Satisfaction feature processing method and device, electronic equipment and storage medium

Abstract

The present disclosure provides a satisfaction feature processing method and device, electronic equipment and storage medium, and relates to the technical field of video recommendation. The specific implementation scheme is: based on the historical consumption behavior information of the video application, a first feature set, a second feature set and a third feature set of a reference user are obtained, the first feature set includes the consumption behavior features of the reference user in the video application within a preset time period before a reference time period; the second feature set includes the consumption behavior features of the reference user in the video application within the reference time period; the third feature set includes the retention features of the reference user in the video application within a future preset time period after the reference time period; based on the three feature sets obtained above, the satisfaction feature representing the satisfaction degree of the reference user to the resources of the video application is analyzed in a causal inference manner. The technology of the present disclosure can objectively, accurately and efficiently obtain the satisfaction feature.

Description

Technical Field

[0001] This disclosure relates to the field of computer technology, specifically to the field of video recommendation and other technical fields, and in particular to a method, apparatus, electronic device and storage medium for processing satisfaction features. Background Technology

[0002] In immersive video recommendation scenarios, videos play automatically without requiring explicit user clicks. The lack of explicit playback signals in such scenarios presents a significant challenge for video recommendation.

[0003] For example, in some scenarios, the duration of a user's playback can be used to characterize their satisfaction with the video. The longer the playback time, the higher the user's satisfaction with the video. Summary of the Invention

[0004] This disclosure provides a method, apparatus, electronic device, and storage medium for processing satisfaction features.

[0005] According to one aspect of this disclosure, a method for processing satisfaction features is provided, comprising:

[0006] Based on historical consumption behavior information of video applications, a first feature set of reference users is obtained. The first feature set includes the consumption behavior characteristics of reference users in video applications within a historical preset time period before the reference time period.

[0007] Based on historical consumption behavior information of video applications, a second feature set of reference users is obtained. The second feature set includes the consumption behavior characteristics of reference users in video applications within a reference time period.

[0008] Based on historical consumption behavior information of video applications, a third feature set of reference users is obtained. The third feature set includes the retention characteristics of reference users in video applications within a future preset time period after the reference time period.

[0009] Based on the first feature set, the second feature set, and the third feature set, a causal inference approach is used to analyze satisfaction features that can characterize the satisfaction level of reference users with the resources of the video application.

[0010] According to another aspect of this disclosure, a processing apparatus for satisfaction features is provided, comprising:

[0011] The first acquisition module is used to acquire a first feature set of a reference user based on the historical consumption behavior information of the video application. The first feature set includes the consumption behavior characteristics of the reference user in the video application within a historical preset time period before the reference time period.

[0012] The second acquisition module is used to acquire a second feature set of the reference user based on the historical consumption behavior information of the video application. The second feature set includes the consumption behavior characteristics of the reference user in the video application within a reference time period.

[0013] The third acquisition module is used to acquire a third feature set of reference users based on the historical consumption behavior information of video applications. The third feature set includes the retention features of reference users in video applications within a future preset time period after the reference time period.

[0014] The analysis module is used to analyze satisfaction features that can characterize the satisfaction level of reference users with the resources of the video application, based on the first feature set, the second feature set, and the third feature set, using a causal inference approach.

[0015] According to another aspect of this disclosure, an electronic device is provided, comprising:

[0016] At least one processor; and

[0017] A memory communicatively connected to the at least one processor; wherein,

[0018] The memory stores instructions that can be executed by the at least one processor to enable the at least one processor to perform the methods described above and any possible implementations.

[0019] According to another aspect of this disclosure, a non-transitory computer-readable storage medium is provided storing computer instructions for causing the computer to perform the methods described above and any possible implementation thereof.

[0020] According to another aspect of this disclosure, a computer program product is provided, comprising a computer program that, when executed by a processor, implements the aspects and any possible implementations described above.

[0021] According to the technology disclosed herein, satisfaction characteristics can be obtained objectively, accurately, and efficiently.

[0022] It should be understood that the description in this section is not intended to identify key or essential features of the embodiments of this disclosure, nor is it intended to limit the scope of this disclosure. Other features of this disclosure will become readily apparent from the following description. Attached Figure Description

[0023] The accompanying drawings are provided to better understand this solution and do not constitute a limitation of this disclosure. Wherein:

[0024] Figure 1This is a schematic diagram based on the first embodiment of the present disclosure;

[0025] Figure 2 This is a schematic diagram according to the second embodiment of the present disclosure;

[0026] Figure 3 This is a publicly available example table showing the importance of features for different user groups.

[0027] Figure 4 This is a schematic diagram according to the third embodiment of the present disclosure;

[0028] Figure 5 This is a schematic diagram according to the fourth embodiment of the present disclosure;

[0029] Figure 6 This is a block diagram of an electronic device used to implement the methods of the embodiments of this disclosure. Detailed Implementation

[0030] The exemplary embodiments of this disclosure are described below with reference to the accompanying drawings, including various details of the embodiments to aid understanding, and should be considered merely exemplary. Therefore, those skilled in the art will recognize that various changes and modifications can be made to the embodiments described herein without departing from the scope and spirit of this disclosure. Similarly, for clarity and brevity, descriptions of well-known functions and structures are omitted in the following description.

[0031] Obviously, the described embodiments are only some, not all, of the embodiments disclosed herein. All other embodiments obtained by those skilled in the art based on the embodiments of this disclosure without inventive effort are within the scope of protection of this disclosure.

[0032] It should be noted that the terminal devices involved in the embodiments of this disclosure may include, but are not limited to, smart devices such as mobile phones, personal digital assistants (PDAs), wireless handheld devices, and tablet computers; the display devices may include, but are not limited to, personal computers, televisions, and other devices with display functions.

[0033] Furthermore, the term "and / or" in this article is merely a description of the relationship between related objects, indicating that three relationships can exist. For example, A and / or B can represent: A existing alone, A and B existing simultaneously, or B existing alone. Additionally, the character " / " in this article generally indicates that the preceding and following related objects have an "or" relationship.

[0034] In existing technologies, video playback duration and physical video length are positively correlated, but this is subject to bias due to the influence of physical length. For example, videos can include different types, such as short videos, medium videos, and long videos. Among them, short videos have the shortest physical length, and long videos have the longest physical length. If existing technologies only use playback duration to represent user satisfaction with videos, it is highly unreasonable. For example, a preset duration threshold can be set; if a user's playback duration reaches this threshold, the user is considered satisfied with the video. However, if the preset duration threshold is greater than the physical length of a short video, it may be mistakenly assumed that even if the user completes the playback of a short video, the user is still considered dissatisfied. Therefore, this method is very unfriendly to short videos with short physical lengths. Thus, the existing technology of using playback duration as a feature of user satisfaction with videos is highly inaccurate. Based on this, this disclosure urgently needs to provide a more accurate feature to represent user satisfaction.

[0035] Figure 1 This is a schematic diagram based on the first embodiment of the present disclosure; as shown Figure 1 As shown in the figure, this embodiment provides a method for processing satisfaction features, which may specifically include the following steps:

[0036] S101. Based on the historical consumption behavior information of video applications, obtain the first feature set of reference users. The first feature set includes the consumption behavior characteristics of reference users in video applications within a historical preset time period before the reference time period.

[0037] S102. Based on the historical consumption behavior information of video applications, obtain a second feature set of reference users. The second feature set includes the consumption behavior characteristics of reference users in video applications within a reference time period.

[0038] S103. Based on the historical consumption behavior information of video applications, obtain the third feature set of reference users. The third feature set includes the retention characteristics of reference users in video applications within a future preset time period after the reference time period.

[0039] The execution subject of the satisfaction feature processing method in this embodiment can be a satisfaction feature processing device. The device can be an electronic entity or a software-integrated application. When in use, it runs on a computer device to process the satisfaction features.

[0040] The historical consumption behavior information of the video application in this embodiment may include the consumption behavior information of users in the video application at various time periods. For example, in this embodiment, in order to facilitate the determination of satisfaction features, taking any reference user as an example, based on the historical consumption behavior information of the video application, the first feature set, the second feature set, and the third feature set of the reference user are first obtained.

[0041] The first, second, and third feature sets respectively include different consumption behavior characteristics of the reference user over different time periods. For example, the first feature set includes the consumption behavior characteristics of the reference user in video applications within a historical preset time period prior to the reference time period. In this embodiment, the reference time period can be from 0:00 to 24:00 on a certain day, for example, it can be denoted as reference time period t. The historical preset time period can be a period of time before the reference time period t, for example, it can be from t-1 days to t-7 days, or other historical preset time periods can be used according to actual needs, which are not limited here.

[0042] The first feature set and the second feature set respectively include the behavioral characteristics of the reference user consuming videos in the video application within a historical preset time period before the reference time period or within the reference time period.

[0043] The third feature set includes the retention characteristics of the reference user in the video application within a future preset time period after the reference time period. This future preset time period can be from t+1 days to t+7 days, or other historical preset time periods can be used according to actual needs, without limitation here. The retention characteristics can identify whether the user retains relevant features.

[0044] S104. Based on the first feature set, the second feature set, and the third feature set, a causal inference approach is used to analyze the satisfaction features that can characterize the reference user's satisfaction with the resources of the video application.

[0045] Based on the three feature sets obtained above, namely the first feature set, the second feature set, and the third feature set, this embodiment can use causal inference to analyze the satisfaction features that can characterize the reference user's satisfaction with the resources in the video application.

[0046] In this embodiment, the number of reference users is not limited. Optionally, in practical applications, according to steps S101-S103, a first feature set, a second feature set, and a third feature set of multiple reference users can be obtained. Then, during the analysis in step S104, the first feature set, the second feature set, and the third feature set of multiple reference users can be referenced simultaneously, and a comprehensive analysis can be performed using causal inference, thereby obtaining satisfaction features that characterize the reference users' satisfaction with the video application's resources.

[0047] The satisfaction feature processing method in this embodiment obtains a first feature set, a second feature set, and a third feature set of reference users. Based on these three feature sets, it uses causal inference to analyze satisfaction features that characterize the reference users' satisfaction with the video application's resources. This method obtains satisfaction features objectively, accurately, and efficiently. Compared to existing technologies that use video playback duration to characterize satisfaction features, this method does not rely on the physical duration of the video, resulting in more accurate satisfaction features.

[0048] Figure 2 This is a schematic diagram based on the second embodiment of this disclosure; the method for processing satisfaction features in this embodiment is described above. Figure 1 Based on the technical solutions of the illustrated embodiments, the technical solutions of this disclosure will be described in further detail. For example... Figure 2 As shown, the method for processing satisfaction features in this embodiment may specifically include the following steps:

[0049] S201. Based on the historical consumption behavior information of video applications, obtain at least one of the following characteristics of the reference user in the video application during the historical preset time period before the reference time period: activity characteristics, resource playback characteristics, resource display characteristics, and interaction characteristics, as well as the basic attribute characteristics of the reference user.

[0050] Step S201 is as described above. Figure 1 This is a specific implementation of step S101 in the illustrated embodiment. Specifically, in this embodiment, the first feature set includes at least one of the following: the reference user's activity characteristics, resource playback characteristics, resource display characteristics, and interaction characteristics within a historical preset time period prior to the reference time period, along with the reference user's basic attribute characteristics. In practical applications, the more features included in the reference user's first feature set, the higher the accuracy of the subsequently determined satisfaction characteristics. Therefore, in step S201, preferably, all of the above-mentioned features are selected.

[0051] The basic attributes of the reference user may include at least one of the following: age, gender, occupation, education level, life stage, and user level. The user level may refer to the user's classification in the video application, such as new user, inactive user, and active user. Furthermore, active users may be further divided into multiple levels based on their activity level.

[0052] Optionally, in one embodiment of this disclosure, step S201, based on the historical consumption behavior information of the video application, obtains the activity characteristics of the reference user in the video application within a historically preset time period prior to the reference time period. Specifically, this may include: obtaining the active duration of the reference user in the video application within a historically preset time period prior to the reference time period based on the historical consumption behavior information of the video application; this active duration can also be considered as the total time the reference user spent watching videos online within the historically preset time period. Based on the active duration and the historically preset time period, an activity ratio is obtained; for example, the ratio of active duration to the historically preset time period can be taken as the activity ratio. The active duration and / or activity ratio are used as the activity characteristics of the reference user in the video application within a historically preset time period prior to the reference time period. The activity characteristics obtained in this way are very accurate and reasonable.

[0053] Optionally, in one embodiment of this disclosure, step S201, based on the historical consumption behavior information of the video application, obtains the resource playback characteristics of the reference user in the video application within a historical preset time period before the reference time period, which may specifically include at least one of the following:

[0054] (1) Based on the historical consumption behavior information of video applications, obtain the resource distribution amount corresponding to each preset duration in at least two preset durations when the reference user plays resources in the video application within the historical preset time period before the reference time period.

[0055] For example, the preset durations can be set to 3s, 5s, 10s, 20s, and 50s, respectively, corresponding to a distribution time of 3s, 5s, 10s, 20s, and 50s. In this embodiment, a 3s distribution corresponds to a video resource playback duration of 3 seconds or more. A 5s distribution corresponds to a video resource playback duration of 5 seconds or more. And so on, with a 50s distribution corresponding to a video resource playback duration of 50 seconds or more.

[0056] Correspondingly, the resource distribution volume for a reference user during a 3-second period of playback within a historical preset time period prior to the reference time period in a video application is the number of resources whose playback duration is greater than or equal to 3 seconds within that historical preset time period. Similarly, the resource distribution volume for a reference user during a 5-second period of playback within that historical preset time period is the number of resources whose playback duration is greater than or equal to 5 seconds within that historical preset time period. Likewise, the resource distribution volume for a reference user during a 50-second period of playback within that historical preset time period is the number of resources whose playback duration is greater than or equal to 50 seconds within that historical preset time period. Optionally, the resource playback characteristics may also include the resource distribution volume for a reference user during a historical preset time period prior to the reference time period, which may be equal to the total number of resources played by the reference user during that historical preset time period.

[0057] Based on the above-mentioned features, and according to the corresponding statistical methods, when the reference user plays resources in the video application within the historical preset time period before the reference time period, the distribution of the video resources played is statistically analyzed according to the resource distribution of each preset duration, and the corresponding features can be accurately obtained.

[0058] (2) Based on the historical consumption behavior information of video applications, obtain the unbiased distribution of resources corresponding to each preset quantile in at least two preset quantiles when the reference user plays resources in the video application within a historical preset time period before the reference time period.

[0059] Since the distribution amount of each preset duration resource is affected by the physical duration of the resource, in order to obtain more objective and accurate characteristics, this embodiment also sets at least two preset percentiles for unbiased resource distribution amounts. For example, 10%, 20%, 30%, ..., 80% percentiles can be preset. That is to say, the preset percentiles can be percentages greater than 0 and less than 100%.

[0060] In practice, all video resources played by the reference user within a historical preset time period prior to the reference time period can be first bucketed according to resource type. Different resource types correspond to different physical duration ranges. For example, based on the different physical durations of the resources, resource types can be divided into short videos, medium videos, and long videos. Specifically, resources with a physical duration less than or equal to a first preset duration threshold can be placed in the short video bucket; resources with a physical duration greater than the first preset duration threshold but less than a second preset duration threshold can be placed in the medium video bucket; and resources with a physical duration greater than the second duration threshold can be placed in the long video bucket. The second preset duration threshold is greater than the first preset duration threshold.

[0061] Within the short video, medium video, and long video buckets, multiple preset quotients are set, such as 10%, 20%, 30%, ..., 80%. Within each bucket, based on the minimum and maximum physical duration of the videos that can be included, a corresponding playback duration threshold is also pre-set for each preset quotient. If the playback duration of a video reaches the playback duration threshold corresponding to that preset quotient, then the video is considered to belong to the unbiased distribution corresponding to that preset quotient.

[0062] Based on the above, it can be concluded that within different buckets, the playback duration threshold corresponding to the same preset position is different. Within the same bucket, as the value of the preset position increases, the corresponding playback duration threshold also increases.

[0063] Based on the aforementioned division of each bucket into preset quantiles, we can first calculate the unbiased resource distribution volume corresponding to each preset quantile within each bucket when the reference user plays resources in the video application during a historical preset time period prior to the reference time period. Then, we sum up the unbiased resource distribution volumes corresponding to the same preset quantile within different buckets to obtain the unbiased resource distribution volume corresponding to that preset quantile when the reference user plays resources in the video application during a historical preset time period prior to the reference time period.

[0064] In this embodiment, by statistically analyzing the characteristics of the unbiased distribution of resources corresponding to each preset quantile, the influence of different physical durations can be eliminated, thereby improving the objectivity, accuracy, and rationality of feature construction.

[0065] (3) Based on the historical consumption behavior information of video applications, obtain the total playback time of resources by the reference user within a historical preset time period before the reference time period when playing resources in the video application; and

[0066] Specifically, by statistically analyzing the playback duration of each resource played by a reference user within a historical preset time period prior to the reference time period, and summing them up, the total playback duration of the corresponding resource can be obtained. The total playback duration characteristic obtained in this way is highly accurate and reasonable.

[0067] (4) Based on the historical consumption behavior information of video applications, obtain the corresponding resource completion volume of the reference user when playing resources in the video application within the historical preset time period before the reference time period.

[0068] Specifically, the number of resources that reference users completed playing within a historical preset time period before the reference time period is used as the corresponding resource completion count.

[0069] Optionally, in one embodiment of this disclosure, step S201, based on the historical consumption behavior information of the video application, obtains the resource display characteristics of the reference user in the video application within a historical preset time period before the reference time period, including: based on the historical consumption behavior information of the video application, obtaining the total number of resources displayed to the reference user in the video application within a historical preset time period before the reference time period.

[0070] In real-world applications, video applications can display more resources to a reference user, who may only choose to watch a portion of those resources. Therefore, the number of resources played by the reference user is less than the number of resources displayed to them within the same time period. However, the video application's historical consumption behavior information records the resources displayed to the reference user at each moment. Therefore, based on this historical consumption behavior information, the total number of resources displayed to the reference user within a pre-defined historical time period prior to the reference time can be calculated. This method of obtaining this feature is highly reasonable and effectively ensures the accuracy of the acquired features.

[0071] Optionally, in one embodiment of this disclosure, step S201, based on the historical consumption behavior information of the video application, to obtain the interaction characteristics of the reference user in the video application within a historical preset time period before the reference time period, may include: based on the historical consumption behavior information of the video application, to obtain the number of resources corresponding to at least one of the following, commenting, liking, collecting and sharing operations of the reference user in the video application within a historical preset time period before the reference time period.

[0072] Specifically, each user, while using a video application, can follow video creators, comment on videos, like videos, favorite videos, or share videos. These actions are recorded in the user's historical video application usage history. Therefore, based on this historical usage information, it's possible to statistically determine the number of videos a user followed, commented on, liked, favorited, and shared within a pre-defined time period prior to a reference timeframe. This method of constructing features is highly reasonable and effectively ensures the accuracy of the acquired features.

[0073] In this embodiment, the features in the first feature set constructed through the above step S201 are very rich and comprehensive. When used, these features can be fully eliminated from their influence on the third feature set, which is conducive to obtaining the satisfaction features more accurately.

[0074] S202. Based on the historical consumption behavior information of video applications, obtain at least one of the following: the resource playback characteristics and resource display characteristics of the reference user in the video application within a reference time period.

[0075] In this embodiment, the second feature set includes at least one of the following: the reference user's resource playback features within the reference time period, and the resource display features. Similarly, in practical applications, the more features included in the reference user's second feature set, the higher the accuracy of the subsequently determined satisfaction features. Therefore, in step S202, preferably, all of the above-mentioned features are selected.

[0076] For example, in one embodiment of this disclosure, step S202, based on the historical consumption behavior information of the video application, obtaining the resource playback characteristics of the reference user in the video application within a reference time period, may include at least one of the following steps:

[0077] (a1) Based on the historical consumption behavior information of video applications, obtain the resource distribution amount corresponding to each preset duration in at least two preset durations when the reference user plays resources in the video application within a reference time period.

[0078] (b1) Based on the historical consumption behavior information of video applications, obtain the unbiased distribution of resources corresponding to each preset quantile in at least two preset quantiles when a reference user plays resources in a video application within a reference time period.

[0079] (c1) Based on historical consumption behavior information of video applications, obtain the total playback duration of resources for a reference user within a reference time period when playing resources in the video application; and

[0080] (d1) Based on the historical consumption behavior information of video applications, obtain the resource completion volume of reference users when playing resources in video applications within a reference time period.

[0081] The specific implementation of steps (a1)-(d1) in this embodiment can be referred to steps (1)-(4) above, and will not be repeated here.

[0082] For example, in one embodiment of this disclosure, step S202, obtaining the resource display characteristics of a reference user in a video application within a reference time period based on the historical consumption behavior information of the video application, may include: obtaining the total number of resources displayed to the reference user in the video application within a reference time period based on the historical consumption behavior information of the video application.

[0083] In this embodiment, the features in the second feature set constructed through the above step S202 are very rich and comprehensive. When used, the correlation between each feature in the second feature set and the third feature set can be fully analyzed, thereby obtaining the satisfaction features more accurately.

[0084] S203. Based on the historical consumption behavior information of video applications, obtain at least one of the following characteristics: whether the reference user is retained in each of the next two future time periods within a future preset time period after the reference time period, and the duration of user retention within the future preset time period.

[0085] In this embodiment, the retention features in the third feature set include at least one of the following: whether the reference user is retained within a future preset time period after the reference time period, within each of at least two future time periods, and the duration of user retention within the future preset time period. Similarly, in practical applications, the more features included in the reference user's third feature set, the higher the accuracy of the subsequently determined satisfaction features. Therefore, in step S203, preferably, all of the above-mentioned features are selected.

[0086] For example, if the reference time period is day t, the future preset time period can be from day t+1 to day t+7. The future time period can be day t+1, from day t+1 to day t+5, from day t+1 to day t+7, etc.

[0087] For example, taking a future time period that only includes day t+1 and days t+1 to t+7, the characteristics for user retention can include: whether the user is retained on day t+1, whether the user is retained from day t+1 to t+7, and the duration of user retention from day t+1 to t+7. The duration of user retention from day t+1 to t+7 can be counted in days; if a user uses the video application on a given day, they are considered retained. If a user uses the video application for 5 days out of 7 days, then the corresponding user retention duration is 5 days. This example only uses three characteristics for user retention; in practical applications, one, two, or more characteristics can be selected based on requirements.

[0088] In this embodiment, the retention features set according to step S203 above are very reasonable and accurate, and can effectively characterize whether a user is retained.

[0089] S204. Using the first feature set as a confounding factor, the second feature set as an explanatory variable, and the third feature set as the explained variable, the influence of the first feature set on the third feature set is eliminated by using causal inference. The feature most relevant to the third feature set is obtained from the second feature set and used as the satisfaction feature.

[0090] For example, the specific implementation of step S204 may include the following steps:

[0091] (a2) Using the first feature set as a confounding factor, the second feature set as an explanatory variable, and the third feature set as the explained variable, a double machine learning (DML) model, a causal forest algorithm, or a survival analysis method is used to eliminate the influence of the first feature set on the third feature set and to analyze the correlation between each feature in the second feature set and each feature in the third feature set.

[0092] The DML model in this embodiment is a pre-trained model, and it does not need to be trained again in this embodiment.

[0093] For example, when using a DML model for analysis, features from the confounding factors, explanatory variables, and explained variables can all be input into the model. Based on causal analysis, the model can eliminate the influence of features in the first feature set on features in the third feature set, and analyze and output the correlation between each feature in the second feature set and each feature in the third feature set. Specifically, it can output the correlation between each feature in the second feature set in step S202 and each retained feature in step S203.

[0094] (b2) Based on the correlation between each feature in the second feature set and each feature in the third feature set, and the preset weights of each feature in the third feature set, obtain the comprehensive correlation between each feature in the second feature set and the third feature set;

[0095] For example, after analyzing the correlation between each feature in the second feature set and each feature in the third feature set, the overall correlation between that feature in the second feature set and the third feature set can be obtained by weighted summation based on the preset weights of each feature in the third feature set. The preset weights of each feature in the third feature set can be set according to the actual scenario requirements and are not limited here.

[0096] For example, for feature A in the second feature set, if the third feature set includes three features: feature 1, feature 2 and feature 3, with preset weights of a1, b1 and c1 respectively, and the correlation between feature A and feature 1 is equal to A1, the correlation between feature A and feature 2 is equal to B1, and the correlation between feature A and feature 3 is equal to C1, then the comprehensive correlation between feature A and the third feature set can be equal to: A1*a1+B1*b1+C1*c1.

[0097] Using the above method, the overall correlation between each feature in the second feature set and the third feature set can be calculated.

[0098] In practical applications, if the third feature set contains only one feature, the correlation between each feature in the second feature set and that feature in the third feature set is the comprehensive correlation.

[0099] (c2) Based on the comprehensive correlation between each feature in the second feature set and the third feature set, the feature most relevant to the third feature set is obtained from the second feature set and used as the satisfaction feature.

[0100] Specifically, the comprehensive relevance of each feature in the second feature set to the third feature set can be sorted in order of magnitude, and the feature in the second feature set corresponding to the highest comprehensive relevance can be used as the satisfaction feature.

[0101] In this embodiment, the satisfaction feature selection is based on features within a reference time period. Features from previous preset time periods are removed to mitigate their impact on retention characteristics in future preset time periods. The system then selects features from the reference time period that are most correlated with retention characteristics in the future preset time period. A higher probability of user retention in the future preset time period indicates a higher level of user satisfaction, as the feature most correlated with retention characteristics in the reference time period and future preset time period can identify that level of satisfaction. Therefore, in this embodiment, the feature most correlated with the third feature set is obtained from the second feature set and used as the satisfaction feature to identify user satisfaction within the reference time period.

[0102] Using the method described in this embodiment, the selected satisfaction characteristics are very objective, reasonable, and highly accurate.

[0103] In this embodiment, when determining the satisfaction characteristics, the selected reference user can be all users in the video application. Alternatively, depending on the needs of the actual scenario, satisfaction characteristics of various types of users can be filtered.

[0104] For example, the satisfaction characteristics of new users on the first day can be determined. The corresponding reference time period could be the day a new user first uses the video application, with the reference users being all new users who use the application on that day. However, since new users on the first day do not have historical preset time periods, there are no features specific to those periods, meaning there are no features in the first feature set. However, the retention characteristics of new users on the first day can be tracked for future preset time periods. Therefore, a satisfaction characteristic can be determined from the features of the new user on the day they first use the video application, thus identifying the satisfaction level of the new user on their first day of use.

[0105] For example, the satisfaction characteristics of new users within 7 days can be determined. The corresponding reference time period can be selected as day t+7, and the reference users are all new users who first use the video application from day t to day t+7. The historical preset time period is from day t to day t+6. Following the method described above, the characteristics of the first feature set of the historical preset time period for the reference users can be obtained. Furthermore, the retention characteristics of these new users within the future preset time period, from day t+8 to day t+14, can be tracked. A satisfaction characteristic can then be determined from the characteristics of the new users on day t+7 to identify their satisfaction level.

[0106] For example, Figure 3 This is an example table showing the importance of characteristics for different user groups, provided in this public disclosure. For example... Figure 3 As shown, columns 1 and 2 analyze all users of the video application, obtaining the comprehensive correlation between each feature in the second feature set and the third feature set. Columns 3 and 4 analyze new users of the video application within one day, obtaining the comprehensive correlation between each feature in the second feature set and the third feature set. Columns 5 and 6 analyze new users of the video application within seven days, obtaining the comprehensive correlation between each feature in the second feature set and the third feature set.

[0107] according to Figure 3 As shown, for all users, the unbiased distribution volume of resources at the 80th percentile and the distribution volume of 50 seconds both have a high overall correlation with the third feature set, and can be used as user satisfaction features to indicate user satisfaction with the video. For new users within 1 day and new users within 7 days, the distribution volume of 50 seconds also has a high overall correlation with the third feature set, and can be used as user satisfaction features to indicate user satisfaction with the video.

[0108] S205. Based on the determined satisfaction characteristics, make resource recommendations.

[0109] For example, in this embodiment, step S205 can be implemented in the following two ways:

[0110] The first implementation method may include the following steps:

[0111] (a3) Based on predetermined satisfaction characteristics, select multiple played resources from the historical playback information of multiple playback users;

[0112] For example, if the satisfaction characteristic is a 50-second distribution time, multiple played resources from multiple users can be selected from the historical playback information of the resource, and each played resource satisfies the 50-second distribution time.

[0113] (b3) The resource recommendation model is trained using each playback user and the corresponding multiple played resources among multiple playback users as positive samples;

[0114] During training, users who do not meet the satisfaction characteristics and their corresponding played resources can be filtered from the historical playback information of the resources as negative samples. For example, played resources can be filtered based on features other than the 50-second distribution time, such as 3-second distribution or unbiased distribution of resources at the 10th percentile.

[0115] Specifically, during training, features of users playing content and features of played resources can be extracted from positive samples and input into the resource recommendation model. The resource recommendation model then predicts the recommendation probability based on the input features. The true recommendation probability for positive samples is 1. Based on the predicted and true recommendation probabilities, the parameters of the resource recommendation model are adjusted to make the predicted recommendation probability consistent with the true recommendation probability.

[0116] Correspondingly, for negative samples, features of the playing users and the played resources can be extracted separately and input into the resource recommendation model. The resource recommendation model then predicts the recommendation probability based on the input features. The true recommendation probability for negative samples is 0. Based on the predicted and true recommendation probabilities, the parameters of the resource recommendation model are adjusted to make the predicted recommendation probability consistent with the true recommendation probability.

[0117] It should be noted that, in this embodiment, the characteristics of the playing user can refer to the basic attribute characteristics of the playing user, including at least one of age, gender, occupation, education level, life stage, and user level. The characteristics of the played resource can include at least one of the type, tag, language, and physical duration of the played resource.

[0118] By using multiple sets of positive and negative samples, the resource recommendation model is trained in the manner described above until the model converges, the parameters of the resource recommendation model are determined, and thus the resource recommendation model is finalized.

[0119] (c3) Based on the trained resource recommendation model, recommend resources from the resource library to the user to be recommended in the video application.

[0120] When recommending resources using the pre-trained resource recommendation model, the features of the user to be recommended and the features of each resource to be recommended in the resource library are first extracted. Then, the resource recommendation model is used to predict the recommendation probability between the user and each resource. The resource with the highest recommendation probability is then selected for recommendation, which can effectively improve the accuracy of resource recommendation.

[0121] The second implementation method may include the following steps:

[0122] (a4) Based on the determined satisfaction characteristics, obtain at least one played resource played by the user to be recommended within a preset time period before the current moment from the user's historical consumption behavior information in the video application.

[0123] (b4) Based on at least one played resource, recommend resources from the resource library to the user to be recommended in the video application.

[0124] For example, if the satisfaction metric is a 50-second distribution time, at least one played resource that meets this satisfaction metric can be selected from the historical consumption behavior information of the user to be recommended within a preset time period prior to the current moment. This preset time period can be 1 day, 2 days, or 1 week, etc., and is not limited here.

[0125] In this embodiment, the features of each played resource and the features of each resource to be recommended in the resource library can be extracted first; then the correlation between the features of each played resource and the features of each resource to be recommended can be calculated; and the resource to be recommended with the highest correlation in the resource library can be selected for recommendation.

[0126] The satisfaction feature processing method in this embodiment, by adopting the above-described technical solution, can simultaneously eliminate the bias between user-side and resource-side features, objectively and accurately obtaining satisfaction features; and further, it can perform more accurate and efficient resource recommendations based on satisfaction features. The method of this embodiment can be applied to real-world large-scale immersive video recommendation scenarios. Experimental verification shows that video recommendations based on these satisfaction features can achieve significant gains in user time and distribution revenue.

[0127] Figure 4 This is a schematic diagram based on the third embodiment of this disclosure; as shown Figure 4 As shown, this embodiment provides a processing device 400 for satisfaction features, including:

[0128] The first acquisition module 401 is used to acquire a first feature set of a reference user based on the historical consumption behavior information of the video application. The first feature set includes the consumption behavior characteristics of the reference user in the video application within a historical preset time period before the reference time period.

[0129] The second acquisition module 402 is used to acquire a second feature set of the reference user based on the historical consumption behavior information of the video application. The second feature set includes the consumption behavior characteristics of the reference user in the video application within a reference time period.

[0130] The third acquisition module 403 is used to acquire a third feature set of the reference user based on the historical consumption behavior information of the video application. The third feature set includes the retention features of the reference user in the video application within a future preset time period after the reference time period.

[0131] Analysis module 404 is used to analyze satisfaction features that can characterize the satisfaction level of reference users with the resources of the video application based on the first feature set, the second feature set and the third feature set, using a causal inference approach.

[0132] The satisfaction feature processing device 400 in this embodiment can realize the implementation principle and technical effect of satisfaction feature processing by adopting the above-mentioned modules. It is the same as the implementation of the above-mentioned related method embodiments. For details, please refer to the description of the above-mentioned related embodiments, which will not be repeated here.

[0133] Figure 5 This is a schematic diagram based on the fourth embodiment of the present disclosure; as shown Figure 5 As shown, this embodiment provides a processing device 500 for satisfaction features, including... Figure 4 The modules with the same name and function shown are: first acquisition module 501, second acquisition module 502, third acquisition module 503, and analysis module 504.

[0134] In one embodiment of this disclosure, the first acquisition module 501 is configured to:

[0135] Based on the historical consumption behavior information of the video application, at least one of the following characteristics of the reference user in the video application during a historical preset time period before the reference time period: activity characteristics, resource playback characteristics, resource display characteristics, and interaction characteristics.

[0136] Further optionally, in one embodiment of this disclosure, the first acquisition module 501 is configured to:

[0137] Based on the historical consumption behavior information of the video application, the active duration of the reference user in the video application within a historical preset time period before the reference time period is obtained;

[0138] The activity ratio is obtained based on the active duration and the historical preset time period;

[0139] The active duration and / or the active ratio are used as the activity characteristics of the reference user in the video application within a historical preset time period prior to the reference time period.

[0140] Further optionally, in one embodiment of this disclosure, the first acquisition module 501 is configured to perform at least one of the following operations:

[0141] Based on the historical consumption behavior information of the video application, the resource distribution amount corresponding to each preset duration in at least two preset durations is obtained when the reference user plays resources in the video application within a historical preset time period before the reference time period.

[0142] Based on the historical consumption behavior information of the video application, obtain the unbiased distribution of resources corresponding to each preset quantile in at least two preset quantiles when the reference user plays resources in the video application within a historical preset time period before the reference time period.

[0143] Based on the historical consumption behavior information of the video application, the total playback time of resources by the reference user within a historical preset time period prior to the reference time period is obtained; and

[0144] Based on the historical consumption behavior information of the video application, the number of completed plays of resources by the reference user in the video application within a historical preset time period before the reference time period is obtained.

[0145] Further optionally, in one embodiment of this disclosure, the first acquisition module 501 is configured to:

[0146] Based on the historical consumption behavior information of the video application, the total number of resources displayed to the reference user in the video application within a historical preset time period prior to the reference time period is obtained.

[0147] Further optionally, in one embodiment of this disclosure, the first acquisition module 501 is configured to:

[0148] Based on the historical consumption behavior information of the video application, the number of resources corresponding to at least one of the following, commenting, liking, collecting, and sharing operations performed by the reference user within a historical preset time period before the reference time period is obtained.

[0149] Further optionally, in one embodiment of this disclosure, the first feature set of the reference user further includes: the basic attribute features of the reference user.

[0150] Further optionally, in one embodiment of this disclosure, the second acquisition module 502 is configured to:

[0151] Based on the historical consumption behavior information of the video application, at least one of the following is obtained: the resource playback characteristics and resource display characteristics of the reference user in the video application within a reference time period.

[0152] Further optionally, in one embodiment of this disclosure, the third acquisition module 503 is configured to:

[0153] Based on the historical consumption behavior information of the video application, at least one of the following is obtained: whether the reference user is retained in each of at least two future time periods within a future preset time period after the reference time period, and the duration of user retention within the future preset time period.

[0154] Further optionally, in one embodiment of this disclosure, the analysis module 504 is configured to:

[0155] Using the first feature set as a confounding factor, the second feature set as an explanatory variable, and the third feature set as the explained variable, a causal inference approach is adopted to eliminate the influence of the first feature set on the third feature set. The feature most relevant to the third feature set is then obtained from the second feature set and used as the satisfaction feature.

[0156] Further optionally, in one embodiment of this disclosure, the analysis module 504 is configured to:

[0157] Using the first feature set as a confounding factor, the second feature set as an explanatory variable, and the third feature set as the explained variable, a dual machine learning model, a causal forest algorithm, or a survival analysis method is employed to eliminate the influence of the first feature set on the third feature set, and to analyze the correlation between each feature in the second feature set and each feature in the third feature set.

[0158] Based on the correlation between each feature in the second feature set and each feature in the third feature set, and the preset weight of each feature in the third feature set, the comprehensive correlation between each feature in the second feature set and the third feature set is obtained;

[0159] Based on the comprehensive correlation between each feature in the second feature set and the third feature set, the feature most relevant to the third feature set is obtained from the second feature set and used as the satisfaction feature.

[0160] Further optional, such as Figure 5 As shown, in one embodiment of this disclosure, the satisfaction feature processing device 500 further includes:

[0161] The resource recommendation module 505 is used to recommend resources based on the determined satisfaction characteristics.

[0162] Further optionally, in one embodiment of this disclosure, the resource recommendation module 505 is used for:

[0163] Based on predetermined satisfaction characteristics, multiple played resources from multiple users are selected from the historical playback information of the resources.

[0164] The resource recommendation model is trained using each of the multiple playback users and the corresponding multiple played resources as positive samples.

[0165] Based on the trained resource recommendation model, resources from the resource library are recommended to the user to be recommended in the video application.

[0166] Further optionally, in one embodiment of this disclosure, the resource recommendation module 505 is used for:

[0167] Based on the determined satisfaction characteristics, at least one played resource played by the user to be recommended within a preset time period before the current moment is obtained from the user's historical consumption behavior information in the video application.

[0168] Based on the at least one played resource, resources from the resource library are recommended to the user to be recommended in the video application.

[0169] The satisfaction feature processing device 500 of this embodiment can realize the implementation principle and technical effect of satisfaction feature processing by adopting the above-mentioned modules. It is the same as the implementation of the above-mentioned related method embodiments. For details, please refer to the description of the above-mentioned related embodiments, which will not be repeated here.

[0170] The acquisition, storage, and application of user personal information involved in the technical solution disclosed herein comply with the provisions of relevant laws and regulations and do not violate public order and good morals.

[0171] According to embodiments of this disclosure, this disclosure also provides an electronic device, a readable storage medium, and a computer program product.

[0172] Figure 6 A schematic block diagram of an example electronic device 600 that can be used to implement embodiments of the present disclosure is shown. The electronic device is intended to represent various forms of digital computers, such as laptop computers, desktop computers, workstations, personal digital assistants, servers, blade servers, mainframe computers, and other suitable computers. The electronic device may also represent various forms of mobile devices, such as personal digital processors, cellular phones, smartphones, wearable devices, and other similar computing devices. The components shown herein, their connections and relationships, and their functions are merely illustrative and are not intended to limit the implementation of the present disclosure described and / or claimed herein.

[0173] like Figure 6 As shown, device 600 includes a computing unit 601, which can perform various appropriate actions and processes based on a computer program stored in read-only memory (ROM) 602 or a computer program loaded from storage unit 608 into random access memory (RAM) 603. RAM 603 may also store various programs and data required for the operation of device 600. The computing unit 601, ROM 602, and RAM 603 are interconnected via bus 604. Input / output (I / O) interface 605 is also connected to bus 604.

[0174] Multiple components in device 600 are connected to I / O interface 605, including: input unit 606, such as keyboard, mouse, etc.; output unit 607, such as various types of monitors, speakers, etc.; storage unit 608, such as disk, optical disk, etc.; and communication unit 609, such as network card, modem, wireless transceiver, etc. Communication unit 609 allows device 600 to exchange information / data with other devices through computer networks such as the Internet and / or various telecommunications networks.

[0175] The computing unit 601 can be a variety of general-purpose and / or special-purpose processing components with processing and computing capabilities. Some examples of the computing unit 601 include, but are not limited to, a central processing unit (CPU), a graphics processing unit (GPU), various special-purpose artificial intelligence (AI) computing chips, various computing units running machine learning model algorithms, a digital signal processor (DSP), and any suitable processor, controller, microcontroller, etc. The computing unit 601 performs the various methods and processes described above, such as the methods of this disclosure. For example, in some embodiments, the methods of this disclosure may be implemented as a computer software program tangibly contained in a machine-readable medium, such as storage unit 608. In some embodiments, part or all of the computer program may be loaded and / or installed on device 600 via ROM 602 and / or communication unit 609. When the computer program is loaded into RAM 603 and executed by the computing unit 601, one or more steps of the methods of this disclosure described above may be performed. Alternatively, in other embodiments, the computing unit 601 may be configured to perform the methods of this disclosure by any other suitable means (e.g., by means of firmware).

[0176] Various embodiments of the systems and techniques described above herein can be implemented in digital electronic circuit systems, integrated circuit systems, field-programmable gate arrays (FPGAs), application-specific integrated circuits (ASICs), application-specific standard products (ASSPs), systems-on-a-chip (SoCs), complex programmable logic devices (CPLDs), computer hardware, firmware, software, and / or combinations thereof. These various embodiments may include implementations in one or more computer programs that can be executed and / or interpreted on a programmable system including at least one programmable processor, which may be a dedicated or general-purpose programmable processor, capable of receiving data and instructions from a storage system, at least one input device, and at least one output device, and transmitting data and instructions to the storage system, the at least one input device, and the at least one output device.

[0177] The program code used to implement the methods of this disclosure may be written in any combination of one or more programming languages. This program code may be provided to a processor or controller of a general-purpose computer, special-purpose computer, or other programmable data processing apparatus, such that when executed by the processor or controller, the program code causes the functions / operations specified in the flowcharts and / or block diagrams to be implemented. The program code may be executed entirely on a machine, partially on a machine, as a standalone software package partially on a machine and partially on a remote machine, or entirely on a remote machine or server.

[0178] In the context of this disclosure, a machine-readable medium can be a tangible medium that may contain or store a program for use by or in conjunction with an instruction execution system, apparatus, or device. A machine-readable medium can be a machine-readable signal medium or a machine-readable storage medium. A machine-readable medium can be, but is not limited to, electronic, magnetic, optical, electromagnetic, infrared, or semiconductor systems, apparatus, or devices, or any suitable combination of the foregoing. More specific examples of machine-readable storage media include electrical connections based on one or more wires, portable computer disks, hard disks, random access memory (RAM), read-only memory (ROM), erasable programmable read-only memory (EPROM or flash memory), optical fiber, portable compact disk read-only memory (CD-ROM), optical storage devices, magnetic storage devices, or any suitable combination of the foregoing.

[0179] To provide interaction with a user, the systems and techniques described herein can be implemented on a computer having: a display device for displaying information to the user (e.g., a CRT (cathode ray tube) or LCD (liquid crystal display) monitor); and a keyboard and pointing device (e.g., a mouse or trackball) through which the user provides input to the computer. Other types of devices can also be used to provide interaction with the user; for example, feedback provided to the user can be any form of sensory feedback (e.g., visual feedback, auditory feedback, or tactile feedback); and input from the user can be received in any form (including sound input, voice input, or tactile input).

[0180] The systems and technologies described herein can be implemented in computing systems that include backend components (e.g., as a data server), or computing systems that include middleware components (e.g., an application server), or computing systems that include frontend components (e.g., a user computer with a graphical user interface or web browser through which a user can interact with embodiments of the systems and technologies described herein), or any combination of such backend, middleware, or frontend components. The components of the system can be interconnected via digital data communication of any form or medium (e.g., a communication network). Examples of communication networks include local area networks (LANs), wide area networks (WANs), and the Internet.

[0181] Computer systems can include clients and servers. Clients and servers are generally located far apart and typically interact via communication networks. Client-server relationships are created by computer programs running on the respective computers and having a client-server relationship with each other. Servers can be cloud servers, servers in distributed systems, or servers incorporating blockchain technology.

[0182] It should be understood that the various forms of processes shown above can be used to rearrange, add, or delete steps. For example, the steps described in this disclosure can be executed in parallel, sequentially, or in different orders, as long as the desired result of the technical solution disclosed in this disclosure can be achieved, and this is not limited herein.

[0183] The specific embodiments described above do not constitute a limitation on the scope of protection of this disclosure. Those skilled in the art should understand that various modifications, combinations, sub-combinations, and substitutions can be made according to design requirements and other factors. Any modifications, equivalent substitutions, and improvements made within the spirit and principles of this disclosure should be included within the scope of protection of this disclosure.

Claims

1. A method for processing satisfaction characteristics, comprising: Based on historical consumption behavior information of video applications, a first feature set of reference users is obtained. The first feature set includes the consumption behavior characteristics of reference users in video applications within a historical preset time period before the reference time period. Based on historical consumption behavior information of video applications, a second feature set of reference users is obtained. The second feature set includes the consumption behavior characteristics of reference users in video applications within a reference time period. Based on historical consumption behavior information of video applications, a third feature set of reference users is obtained. The third feature set includes the retention characteristics of reference users in video applications within a future preset time period after the reference time period. Based on the first feature set, the second feature set, and the third feature set, a causal inference approach is used to analyze satisfaction features that can characterize the satisfaction level of reference users with the resources of the video application.

2. The method of claim 1, wherein, Based on historical consumption behavior information from video applications, a first set of features for reference users is obtained, including: Based on the historical consumption behavior information of the video application, at least one of the following characteristics of the reference user in the video application during a historical preset time period before the reference time period: activity characteristics, resource playback characteristics, resource display characteristics, and interaction characteristics.

3. The method according to claim 2, wherein, Based on the historical consumption behavior information of the video application, the activity characteristics of the reference user in the video application within a historical preset time period prior to the reference time period are obtained, including: Based on the historical consumption behavior information of the video application, the active duration of the reference user in the video application within a historical preset time period before the reference time period is obtained; The activity ratio is obtained based on the active duration and the historical preset time period; The active duration and / or the active ratio are used as the activity characteristics of the reference user in the video application within a historical preset time period prior to the reference time period.

4. The method according to claim 2, wherein, Based on the historical consumption behavior information of the video application, the resource playback characteristics of the reference user in the video application within a historical preset time period before the reference time period are obtained, including at least one of the following: Based on the historical consumption behavior information of the video application, the resource distribution amount corresponding to each preset duration in at least two preset durations is obtained when the reference user plays resources in the video application within a historical preset time period before the reference time period. Based on the historical consumption behavior information of the video application, obtain the unbiased distribution of resources corresponding to each preset quantile in at least two preset quantiles when the reference user plays resources in the video application within a historical preset time period before the reference time period. Based on the historical consumption behavior information of the video application, the total playback time of the resources of the reference user in the video application during the historical preset time period before the reference time period is obtained. as well as Based on the historical consumption behavior information of the video application, the number of completed plays of resources by the reference user in the video application within a historical preset time period before the reference time period is obtained.

5. The method according to claim 2, wherein, Based on the historical consumption behavior information of the video application, the resource display characteristics of the reference user in the video application within a historical preset time period prior to the reference time period are obtained, including: Based on the historical consumption behavior information of the video application, the total number of resources displayed to the reference user in the video application within a historical preset time period prior to the reference time period is obtained.

6. The method according to claim 2, wherein, Based on the historical consumption behavior information of the video application, the interaction characteristics of the reference user in the video application within a historical preset time period prior to the reference time period are obtained, including: Based on the historical consumption behavior information of the video application, the number of resources corresponding to at least one of the following, commenting, liking, collecting, and sharing operations performed by the reference user within a historical preset time period before the reference time period is obtained.

7. The method according to claim 2, wherein, The first feature set of the reference user also includes: the basic attribute features of the reference user.

8. The method according to claim 1, wherein, Based on historical consumption behavior information from video applications, a second set of features for reference users is obtained, including: Based on the historical consumption behavior information of the video application, at least one of the following is obtained: the resource playback characteristics and resource display characteristics of the reference user in the video application within a reference time period.

9. The method according to claim 1, wherein, Based on historical consumption behavior information from video applications, a third set of features for reference users is obtained, including: Based on the historical consumption behavior information of the video application, at least one of the following is obtained: whether the reference user is retained in each of at least two future time periods within a future preset time period after the reference time period, and the duration of user retention within the future preset time period.

10. The method according to claim 1, wherein, Based on the first feature set, the second feature set, and the third feature set, a causal inference approach is used to analyze satisfaction features that characterize the reference user's satisfaction with the resources of the video application, including: Using the first feature set as a confounding factor, the second feature set as an explanatory variable, and the third feature set as the explained variable, a causal inference approach is adopted to eliminate the influence of the first feature set on the third feature set. The feature most relevant to the third feature set is then obtained from the second feature set and used as the satisfaction feature.

11. The method according to claim 10, wherein, Using the first feature set as a confounding factor, the second feature set as an explanatory variable, and the third feature set as the explained variable, a causal inference approach is employed to eliminate the influence of the first feature set on the third feature set. The feature most relevant to the third feature set is then obtained from the second feature set as the satisfaction feature, including: Using the first feature set as a confounding factor, the second feature set as an explanatory variable, and the third feature set as the explained variable, a dual machine learning model, a causal forest algorithm, or a survival analysis method is employed to eliminate the influence of the first feature set on the third feature set, and to analyze the correlation between each feature in the second feature set and each feature in the third feature set. Based on the correlation between each feature in the second feature set and each feature in the third feature set, and the preset weight of each feature in the third feature set, the comprehensive correlation between each feature in the second feature set and the third feature set is obtained; Based on the comprehensive correlation between each feature in the second feature set and the third feature set, the feature most relevant to the third feature set is obtained from the second feature set and used as the satisfaction feature.

12. The method according to any one of claims 1-11, wherein, Based on the first feature set, the second feature set, and the third feature set, and after analyzing satisfaction features that characterize the satisfaction level of reference users with the resources of the video application using causal inference, the method further includes: Based on the determined satisfaction characteristics, resource recommendations are made.

13. The method according to claim 12, wherein, Based on the determined satisfaction characteristics, resource recommendations are made, including: Based on predetermined satisfaction characteristics, multiple played resources from multiple users are selected from the historical playback information of the resources. The resource recommendation model is trained using each of the multiple playback users and the corresponding multiple played resources as positive samples. Based on the trained resource recommendation model, resources from the resource library are recommended to the user to be recommended in the video application.

14. The method according to claim 13, wherein, Based on the determined satisfaction characteristics, resource recommendations are made, including: Based on the determined satisfaction characteristics, at least one played resource played by the user to be recommended within a preset time period before the current moment is obtained from the user's historical consumption behavior information in the video application. Based on the at least one played resource, resources from the resource library are recommended to the user to be recommended in the video application.

15. A processing apparatus for satisfaction characteristics, comprising: The first acquisition module is used to acquire a first feature set of a reference user based on the historical consumption behavior information of the video application. The first feature set includes the consumption behavior characteristics of the reference user in the video application within a historical preset time period before the reference time period. The second acquisition module is used to acquire a second feature set of the reference user based on the historical consumption behavior information of the video application. The second feature set includes the consumption behavior characteristics of the reference user in the video application within a reference time period. The third acquisition module is used to acquire a third feature set of reference users based on the historical consumption behavior information of video applications. The third feature set includes the retention features of reference users in video applications within a future preset time period after the reference time period. The analysis module is used to analyze satisfaction features that can characterize the satisfaction level of reference users with the resources of the video application, based on the first feature set, the second feature set, and the third feature set, using a causal inference approach.

16. The apparatus according to claim 15, wherein, The first acquisition module is used for: Based on the historical consumption behavior information of the video application, at least one of the following characteristics of the reference user in the video application during a historical preset time period before the reference time period: activity characteristics, resource playback characteristics, resource display characteristics, and interaction characteristics.

17. The apparatus according to claim 16, wherein, The first acquisition module is used for: Based on the historical consumption behavior information of the video application, the active duration of the reference user in the video application within a historical preset time period before the reference time period is obtained; The activity ratio is obtained based on the active duration and the historical preset time period; The active duration and / or the active ratio are used as the activity characteristics of the reference user in the video application within a historical preset time period prior to the reference time period.

18. The apparatus according to claim 16, wherein, The first acquisition module is configured to perform at least one of the following operations: Based on the historical consumption behavior information of the video application, the resource distribution amount corresponding to each preset duration in at least two preset durations is obtained when the reference user plays resources in the video application within a historical preset time period before the reference time period. Based on the historical consumption behavior information of the video application, obtain the unbiased distribution of resources corresponding to each preset quantile in at least two preset quantiles when the reference user plays resources in the video application within a historical preset time period before the reference time period. Based on the historical consumption behavior information of the video application, the total playback time of the resources of the reference user in the video application during the historical preset time period before the reference time period is obtained. as well as Based on the historical consumption behavior information of the video application, the number of completed plays of resources by the reference user in the video application within a historical preset time period before the reference time period is obtained.

19. The apparatus according to claim 16, wherein, The first acquisition module is used for: Based on the historical consumption behavior information of the video application, the total number of resources displayed to the reference user in the video application within a historical preset time period prior to the reference time period is obtained.

20. The apparatus according to claim 16, wherein, The first acquisition module is used for: Based on the historical consumption behavior information of the video application, the number of resources corresponding to at least one of the following, commenting, liking, collecting, and sharing operations performed by the reference user within a historical preset time period before the reference time period is obtained.

21. The apparatus according to claim 16, wherein, The first feature set of the reference user also includes: the basic attribute features of the reference user.

22. The apparatus according to claim 15, wherein, The second acquisition module is used for: Based on the historical consumption behavior information of the video application, at least one of the following is obtained: the resource playback characteristics and resource display characteristics of the reference user in the video application within a reference time period.

23. The apparatus according to claim 15, wherein, The third acquisition module is used for: Based on the historical consumption behavior information of the video application, at least one of the following is obtained: whether the reference user is retained in each of at least two future time periods within a future preset time period after the reference time period, and the duration of user retention within the future preset time period.

24. The apparatus according to claim 15, wherein, The analysis module is used for: Using the first feature set as a confounding factor, the second feature set as an explanatory variable, and the third feature set as the explained variable, a causal inference approach is adopted to eliminate the influence of the first feature set on the third feature set. The feature most relevant to the third feature set is then obtained from the second feature set and used as the satisfaction feature.

25. The apparatus according to claim 24, wherein, The analysis module is used for: Using the first feature set as a confounding factor, the second feature set as an explanatory variable, and the third feature set as the explained variable, a dual machine learning model, a causal forest algorithm, or a survival analysis method is employed to eliminate the influence of the first feature set on the third feature set, and to analyze the correlation between each feature in the second feature set and each feature in the third feature set. Based on the correlation between each feature in the second feature set and each feature in the third feature set, and the preset weight of each feature in the third feature set, the comprehensive correlation between each feature in the second feature set and the third feature set is obtained; Based on the comprehensive correlation between each feature in the second feature set and the third feature set, the feature most relevant to the third feature set is obtained from the second feature set and used as the satisfaction feature.

26. The apparatus according to any one of claims 15-25, wherein, The device further includes: The resource recommendation module is used to recommend resources based on the determined satisfaction characteristics.

27. The apparatus according to claim 26, wherein, The resource recommendation module is used for: Based on predetermined satisfaction characteristics, multiple played resources from multiple users are selected from the historical playback information of the resources. The resource recommendation model is trained using each of the multiple playback users and the corresponding multiple played resources as positive samples. Based on the trained resource recommendation model, resources from the resource library are recommended to the user to be recommended in the video application.

28. The apparatus according to claim 26, wherein, The resource recommendation module is used for: Based on the determined satisfaction characteristics, at least one played resource played by the user to be recommended within a preset time period before the current moment is obtained from the user's historical consumption behavior information in the video application. Based on the at least one played resource, resources from the resource library are recommended to the user to be recommended in the video application.

29. An electronic device comprising: At least one processor; as well as A memory communicatively connected to the at least one processor; wherein, The memory stores instructions that can be executed by the at least one processor to enable the at least one processor to perform the method of any one of claims 1-14.

30. A non-transitory computer-readable storage medium storing computer instructions, wherein, The computer instructions are used to cause the computer to perform the method according to any one of claims 1-14.

31. A computer program product comprising a computer program that, when executed by a processor, implements the method according to any one of claims 1-14.

Images