest diversity in the video series to be predicted. That is, the 1 st to 3 rd sets are non-membership right diversity, and the 4 th to 9 th sets are membership right diversity.
In another alternative embodiment, the process of determining the membership right diversity in the video episode to be predicted based on the membership right information of the video episode to be predicted may include: and performing aggregation operation on the member conversion rates to obtain at least two sets, selecting the set with the maximum member conversion rate sum, and using the video diversity to be predicted of the next set of the video diversity to be predicted in the selected set as the member rights diversity in the video series set to be predicted.
For example, the member conversion rates of the 1 st to 9 th sets in the above-mentioned example are aggregated, and the 1 st, 2 nd, 5 th, 8 th and 9 th sets may be aggregated into a first set, and the 3 rd, 4 th, 6 th and 7 th sets may be aggregated into a second set. The sum of the membership conversion rates of the second set is the maximum, so that the 4 th set, the 5 th set, the 7 th set and the 8 th set of the 3 rd set, the 4 th set, the 6 th set and the 7 th set in the second set are used as membership rights and interests in the video play set to be predicted; and taking the diversity of other videos to be predicted, namely the 1 st set, the 2 nd set, the 3 rd set, the 6 th set and the 9 th set, as the non-member rights and interests diversity in the video series set to be predicted.
Alternatively, the aggregation operation may be performed by k-means clustering or the like.
FIG. 4 is a flowchart illustrating a method for predicting membership rights in an embodiment of the present invention.
As shown in fig. 4, the membership rights prediction method may include the steps of:
And 403, performing conversion processing on the feature data to be predicted to obtain new feature data to be predicted.
And 404, selecting target feature data to be predicted corresponding to the target feature type from the feature data to be predicted and the newly added feature data to be predicted.
The target feature type is obtained by calculation in the training process of the prediction model, and the feature type of which the correlation coefficient with the output of the prediction model is larger than a preset correlation threshold value. Reference may be made specifically to the description of step 202 above.
And 405, taking the target characteristic data to be predicted as the input of the prediction model to obtain the membership right information of the video diversity to be predicted, which is output by the prediction model.
And aiming at each video diversity to be predicted in the video series set to be predicted, taking the target characteristic data to be predicted of the video diversity to be predicted as the input of the prediction model, and obtaining the membership rights and interests information of the video diversity to be predicted, which is output by the prediction model.
And step 406, determining the membership right diversity in the video drama set to be predicted based on the membership right information of the video diversity to be predicted.
For the specific process of step 406, reference may be made to the related description in step 305 above, and the embodiment of the present invention will not be discussed in detail here.
Alternatively, the effect of the member interest prediction method may also be verified. For example, an experimental group and a control group can be selected. The experimental group comprises a plurality of experimental video episodes, and the control group comprises a plurality of control video episodes. And adjusting the first set membership interest diversity of the experimental video episode in the experimental group into the optimal diversity output by the prediction model, and maintaining the control video episode in the control group in the original state. After the experiment lasts for a certain period of time, the experimental group can be found to achieve a certain improvement of the member conversion rate compared with the control group.
In the embodiment of the invention, the member interest initial set which can reach the highest member conversion rate can be searched in a model prediction mode, and correspondingly, the content which is most wonderful and most attractive to the user to see can be found, so that some content operation popularization and the like can be carried out on the content, the content can be watched to the maximum extent, the watching duration of the user can be prolonged, and the user can stay on the platform for a longer time.
Fig. 5 is a block diagram illustrating a membership rights prediction apparatus according to an embodiment of the present invention.
As shown in fig. 5, the membership rights prediction apparatus may include the following modules:
a first obtaining module 501, configured to obtain a pre-trained prediction model; the prediction model is obtained by training based on a plurality of sample data, wherein the sample data comprises sample characteristic data of sample video diversity and actual membership rights and interests information of the sample video diversity;
a second obtaining module 502, configured to obtain, for each to-be-predicted video diversity in a to-be-predicted video drama set, to-be-predicted feature data of the to-be-predicted video diversity;
the prediction module 503 is configured to use the feature data to be predicted as an input of the prediction model to obtain membership right information of the video diversity to be predicted, which is output by the prediction model;
a determining module 504, configured to determine, based on the membership right information of the video episode to be predicted, the membership right information in the video episode to be predicted.
Fig. 6 is a block diagram illustrating a membership rights prediction apparatus according to an embodiment of the present invention.
As shown in fig. 6, the membership rights prediction apparatus may include the following modules:
a first obtaining module 601, configured to obtain a pre-trained prediction model; the prediction model is obtained by training based on a plurality of sample data, wherein the sample data comprises sample characteristic data of sample video diversity and actual membership rights and interests information of the sample video diversity;
a second obtaining module 602, configured to obtain, for each to-be-predicted video diversity in a to-be-predicted video drama set, to-be-predicted feature data of the to-be-predicted video diversity;
the prediction module 603 is configured to use the feature data to be predicted as an input of the prediction model, so as to obtain membership right information of the video diversity to be predicted, which is output by the prediction model;
a determining module 604, configured to determine, based on the member interest information of the video episode to be predicted, the member interest diversity in the video episode to be predicted.
Optionally, the feature data to be predicted includes a static feature to be predicted, which is not related to the membership rights and interests, and a dynamic feature to be predicted, which is related to the membership rights and interests; the static features to be predicted comprise at least one of: the video episode to be predicted comprises the independent broadcasting information of the video episode to be predicted, the search index of the video episode to be predicted and the score of the video episode to be predicted, wherein the independent broadcasting information is used for indicating whether the video episode to be predicted is independently broadcast or not; the dynamic features to be predicted comprise at least one of the following: the number of the non-member right and interest diversity in the video episode to be predicted, the number of the member right and interest diversity in the video episode to be predicted, the retention rate of the video diversity to be predicted, the average playing completion degree of the video diversity to be predicted, the playing proportion of the video diversity to be predicted exceeding the preset playing completion degree, and the jumping rate of the video diversity to be predicted within the preset time length.
Optionally, the member interest information is used to indicate a member conversion rate when the video diversity to be predicted is used as a last set of non-member interest diversity; the determining module 603 is specifically configured to select a next set of to-be-predicted video diversity of the to-be-predicted video diversity with the largest member conversion rate, and use the selected to-be-predicted video diversity as a first set of member right diversity in the to-be-predicted video drama set.
Optionally, the apparatus further comprises: a first conversion module 605, configured to perform conversion processing on the feature data to be predicted to obtain new feature data to be predicted; the conversion processing comprises four arithmetic operations and/or logarithmic transformation; the prediction module 603 is specifically configured to use the feature data to be predicted and the newly added feature data to be predicted as inputs of the prediction model.
Optionally, the apparatus further comprises: a first selecting module 606, configured to select, from the feature data to be predicted and the newly added feature data to be predicted, target feature data to be predicted corresponding to a target feature type; the target feature type is obtained by calculation in the training process of the prediction model, and the correlation coefficient with the output of the prediction model is larger than a preset correlation threshold; the prediction module 603 is specifically configured to use the target feature data to be predicted as an input of the prediction model.
Optionally, the predictive model is trained by: a third obtaining module 607, configured to obtain the sample data; a training module 608, configured to train a preset model to be trained by using the sample data; the input of the model to be trained is the sample characteristic data, and the output is the predicted membership rights and interests information of the sample video diversity; and the judging module 609 is configured to judge that training is completed based on the actual member equity information and the predicted member equity information, and then use a trained model as the prediction model.
Optionally, the apparatus further comprises: the second conversion module 610 is configured to perform conversion processing on the sample feature data to obtain new sample feature data; a calculating module 611, configured to calculate, for each feature type in the sample feature data and the newly added sample feature data, a correlation coefficient between the feature type and an output of the prediction model, and select, as a target feature type, a feature type whose correlation coefficient is greater than a preset correlation threshold; a second selecting module 612, configured to select, from the sample feature data and the newly added sample feature data, target sample feature data corresponding to the target feature type; and the input of the model to be trained is the target sample characteristic data.
Optionally, the predictive model is a partial least squares regression, PLS, model.
In the embodiment of the invention, the diversity of the videos to be predicted in the video episode can be analyzed more pertinently aiming at different video episodes, so that the more appropriate and accurate member rights and interests diversity of the video episode can be determined, the member can be more powerfully promoted by a user, and the user activity can be effectively improved.
An embodiment of the present invention further provides an electronic device, as shown in fig. 7, including a processor 701, a communication interface 702, a memory 703 and a communication bus 704, where the processor 701, the communication interface 702, and the memory 703 complete mutual communication through the communication bus 704.
A memory 703 for storing a computer program;
the processor 701 is configured to implement the following steps when executing the program stored in the memory 703:
obtaining a pre-trained prediction model; the prediction model is obtained by training based on a plurality of sample data, wherein the sample data comprises sample characteristic data of sample video diversity and actual membership rights and interests information of the sample video diversity;
acquiring characteristic data to be predicted of the video diversity to be predicted aiming at each video diversity to be predicted in a video series set to be predicted, and taking the characteristic data to be predicted as the input of the prediction model to obtain member rights and interests information of the video diversity to be predicted, which is output by the prediction model;
and determining the membership right diversity in the video drama set to be predicted based on the membership right information of the video diversity to be predicted.
Optionally, the feature data to be predicted includes a static feature to be predicted, which is not related to the membership rights and interests, and a dynamic feature to be predicted, which is related to the membership rights and interests; the static features to be predicted comprise at least one of: the method comprises the steps of the recording of the video episode to be predicted, the recording of the video episode to be predicted and the scoring of the video episode to be predicted, wherein the independent broadcasting information is used for indicating whether the video episode to be predicted is independently broadcast; the dynamic features to be predicted comprise at least one of the following: the number of the non-member right and interest diversity in the video episode to be predicted, the number of the member right and interest diversity in the video episode to be predicted, the retention rate of the video diversity to be predicted, the average playing completion degree of the video diversity to be predicted, the playing proportion of the video diversity to be predicted exceeding the preset playing completion degree, and the jumping rate of the video diversity to be predicted within the preset time length.
Optionally, the membership right information is used to indicate a membership conversion rate when the video diversity to be predicted is used as a last set of non-membership right diversity; the determining the membership right diversity in the video drama set to be predicted based on the membership right information of the video drama set to be predicted comprises the following steps: and selecting the video diversity to be predicted of the next set of the video diversity to be predicted with the maximum membership conversion rate, and taking the selected video diversity to be predicted as the first set membership right diversity in the video drama set to be predicted.
Optionally, after obtaining the to-be-predicted feature data of the to-be-predicted video diversity, the method further includes: converting the characteristic data to be predicted to obtain newly added characteristic data to be predicted; the conversion processing comprises four arithmetic operations and/or logarithmic transformation; the using the feature data to be predicted as the input of the prediction model comprises: and taking the feature data to be predicted and the newly added feature data to be predicted as the input of the prediction model.
Optionally, after performing conversion processing on the feature data to be predicted to obtain new feature data to be predicted, the method further includes: selecting target feature data to be predicted corresponding to the target feature type from the feature data to be predicted and the newly added feature data to be predicted; the target feature type is obtained by calculation in the training process of the prediction model, and the correlation coefficient with the output of the prediction model is larger than a preset correlation threshold; the using the feature data to be predicted as the input of the prediction model comprises: and taking the target characteristic data to be predicted as the input of the prediction model.
Optionally, the predictive model is trained by: acquiring the sample data; training a preset model to be trained by using the sample data; the input of the model to be trained is the sample characteristic data, and the output is the predicted membership rights and interests information of the sample video diversity; and after training is judged to be completed based on the actual member rights and interests information and the predicted member rights and interests information, taking a trained model as the predicted model.
Optionally, after obtaining the sample data, the method further includes: converting the sample characteristic data to obtain newly added sample characteristic data; calculating a correlation coefficient between the feature type and the output of the prediction model aiming at each feature type in the sample feature data and the newly added sample feature data, and selecting the feature type with the correlation coefficient larger than a preset correlation threshold value as a target feature type; selecting target sample characteristic data corresponding to the target characteristic type from the sample characteristic data and the newly added sample characteristic data; and the input of the model to be trained is the target sample characteristic data.
The communication bus mentioned in the above terminal may be a Peripheral Component Interconnect (PCI) bus, an Extended Industry Standard Architecture (EISA) bus, or the like. The communication bus may be divided into an address bus, a data bus, a control bus, etc. For ease of illustration, only one thick line is shown, but this is not intended to represent only one bus or type of bus.
The communication interface is used for communication between the terminal and other equipment.
The Memory may include a Random Access Memory (RAM) or a non-volatile Memory (non-volatile Memory), such as at least one disk Memory. Alternatively, the memory may be at least one memory device located remotely from the processor.
The Processor may be a general-purpose Processor, including a Central Processing Unit (CPU), a Network Processor (NP), and the like; the Integrated Circuit may also be a Digital Signal Processor (DSP), an Application Specific Integrated Circuit (ASIC), a Field Programmable Gate Array (FPGA) or other Programmable logic device, a discrete Gate or transistor logic device, or a discrete hardware component.
In yet another embodiment of the present invention, a computer-readable storage medium is further provided, which stores instructions that, when executed on a computer, cause the computer to implement the membership rights prediction method according to any one of the above embodiments.
In yet another embodiment of the present invention, there is also provided a computer program product containing instructions which, when run on a computer, cause the computer to implement the membership rights prediction method as described in any of the above embodiments.
In the above embodiments, the implementation may be wholly or partially realized by software, hardware, firmware, or any combination thereof. When implemented in software, may be implemented in whole or in part in the form of a computer program product. The computer program product includes one or more computer instructions. When loaded and executed on a computer, cause the processes or functions described in accordance with the embodiments of the invention to occur, in whole or in part. The computer may be a general purpose computer, a special purpose computer, a network of computers, or other programmable device. The computer instructions may be stored in a computer readable storage medium or transmitted from one computer readable storage medium to another, for example, from one website site, computer, server, or data center to another website site, computer, server, or data center via wired (e.g., coaxial cable, fiber optic, digital Subscriber Line (DSL)) or wireless (e.g., infrared, wireless, microwave, etc.). The computer-readable storage medium can be any available medium that can be accessed by a computer or a data storage device, such as a server, a data center, etc., that incorporates one or more of the available media. The usable medium may be a magnetic medium (e.g., floppy Disk, hard Disk, magnetic tape), an optical medium (e.g., DVD), or a semiconductor medium (e.g., solid State Disk (SSD)), among others.
It should be noted that, in this document, relational terms such as first and second, and the like are used solely to distinguish one entity or action from another entity or action without necessarily requiring or implying any actual such relationship or order between such entities or actions. Also, the terms "comprises," "comprising," or any other variation thereof, are intended to cover a non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements does not include only those elements but may include other elements not expressly listed or inherent to such process, method, article, or apparatus. Without further limitation, an element defined by the phrases "comprising a," "...," or "comprising" does not exclude the presence of additional like elements in a process, method, article, or apparatus that comprises the element.
All the embodiments in the present specification are described in a related manner, and the same and similar parts among the embodiments may be referred to each other, and each embodiment focuses on the differences from the other embodiments. In particular, for the system embodiment, since it is substantially similar to the method embodiment, the description is simple, and for the relevant points, reference may be made to the partial description of the method embodiment.
The above description is only for the preferred embodiment of the present invention, and is not intended to limit the scope of the present invention. Any modification, equivalent replacement, or improvement made within the spirit and principle of the present invention shall fall within the protection scope of the present invention.
Claims (9)
1. A method for predicting membership rights, comprising:
acquiring a pre-trained prediction model; the prediction model is obtained by training based on a plurality of sample data, and the sample data comprises sample characteristic data of sample video diversity and actual membership interest information of the sample video diversity;
acquiring feature data to be predicted of each video diversity to be predicted in a video drama set to be predicted, and taking the feature data to be predicted as input of a prediction model to obtain member interest information of the video diversity to be predicted, wherein the feature data to be predicted comprises static features to be predicted, which are irrelevant to member interest, and dynamic features to be predicted, which are relevant to the member interest;
the static feature to be predicted comprises at least one of the following: the video episode to be predicted comprises the independent broadcasting information of the video episode to be predicted, the search index of the video episode to be predicted and the score of the video episode to be predicted, wherein the independent broadcasting information is used for indicating whether the video episode to be predicted is independently broadcast or not;
the dynamic features to be predicted comprise at least one of the following: the number of non-member right and interest diversity in the video episode to be predicted, the number of member right and interest diversity in the video episode to be predicted, the retention rate of the video diversity to be predicted, the average playing completion degree of the video diversity to be predicted, the playing proportion of the video diversity to be predicted exceeding the preset playing completion degree, and the jumping rate of the video diversity to be predicted within the preset time length;
and determining the membership right diversity in the video drama set to be predicted based on the membership right information of the video diversity to be predicted.
2. The method of claim 1, wherein the membership interest information is used to indicate membership conversion rate when the video diversity to be predicted is used as a last set of non-membership interest diversity; the determining the membership right diversity in the video drama set to be predicted based on the membership right information of the video diversity to be predicted comprises the following steps:
and selecting the video diversity to be predicted of the next set of the video diversity to be predicted with the maximum member conversion rate, and taking the selected video diversity to be predicted as the first set of member interest diversity in the video drama set to be predicted.
3. The method according to claim 1, further comprising, after obtaining the to-be-predicted feature data of the to-be-predicted video diversity, the step of:
converting the characteristic data to be predicted to obtain newly added characteristic data to be predicted; the conversion processing comprises four arithmetic operations and/or logarithmic transformation;
the using the feature data to be predicted as the input of the prediction model comprises: and taking the feature data to be predicted and the newly added feature data to be predicted as the input of the prediction model.
4. The method of claim 3, further comprising, after performing conversion processing on the feature data to be predicted to obtain new feature data to be predicted:
selecting target feature data to be predicted corresponding to the target feature type from the feature data to be predicted and the newly added feature data to be predicted; the target feature type is obtained by calculation in the training process of the prediction model, and the correlation coefficient with the output of the prediction model is larger than a preset correlation threshold;
the using the feature data to be predicted as the input of the prediction model comprises: and taking the target characteristic data to be predicted as the input of the prediction model.
5. The method of claim 1, wherein the predictive model is trained by:
acquiring the sample data;
training a preset model to be trained by using the sample data; the input of the model to be trained is the sample characteristic data, and the output is the predicted membership interest information of the sample video diversity;
and after training is judged to be completed based on the actual member rights and interests information and the predicted member rights and interests information, taking a trained model as the predicted model.
6. The method of claim 5, further comprising, after obtaining the sample data:
converting the sample characteristic data to obtain newly added sample characteristic data;
calculating a correlation coefficient between the characteristic type and the output of the prediction model aiming at each characteristic type in the sample characteristic data and the newly added sample characteristic data, and selecting the characteristic type with the correlation coefficient larger than a preset correlation threshold value as a target characteristic type;
selecting target sample characteristic data corresponding to the target characteristic type from the sample characteristic data and the newly added sample characteristic data; and the input of the model to be trained is the target sample characteristic data.
7. A membership rights prediction apparatus comprising:
the first acquisition module is used for acquiring a pre-trained prediction model; the prediction model is obtained by training based on a plurality of sample data, wherein the sample data comprises sample characteristic data of sample video diversity and actual membership rights and interests information of the sample video diversity;
the second acquisition module is used for acquiring the characteristic data to be predicted of the video diversity to be predicted aiming at each video diversity to be predicted in the video diversity to be predicted;
the prediction module is used for taking the characteristic data to be predicted as the input of the prediction model to obtain the member interest information of the video diversity to be predicted, which is output by the prediction model, wherein the characteristic data to be predicted comprises a static characteristic to be predicted, which is irrelevant to the member interest, and a dynamic characteristic to be predicted, which is relevant to the member interest;
the static feature to be predicted comprises at least one of the following: the video episode to be predicted comprises the independent broadcasting information of the video episode to be predicted, the search index of the video episode to be predicted and the score of the video episode to be predicted, wherein the independent broadcasting information is used for indicating whether the video episode to be predicted is independently broadcast or not;
the dynamic features to be predicted comprise at least one of the following: the number of non-member right and interest diversity in the video episode to be predicted, the number of member right and interest diversity in the video episode to be predicted, the retention rate of the video diversity to be predicted, the average playing completion degree of the video diversity to be predicted, the playing proportion of the video diversity to be predicted exceeding the preset playing completion degree, and the jumping rate of the video diversity to be predicted within the preset time length;
and the determining module is used for determining the membership right diversity in the video drama set to be predicted based on the membership right information of the video diversity to be predicted.
8. An electronic device is characterized by comprising a processor, a communication interface, a memory and a communication bus, wherein the processor and the communication interface are used for realizing mutual communication by the memory through the communication bus;
a memory for storing a computer program;
a processor for implementing the method of any one of claims 1 to 6 when executing a program stored in the memory.
9. 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 6.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202110790348.5A CN113676770B (en) | 2021-07-13 | 2021-07-13 | Member rights prediction method, member rights prediction device, electronic equipment and storage medium |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202110790348.5A CN113676770B (en) | 2021-07-13 | 2021-07-13 | Member rights prediction method, member rights prediction device, electronic equipment and storage medium |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN113676770A CN113676770A (en) | 2021-11-19 |
| CN113676770B true CN113676770B (en) | 2023-04-18 |
Family
ID=78539109
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202110790348.5A Active CN113676770B (en) | 2021-07-13 | 2021-07-13 | Member rights prediction method, member rights prediction device, electronic equipment and storage medium |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN113676770B (en) |
Families Citing this family (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN114339402A (en) * | 2021-12-31 | 2022-04-12 | 北京字节跳动网络技术有限公司 | Video playing completion rate prediction method, device, medium and electronic equipment |
Family Cites Families (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| EP2455904A4 (en) * | 2009-07-14 | 2013-01-16 | Rakuten Inc | Affiliation apportionment device, affiliation apportionment system, affiliation apportionment method, affiliation apportionment program, and computer-readable recording medium |
| CN106303590B (en) * | 2016-08-08 | 2020-08-18 | 腾讯科技（深圳）有限公司 | Method and device for inviting to watch video film |
| CN110933473A (en) * | 2019-12-10 | 2020-03-27 | 北京爱奇艺科技有限公司 | Video playing heat determining method and device |
| CN111062527B (en) * | 2019-12-10 | 2023-12-05 | 北京爱奇艺科技有限公司 | Video traffic collection prediction method and device |
| CN113052375A (en) * | 2021-03-19 | 2021-06-29 | 上海森宇文化传媒股份有限公司 | Method and device for predicting play volume of episode |
-
2021
- 2021-07-13 CN CN202110790348.5A patent/CN113676770B/en active Active
Also Published As
| Publication number | Publication date |
|---|---|
| CN113676770A (en) | 2021-11-19 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| Hoiles et al. | Engagement and popularity dynamics of YouTube videos and sensitivity to meta-data | |
| US20200034383A1 (en) | Content Recommendation System | |
| CN108875022B (en) | Video recommendation method and device | |
| CN110909205B (en) | Video cover determination method and device, electronic equipment and readable storage medium | |
| CN108989889B (en) | Video playing amount prediction method and device and electronic equipment | |
| CN111062527B (en) | Video traffic collection prediction method and device | |
| CN110991476A (en) | Training method and device for decision classifier, recommendation method and device for audio and video, and storage medium | |
| CN108629047B (en) | Song list generation method and terminal equipment | |
| CN110933492B (en) | Method and device for predicting playing time | |
| CN112395496A (en) | Information recommendation method and device, electronic equipment and storage medium | |
| CN112579913A (en) | Video recommendation method, device, equipment and computer-readable storage medium | |
| CN113656681A (en) | Object evaluation method, device, equipment and storage medium | |
| CN113676770B (en) | Member rights prediction method, member rights prediction device, electronic equipment and storage medium | |
| CN110198460B (en) | Method and device for selecting media information, storage medium and electronic device | |
| CN113204699B (en) | Information recommendation method and device, electronic equipment and storage medium | |
| CN115794898B (en) | Financial information recommendation method and device, electronic equipment and storage medium | |
| CN110971973A (en) | Video pushing method and device and electronic equipment | |
| CN112804566A (en) | Program recommendation method, device and computer readable storage medium | |
| CN110569447A (en) | network resource recommendation method and device and storage medium | |
| KR20200145346A (en) | A program for the product or content recommendation service | |
| CN111669656B (en) | Method and device for determining wonderful degree of video clip | |
| CN113313511A (en) | Video traffic prediction method, device, electronic equipment and medium | |
| CN113836412A (en) | Information recommendation method and device, electronic equipment and storage medium | |
| CN113269259A (en) | Target information prediction method and device | |
| CN112115300A (en) | Text processing method and device, electronic equipment and readable 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 |