Data processing method, device and equipment and readable storage medium

Abstract

The invention discloses a data processing method, device and equipment and a readable storage medium, and the method comprises the steps: generating interest weight data of an object for each media domain type based on the interaction operation of the object for media data under at least two media domain types in a target time period in log data of the object; performing interest matching in the candidate media data set based on the interest weight data to obtain a media push list for the object; if the key interaction operation of the object is detected, performing normalization processing on interest weight data corresponding to the at least two media domain types to obtain weight proportions corresponding to the at least two media domain types; and adjusting the weight ratio based on the key interaction operation, and updating the media push list according to the adjusted weight ratio. By adopting the method and the device, the media push list can be updated in real time, the matching degree of the media push list and the film watching interest of the object is improved, and the flow consumption of the object is reduced.

Description

Technical Field

[0001] The present application relates to the field of computer technology, and in particular to a data processing method, apparatus, device, and readable storage medium. Background Art

[0002] By analyzing the content characteristics of media data in a user's viewing history, media platforms can push media data that matches the user's viewing interests to the user. However, users' viewing interests may change over time, and media platforms struggle to track changes in their viewing interests in real time through their viewing history. This makes it difficult for the pushed media data to meet the user's current viewing interests. Users may frequently switch between media data, resulting in significant data consumption and a poor viewing experience. Summary of the Invention

[0003] The embodiments of the present application provide a data processing method, apparatus, device, and readable storage medium, which can update the media push list in real time, improve the matching degree between the media push list and the subject's viewing interests, and reduce the subject's traffic consumption.

[0004] On the one hand, an embodiment of the present application provides a data processing method, including:

[0005] Get log data for an object;

[0006] Based on the interactive operations of the subject on media data of at least two media domain types within a target time period in the log data, generating interest weight data of the subject for each media domain type;

[0007] Based on the interest weight data corresponding to at least two media field types, interest matching is performed in the candidate media data set to obtain a media push list for the object; the media field type corresponding to the media field type with the largest interest weight data in the media push list has the largest proportion of media data;

[0008] If a key interactive operation of the object on media data under at least two media field types is detected, normalizing the interest weight data corresponding to the at least two media field types to obtain weight ratios corresponding to the at least two media field types;

[0009] At least two weight ratios are adjusted based on the key interaction operation, and the media push list is updated according to the at least two adjusted weight ratios.

[0010] Wherein, at least two media domain types include media domain type S i, i is a positive integer; based on the interactive operations of the object on media data of at least two media domain types within the target time period in the log data, the object's interest weight data for each media domain type is generated, including:

[0011] Based on the objects in the log data within the target time period for the media domain type S i generating a first dimension parameter based on a behavioral relationship between the interactive behavior of the media data under the target time period and the interactive behavior of the object with respect to historical media data sets under at least two media domain types within the target time period;

[0012] Based on the objects in the log data within the target time period for the media domain type S i The second dimension parameter is generated by the relationship between the browsing time of the media data under the target time period and the browsing time of the historical media data set of the object in the target time period;

[0013] Based on the objects in the log data within the target time period for the media domain type S i The frequency relationship between the interaction frequency of the media data under the target time period and the interaction frequency of the object with respect to the historical media data set in the target time period is used to generate a third dimension parameter;

[0014] According to the first dimension parameter, the second dimension parameter and the third dimension parameter, an object is generated for the media domain type S i Interest weight data.

[0015] Among them, the first dimension parameters include behavior impact parameters; based on the objects in the log data in the target time period for the media field type S i The first dimension parameters are generated based on the behavioral relationship between the interactive behavior of the media data under the target time period and the interactive behavior of the object for the historical media data sets under at least two media domain types within the target time period, including:

[0016] Get the objects in the log data for the media field type S within the target time period i The first time decay factor, first behavior type relevance and first behavior interaction depth of the media data under the first time decay factor is used to characterize the object for the media field type S i The first behavior type relevance is used to characterize the media data involved in the object's interactive behavior and the media domain type S i The first behavior interaction depth is used to characterize the influence of the object's interaction behavior;

[0017] Based on the first time attenuation factor, the first behavior type relevance and the first behavior interaction depth, a target for the media domain type S is generated. i First interactive behavior data of the media data under;

[0018] Obtaining a second time decay factor, a second behavior type relevance, and a second behavior interaction depth of an object in the log data for a historical media dataset under at least two media field types within a target time period;

[0019] generating second interactive behavior data for at least two media domain types based on a second time decay factor, a second behavior type relevance, and a second behavior interaction depth;

[0020] A ratio of the first interactive behavior data to the second interactive behavior data is determined as a behavior influence parameter.

[0021] The first dimension parameter includes the intensity dimension parameter; based on the object in the log data in the target time period for the media field type S i The first dimension parameters are generated based on the behavioral relationship between the interactive behavior of the media data under the target time period and the interactive behavior of the object for the historical media data sets under at least two media domain types within the target time period, including:

[0022] Get the objects in the log data for the media field type S within the target time period i a plurality of first single interaction intensity data of the media data under the image, summing the plurality of first single interaction intensity data to obtain first interaction intensity data; the first single interaction intensity data is used to characterize the influence degree of the interactive behavior of the object, and different interactive behaviors correspond to different first single interaction intensity data;

[0023] Acquire multiple second single interaction intensity data of the object in the log data with respect to the historical media dataset within the target time period, and sum the multiple second single interaction intensity data to obtain second interaction intensity data;

[0024] A ratio of the first interaction intensity data to the second interaction intensity data is determined as an intensity dimension parameter.

[0025] Among them, based on the objects in the log data within the target time period for the media field type S i The second dimension parameters are generated by the relationship between the browsing time of the media data under the target time period and the browsing time of the historical media data set of the object in the target time period, including:

[0026] Get the objects in the log data for the media field type S within the target time period i multiple first single browsing durations of the media data under the above conditions, summing the multiple first single browsing durations to obtain a first browsing duration;

[0027] Acquire multiple second single browsing durations of the object in the log data for the historical media dataset within the target time period, and sum the multiple second single browsing durations to obtain a second browsing duration;

[0028] The ratio of the first browsing time to the second browsing time is determined as the second dimension parameter.

[0029] Among them, based on the objects in the log data within the target time period for the media field type S i The frequency relationship between the interaction frequency of the media data under the target time period and the interaction frequency of the object for the historical media data set in the target time period is used to generate the third dimension parameters, including:

[0030] Get the objects in the log data for the media field type S within the target time period i The first number of interactions of the media data under the target time period, and the second number of interactions of the object in the log data with respect to the historical media data set within the target time period;

[0031] The ratio of the first interaction number to the second interaction number is determined as the third dimension parameter.

[0032] The object's interest weight data for each media field type is generated based on the first dimension parameter, the second dimension parameter, and the third dimension parameter, including:

[0033] Obtain at least two adjustment parameters;

[0034] Based on at least two adjustment parameters, the first dimension parameter, the second dimension parameter and the third dimension parameter are weighted and summed to obtain the object for the media field type S i interest weight data; at least two adjustment parameters are used to determine the degree of influence of the first dimension parameter, the second dimension parameter and the third dimension parameter on the interest weight data.

[0035] The first dimension parameters include a behavior impact parameter related to the real-time behavior and an intensity dimension parameter for characterizing the interaction intensity; and the method further includes:

[0036] If it is detected that the data volume of the log data is less than the data volume threshold, then reducing the adjustment parameters associated with the behavior impact parameter and the second dimension parameter respectively among the at least two adjustment parameters, and simultaneously increasing the adjustment parameters associated with the third dimension parameter and the intensity dimension parameter respectively among the at least two adjustment parameters;

[0037] If it is detected that the data volume of the log data is greater than or equal to the data volume threshold, the adjustment parameters of at least two adjustment parameters respectively associated with the behavior impact parameter and the second dimension parameter are increased, and the adjustment parameters of at least two adjustment parameters respectively associated with the third dimension parameter and the intensity dimension parameter are reduced simultaneously.

[0038] At least two media domain types include a target media domain type; based on interest weight data corresponding to the at least two media domain types, interest matching is performed in the candidate media data set to obtain a media push list for the target, including:

[0039] If the interest weight data related to the target media field type corresponding to the object is greater than the weight threshold, the interest weight data related to the target media field type corresponding to at least two historical objects are determined as target interest weight data, and the historical objects corresponding to the target interest weight data greater than or equal to the weight threshold are determined as candidate objects;

[0040] Obtaining a first historical interaction behavior of the candidate object with respect to the candidate media dataset, and obtaining a second historical interaction behavior of the object in the log data;

[0041] determining an object similarity between the candidate object and the object based on the first historical interaction behavior and the second historical interaction behavior;

[0042] If the object similarity is greater than the similarity threshold, a media push list for the object is generated based on the media data that the candidate object has interacted with and the object has not interacted with.

[0043] At least two media domain types include a target media domain type; based on interest weight data corresponding to the at least two media domain types, interest matching is performed in the candidate media data set to obtain a media push list for the target, including:

[0044] If the interest weight data corresponding to the object and related to the target media domain type is greater than the weight threshold, then feature fusion is performed on the media data under the target media domain type that the object has interacted with to obtain a target feature vector;

[0045] Perform feature extraction on each media data under the target media domain type in the candidate media data set to obtain a set of candidate feature vectors;

[0046] The target feature vector is similar to each candidate feature vector in the candidate feature vector set, and a media push list for the object is generated based on the media data corresponding to the candidate feature vectors whose similarity is greater than or equal to the similarity threshold.

[0047] The step of adjusting at least two weight ratios based on the key interaction operation and updating the media push list according to the at least two adjusted weight ratios includes:

[0048] If the key interactive operation is a forward interactive operation, the weight ratio of the key media domain type is increased, and the weight ratios other than the weight ratio of the key media domain type in at least two weight ratios are simultaneously reduced; the key media domain type refers to the media domain type to which the media data operated by the key interactive operation belongs;

[0049] Based on at least two adjusted weight ratios, the amount of media data belonging to the key media domain type in the media push list is increased, and the amount of media data belonging to other media domain types in the media push list is simultaneously reduced; the other media domain types refer to media domain types other than the key media domain type in the at least two media domain types;

[0050] If the key interaction operation is a negative interaction operation, the weight ratio of the key media field type will be reduced, and the weight ratios of other types will be increased simultaneously;

[0051] Based on at least two adjusted weight ratios, the amount of media data belonging to the key media domain type in the media push list is reduced, and the amount of media data belonging to other media domain types in the media push list is increased simultaneously.

[0052] At least two weight ratios are adjusted based on key interaction operations, including:

[0053] If the key interactive operation is a trigger operation for a feedback control, obtaining a ratio adjustment value based on the feedback intent indicated by the feedback control, performing a first ratio adjustment on the weight ratio of the key media domain type in the at least two weight ratios based on the ratio adjustment value, and performing a second ratio adjustment on the other weight ratios in the at least two weight ratios except the weight ratio of the key media domain type; the key media domain type refers to the media domain type to which the media data operated by the key interactive operation belongs;

[0054] If the key interactive operation is a sharing operation on a social platform, the platform priority of the shared social platform is obtained, and the weight ratio of the key media field type is increased based on the proportion mapped by the platform priority, while the other weight ratios are reduced simultaneously.

[0055] The updating of the media push list according to at least two adjusted weight ratios includes:

[0056] Interest matching is performed in the candidate media data set according to at least two adjusted weight ratios to obtain an updated media push list for the object, and the media push list is replaced with the updated media push list.

[0057] Among them, also include:

[0058] Obtaining playback network status of target media data for at least two participating objects within a geographical area; the at least two participating objects include an object, and the media push list includes the target media data;

[0059] If it is detected that the number of participating objects whose playback network status is a network freeze state is greater than the object quantity threshold, the definition recommended version for the target media data in the media push list is adjusted.

[0060] In one aspect, an embodiment of the present application provides a data processing device, including:

[0061] The transceiver module is used to obtain log data for the object;

[0062] A data generation module, configured to generate interest weight data of the object for each media field type based on the object's interactive operations on media data of at least two media field types within a target time period in the log data;

[0063] A list generation module is configured to perform interest matching in a candidate media data set based on interest weight data corresponding to at least two media domain types, and obtain a media push list for the target; in the media push list, the media domain type corresponding to the data with the largest interest weight has the largest proportion of media data;

[0064] A ratio generation module is configured to, upon detecting a key interaction operation of an object on media data of at least two media domain types, normalize the interest weight data corresponding to the at least two media domain types to obtain weight ratios corresponding to the at least two media domain types;

[0065] The list updating module is used to adjust at least two weight ratios based on the key interactive operation, and update the media push list according to the at least two adjusted weight ratios.

[0066] In one possible implementation, the at least two media domain types include a media domain type S i , i is a positive integer; the data generation module is used to generate the object's interest weight data for each media field type based on the object's interactive operations on media data under at least two media field types within a target time period in the log data, specifically for performing the following operations:

[0067] Based on the objects in the log data within the target time period for the media domain type S i generating a first dimension parameter based on a behavioral relationship between the interactive behavior of the media data under the target time period and the interactive behavior of the object with respect to historical media data sets under at least two media domain types within the target time period;

[0068] Based on the objects in the log data within the target time period for the media domain type S i The second dimension parameter is generated by the relationship between the browsing time of the media data under the target time period and the browsing time of the historical media data set of the object in the target time period;

[0069] Based on the objects in the log data within the target time period for the media domain type S i The frequency relationship between the interaction frequency of the media data under the target time period and the interaction frequency of the object with respect to the historical media data set in the target time period is used to generate a third dimension parameter;

[0070] According to the first dimension parameter, the second dimension parameter and the third dimension parameter, an object is generated for the media domain type S i Interest weight data.

[0071] In a possible implementation, the first dimension parameter includes a behavior impact parameter; the data generation module is used to generate a behavior impact parameter based on the behavior of the object in the log data within the target time period for the media domain type S i The behavioral relationship between the interactive behavior of the media data under the target time period and the interactive behavior of the object for the historical media data sets under at least two media field types during the target time period is used to generate the first dimension parameter, specifically for performing the following operations:

[0072] Get the objects in the log data for the media field type S within the target time period i The first time decay factor, first behavior type relevance and first behavior interaction depth of the media data under the first time decay factor is used to characterize the object for the media field type S i The first behavior type relevance is used to characterize the media data involved in the object's interactive behavior and the media domain type S i The first behavior interaction depth is used to characterize the influence of the object's interaction behavior;

[0073] Based on the first time attenuation factor, the first behavior type relevance and the first behavior interaction depth, a target for the media domain type S is generated. i First interactive behavior data of the media data under;

[0074] Obtaining a second time decay factor, a second behavior type relevance, and a second behavior interaction depth of an object in the log data for a historical media dataset under at least two media field types within a target time period;

[0075] generating second interactive behavior data for at least two media domain types based on a second time decay factor, a second behavior type relevance, and a second behavior interaction depth;

[0076] A ratio of the first interactive behavior data to the second interactive behavior data is determined as a behavior influence parameter.

[0077] In a possible implementation, the first dimension parameter includes an intensity dimension parameter; the data generation module is configured to generate an intensity dimension parameter based on the object in the log data within the target time period for the media domain type S. i The behavioral relationship between the interactive behavior of the media data under the target time period and the interactive behavior of the object for the historical media data sets under at least two media field types during the target time period is used to generate the first dimension parameter, specifically for performing the following operations:

[0078] Get the objects in the log data for the media field type S within the target time period i a plurality of first single interaction intensity data of the media data under the image, summing the plurality of first single interaction intensity data to obtain first interaction intensity data; the first single interaction intensity data is used to characterize the influence degree of the interactive behavior of the object, and different interactive behaviors correspond to different first single interaction intensity data;

[0079] Acquire multiple second single interaction intensity data of the object in the log data with respect to the historical media dataset within the target time period, and sum the multiple second single interaction intensity data to obtain second interaction intensity data;

[0080] A ratio of the first interaction intensity data to the second interaction intensity data is determined as an intensity dimension parameter.

[0081] In a possible implementation, the data generation module is used to generate data for the media domain type S based on the objects in the log data within the target time period. i The duration relationship between the browsing time of the media data under the target time period and the browsing time of the historical media data set of the object in the target time period is used to generate the second dimension parameter, specifically for performing the following operations:

[0082] Get the objects in the log data for the media field type S within the target time period i multiple first single browsing durations of the media data under the above conditions, summing the multiple first single browsing durations to obtain a first browsing duration;

[0083] Acquire multiple second single browsing durations of the object in the log data for the historical media dataset within the target time period, and sum the multiple second single browsing durations to obtain a second browsing duration;

[0084] The ratio of the first browsing time to the second browsing time is determined as the second dimension parameter.

[0085] In a possible implementation, the data generation module is used to generate data for the media domain type S based on the objects in the log data within the target time period.i The frequency relationship between the interaction frequency of the media data under the target time period and the interaction frequency of the object for the historical media data set in the target time period is used to generate the third dimension parameter, specifically for the following operations:

[0086] Get the objects in the log data for the media field type S within the target time period i The first number of interactions of the media data under the target time period, and the second number of interactions of the object in the log data with respect to the historical media data set within the target time period;

[0087] The ratio of the first interaction number to the second interaction number is determined as the third dimension parameter.

[0088] In a possible implementation, when the data generation module is used to generate the object's interest weight data for each media field type based on the first dimension parameter, the second dimension parameter, and the third dimension parameter, it is specifically used to perform the following operations:

[0089] Obtain at least two adjustment parameters;

[0090] Based on at least two adjustment parameters, the first dimension parameter, the second dimension parameter and the third dimension parameter are weighted and summed to obtain the object for the media field type S i interest weight data; at least two adjustment parameters are used to determine the degree of influence of the first dimension parameter, the second dimension parameter and the third dimension parameter on the interest weight data.

[0091] In a possible implementation, the first dimension parameter includes a behavior impact parameter related to the real-time behavior and an intensity dimension parameter for characterizing interaction intensity; and the data generation module is further configured to perform the following operations:

[0092] If it is detected that the data volume of the log data is less than the data volume threshold, then reducing the adjustment parameters associated with the behavior impact parameter and the second dimension parameter respectively among the at least two adjustment parameters, and simultaneously increasing the adjustment parameters associated with the third dimension parameter and the intensity dimension parameter respectively among the at least two adjustment parameters;

[0093] If it is detected that the data volume of the log data is greater than or equal to the data volume threshold, the adjustment parameters of at least two adjustment parameters respectively associated with the behavior impact parameter and the second dimension parameter are increased, and the adjustment parameters of at least two adjustment parameters respectively associated with the third dimension parameter and the intensity dimension parameter are reduced simultaneously.

[0094] In one possible implementation, at least two media domain types include a target media domain type; and the list generation module is configured to perform interest matching in a candidate media data set based on interest weight data corresponding to the at least two media domain types, and to obtain a media push list for the target, specifically performing the following operations:

[0095] If the interest weight data related to the target media field type corresponding to the object is greater than the weight threshold, the interest weight data related to the target media field type corresponding to at least two historical objects are determined as target interest weight data, and the historical objects corresponding to the target interest weight data greater than or equal to the weight threshold are determined as candidate objects;

[0096] Obtaining a first historical interaction behavior of the candidate object with respect to the candidate media dataset, and obtaining a second historical interaction behavior of the object in the log data;

[0097] determining an object similarity between the candidate object and the object based on the first historical interaction behavior and the second historical interaction behavior;

[0098] If the object similarity is greater than the similarity threshold, a media push list for the object is generated based on the media data that the candidate object has interacted with and the object has not interacted with.

[0099] In one possible implementation, at least two media domain types include a target media domain type; and the list generation module is configured to perform interest matching in a candidate media data set based on interest weight data corresponding to the at least two media domain types, and to obtain a media push list for the target, specifically performing the following operations:

[0100] If the interest weight data corresponding to the object and related to the target media domain type is greater than the weight threshold, then feature fusion is performed on the media data under the target media domain type that the object has interacted with to obtain a target feature vector;

[0101] Perform feature extraction on each media data under the target media domain type in the candidate media data set to obtain a set of candidate feature vectors;

[0102] The target feature vector is similar to each candidate feature vector in the candidate feature vector set, and a media push list for the object is generated based on the media data corresponding to the candidate feature vectors whose similarity is greater than or equal to the similarity threshold.

[0103] In one possible implementation, the list updating module is configured to adjust at least two weight ratios based on the key interaction operation, and when updating the media push list according to the at least two adjusted weight ratios, specifically to perform the following operations:

[0104] If the key interactive operation is a forward interactive operation, the weight ratio of the key media domain type is increased, and the weight ratios other than the weight ratio of the key media domain type in at least two weight ratios are simultaneously reduced; the key media domain type refers to the media domain type to which the media data operated by the key interactive operation belongs;

[0105] Based on at least two adjusted weight ratios, the amount of media data belonging to the key media domain type in the media push list is increased, and the amount of media data belonging to other media domain types in the media push list is simultaneously reduced; the other media domain types refer to media domain types other than the key media domain type in the at least two media domain types;

[0106] If the key interaction operation is a negative interaction operation, the weight ratio of the key media field type will be reduced, and the weight ratios of other types will be increased simultaneously;

[0107] Based on at least two adjusted weight ratios, the amount of media data belonging to the key media domain type in the media push list is reduced, and the amount of media data belonging to other media domain types in the media push list is increased simultaneously.

[0108] In a possible implementation, when the list updating module is used to adjust at least two weight ratios based on the key interaction operation, it is specifically used to perform the following operations:

[0109] If the key interactive operation is a trigger operation for a feedback control, obtaining a ratio adjustment value based on the feedback intent indicated by the feedback control, performing a first ratio adjustment on the weight ratio of the key media domain type in the at least two weight ratios based on the ratio adjustment value, and performing a second ratio adjustment on the other weight ratios in the at least two weight ratios except the weight ratio of the key media domain type; the key media domain type refers to the media domain type to which the media data operated by the key interactive operation belongs;

[0110] If the key interactive operation is a sharing operation on a social platform, the platform priority of the shared social platform is obtained, and the weight ratio of the key media field type is increased based on the proportion mapped by the platform priority, while the other weight ratios are reduced simultaneously.

[0111] In a possible implementation, when the list updating module is used to update the media push list according to at least two adjusted weight ratios, it is specifically used to perform the following operations:

[0112] Interest matching is performed in the candidate media data set according to at least two adjusted weight ratios to obtain an updated media push list for the object, and the media push list is replaced with the updated media push list.

[0113] In a possible implementation, the list updating module is further configured to perform the following operations:

[0114] Obtaining playback network status of target media data for at least two participating objects within a geographical area; the at least two participating objects include an object, and the media push list includes the target media data;

[0115] If it is detected that the number of participating objects whose playback network status is a network freeze state is greater than the object quantity threshold, the definition recommended version for the target media data in the media push list is adjusted.

[0116] An embodiment of the present application provides a computer device, including: a processor, a memory, and a network interface;

[0117] The processor is connected to the memory and the network interface, wherein the network interface is used to provide data communication functions, and the memory is used to store computer programs. When the computer program is executed by the processor, the computer device executes the method provided in the embodiment of the present application.

[0118] On the one hand, an embodiment of the present application provides a computer-readable storage medium, which stores a computer program. The computer program is suitable for being loaded and executed by a processor so that a computer device having the processor executes the method provided by the embodiment of the present application.

[0119] In one aspect, an embodiment of the present application provides a computer program product, comprising a computer program stored in a computer-readable storage medium. A processor of a computer device reads the computer program from the computer-readable storage medium and executes the computer program, causing the computer device to perform the method provided in the embodiment of the present application.

[0120] The embodiment of the present application can generate the object's interest weight data for each media field type by obtaining log data for the object, and based on the object's interactive operations on media data under at least two media field types in the log data within the target time period, interest matching is performed in the candidate media data set according to the interest weight data corresponding to the at least two media field types, and a media push list for the object can be obtained. When a key interactive operation of the object for media data under at least two media field types is detected, the interest weight data corresponding to the at least two media field types are normalized to obtain the weight ratios corresponding to the at least two media field types. After adjusting the at least two weight ratios through the key interactive operation, the media push list can be updated according to the at least two adjusted weight ratios. Since the interest weight data is generated based on the object's interactive operations on media data under at least two media field types within the target time period, when the object's interactive operations change, the interest weight data will also change accordingly. It can be seen that the interest weight data can accurately reflect the object's interest habits, so media data that meets the object's viewing interests can be more accurately selected based on the interest weight data. Through the key interactive operations of the object, the weight ratios corresponding to at least two media field types can be adjusted in real time, and the media push list can be updated according to the at least two adjusted weight ratios. Therefore, the changes in the object's viewing interests can be obtained in real time, and then the media push list can be updated in real time based on the changes in the object's viewing interests, so as to ensure that the real-time updated media push list is sufficiently matched with the object's viewing interests, so as to improve the object's viewing experience, and also reduce the traffic consumption caused by the object's frequent switching of media data due to lack of interest. BRIEF DESCRIPTION OF THE DRAWINGS

[0121] In order to more clearly illustrate the embodiments of the present application or the technical solutions in the prior art, the following briefly introduces the drawings required for use in the embodiments or the description of the prior art. Obviously, the drawings described below are only some embodiments of the present application. For ordinary technicians in this field, other drawings can be obtained based on these drawings without any creative work.

[0122] Figure 1 This is a schematic diagram of a network architecture provided by an embodiment of the present application;

[0123] Figure 2 This is a data processing scenario provided by the embodiment of the present application. Figure 1 ;

[0124] Figure 3 This is a flow diagram of a data processing method provided in an embodiment of the present application. Figure 1 ;

[0125] Figure 4 This is a flow diagram of a data processing method provided in an embodiment of the present application. Figure 2 ;

[0126] Figure 5 This is a data processing scenario provided by the embodiment of the present application. Figure 2 ;

[0127] Figure 6 This is a data processing scenario provided by the embodiment of the present application. Figure 3 ;

[0128] Figure 7 is a structural diagram of a data processing device provided in an embodiment of the present application;

[0129] Figure 8 It is a structural diagram of a computer device provided in an embodiment of the present application. DETAILED DESCRIPTION

[0130] The following will be combined with the drawings in the embodiments of this application to clearly and completely describe the technical solutions in the embodiments of this application. Obviously, the embodiments described are only part of the embodiments of this application, not all of the embodiments. Based on the embodiments in this application, all other embodiments obtained by ordinary technicians in this field without making creative efforts are within the scope of protection of this application.

[0131] It is understandable that in the specific implementation of this application, the user data involved, when the following embodiments of this application are applied to specific products or technologies, needs to obtain user permission or consent, and the collection, use and processing of relevant data need to comply with relevant laws, regulations and standards of the relevant regions.

[0132] Among them, if it is necessary to collect object (such as user, etc.) data in this application, a prompt interface or pop-up window will be displayed before or during the collection. The prompt interface or pop-up window is used to remind the user that certain data is currently being collected. Only after the user confirms the prompt interface or pop-up window, the relevant steps for data acquisition will be started, otherwise the process will end. Moreover, the acquired user data will be used in reasonable and legal scenarios or purposes. Optionally, in some scenarios where user data needs to be used but has not been authorized by the user, authorization can be requested from the user, and the user data can be used when the authorization is passed.

[0133] See Figure 1 , Figure 1 This is a network architecture diagram provided by an embodiment of the present application. Figure 1As shown, the network structure may include a service server 100 and a terminal device 200. The service server 100 may be in communication with the terminal device 200. The communication connection is not limited to a connection method and may be directly or indirectly connected via a wired communication method, or directly or indirectly connected via a wireless communication method, or in other ways, which are not limited in this application.

[0134] It should be understood that Figure 1 The terminal device 200 shown in the figure may be installed with a business application. When the business application is running in the terminal device 200, it can be used together with the above-mentioned Figure 1 Data is exchanged between the business servers 100 shown. The business application can be an application client with image, video and other data information functions, such as a game application, video editing application, social application, instant messaging application, live broadcast application, short video application, video application, music application, shopping application, novel application, payment application, browser, etc. The application client can be an independent client or an embedded sub-client integrated in a client (such as an instant messaging client, social client, video client, etc.), which is not limited here.

[0135] like Figure 1 As shown, the terminal device 200 can send interactive information such as the object's operation information and browsing information to the business server 100. The business server 100 can generate log data based on the interactive information, and generate interest weight data for the object for each media field type based on the interactive operation of the object for media data of at least two media field types in the log data, and a single interest weight data can reflect the change in the object's interest in the interactive process of a certain media field type. Through interest matching in the candidate media data set through the interest weight data (that is, the media data under the media field type with a relatively large interest weight data can be prioritized and focused on in the candidate media data set), a media push list can be generated. Since the interest weight data can reflect the object's interest level in a certain media field type, the number of media data under the media field type corresponding to the largest interest weight data in the media push list is the largest, that is, the largest proportion.

[0136] The terminal device 200 mentioned above may be an electronic device, including but not limited to a mobile phone, a tablet computer, a desktop computer, a laptop computer, a PDA, an in-vehicle device, an augmented reality / virtual reality (AR / VR) device, a helmet display, a smart TV, a wearable device, a smart speaker, a digital camera, a camera and other mobile internet devices (MID) with network access capabilities, or terminal devices in scenarios such as trains, ships, and flights. The business server 100 mentioned above may be an independent physical server, or a server cluster or distributed system composed of multiple physical servers, or a cloud server that provides basic cloud computing services such as cloud services, cloud databases, cloud computing, cloud functions, cloud storage, network services, cloud communications, middleware services, domain name services, security services, vehicle-road collaboration, content delivery networks (CDNs), and big data and artificial intelligence platforms.

[0137] The business server 100 can send the media push list to the terminal device 200, and the object can perform interactive operations (such as playing in sequence) on the media data in the media push list in the terminal device 200. Since the interest weight data is generated based on the object's interactive operations on media data under at least two media field types within the target time period, when the object's interactive operations change, the interest weight data will also change accordingly. It can be seen that the interest weight data can accurately reflect the object's interest habits, and therefore media data that meets the object's viewing interests can be more accurately selected based on the interest weight data.

[0138] See Figure 2 , Figure 2 This is a schematic diagram of a data processing scenario provided by an embodiment of the present application. Figure 2As shown, the business server 100 can collect the log data of the object in the terminal device 200 by means of multi-terminal burial points, that is, collect log data once at a certain time interval (such as 1 minute), and integrate and store the collected log data. Then, before calculating the interest weight data, the business server 100 can obtain the log data within the target time period (such as the past hour), such as obtaining the object's interactive operations (such as clicks, comments, and other interactive behaviors) for media data (such as movies and TV shows, etc.) under at least two media field types (such as comedy, science fiction drama, etc.) within the target time period. The business server 100 calculates the weights of various interactive operations of the object for media data under at least two media field types within the target time period based on the log data, and obtains the object's interest weight data for each media field type. For example, the interest weight data for comedy can be 0.1, the interest weight data for science fiction drama can be 0.7, etc., indicating that the object's interest in science fiction drama is higher than that in comedy.

[0139] By performing interest matching on the interest weight data corresponding to at least two media field types and the candidate media data in the candidate media data set, an interest matching result is obtained. The business server 100 can generate a media push list for the object based on the interest matching result, wherein the media push list can include media data under at least two media field types, such as at least two media field types can include comedies with interest weight data of 0.05, tragedies with interest weight data of 0.1, and science fiction dramas with interest weight data of 0.35, and the number of media data under the media field type corresponding to the largest interest weight data in the media push list accounts for the largest proportion, that is, the number of science fiction dramas accounts for the largest proportion in the media push list.

[0140] When the business server 100 detects that the object performs key interactive operations on media data under at least two media field types in the terminal device 200, the interest weight data corresponding to the at least two media field types can be normalized to obtain the weight ratios corresponding to the at least two media field types, and the sum of the weight ratios of each media field type is 1. For example, if it is detected that the object frequently fast-forwards when watching a comedy film or TV show, it indicates that the object has a low interest in watching comedies, or if it is detected that the object repeatedly watches a certain clip when watching a science fiction drama or TV show, it indicates that the object has a high interest in watching science fiction dramas. In this case, the business server 100 can normalize the interest weight data corresponding to comedies, tragedies, and science fiction dramas, respectively, to obtain the weight ratios corresponding to comedies, tragedies, and science fiction dramas, respectively, i.e., a weight ratio of 0.2 for comedies, a weight ratio of 0.2 for tragedies, and a weight ratio of 0.6 for science fiction dramas.

[0141] The business server 100 can adjust at least two weight ratios based on key interactive operations, and then update the media push list based on the at least two adjusted weight ratios. For example, by frequently fast-forwarding a subject to comedy films and television programs, it can be identified that the subject is not interested in comedy. Therefore, the weight ratio of comedy can be reduced by 0.1, and the weight ratios of tragedy and science fiction dramas can be increased by 0.05 respectively. The adjusted weight ratios of each media field type are obtained, that is, the weight ratio corresponding to comedy is 0.1, the weight ratio corresponding to tragedy is 0.25, and the weight ratio corresponding to science fiction drama is 0.65. In this way, the media push list can be updated based on the adjusted weight ratios of each media field type, that is, the number of comedy films and television programs in the media push list is reduced, and the number of tragedy films and television programs and science fiction drama films and television programs are increased simultaneously.

[0142] It can be understood that the embodiment of the present application can generate the object's interest weight data for each media field type by obtaining the log data for the object, and based on the object's interactive operations on media data under at least two media field types in the log data within the target time period, interest matching is performed in the candidate media data set according to the interest weight data corresponding to the at least two media field types, and a media push list for the object can be obtained. When a key interactive operation of the object for media data under at least two media field types is detected, the interest weight data corresponding to the at least two media field types are normalized to obtain the weight ratios corresponding to the at least two media field types. After adjusting the at least two weight ratios through the key interactive operation, the media push list can be updated according to the at least two adjusted weight ratios. Since the interest weight data is generated based on the object's interactive operations on media data under at least two media field types within the target time period, when the object's interactive operations change, the interest weight data will also change accordingly. It can be seen that the interest weight data can accurately reflect the object's interest habits, so media data that meets the object's viewing interests can be more accurately selected based on the interest weight data. Through the key interactive operations of the object, the weight ratios corresponding to at least two media field types can be adjusted in real time, and the media push list can be updated according to the at least two adjusted weight ratios. Therefore, the changes in the object's viewing interests can be obtained in real time, and then the media push list can be updated in real time based on the changes in the object's viewing interests, so as to ensure that the real-time updated media push list is sufficiently matched with the object's viewing interests, so as to improve the object's viewing experience, and also reduce the traffic consumption caused by the object's frequent switching of media data due to lack of interest.

[0143] See Figure 3 , Figure 3 This is a flow diagram of a data processing method provided in an embodiment of the present application. Figure 1The data processing method can be executed by a computer device, which can be Figure 1 The service server 100 or terminal device 200 shown may also be a system composed of the service server 100 and the terminal device 200; the following description will be made using the computer device as the service server 100 as an example. The data processing method may include at least the following steps S101 to S104:

[0144] Step S101: Obtain log data for an object; based on the object's interactive operations on media data under at least two media domain types within a target time period in the log data, generate interest weight data for the object for each media domain type;

[0145] Specifically, the computer device can perform comprehensive tracking on the application side or web page side of the terminal device of the object, so as to automatically collect the log data of the object (such as page browsing log data and click interaction log data, etc.), and transmit the log data to the log database for storage through a message queue or other means. The computer device obtains the interactive operations of the object on media data (such as media data A1 under media field type, media data B1 under media field type B, etc.) under at least two media field types (such as media field type A, media field type B, etc.) in the log database within the target time period. The computer device can generate interest weight data for each media field type based on the interactive operations, such as generating the object's interest weight data for media field type A based on the interactive operations such as clicks, plays, etc. on multiple media data under media field type A in the past hour and the interactive operations on media data that the object has interacted with under all media field types in the past hour.

[0146] Step S102: performing interest matching in the candidate media data set based on the interest weight data corresponding to at least two media field types, and obtaining a media push list for the target;

[0147] Specifically, the computer device can perform interest matching in the candidate media data set based on the interest weight data A2 corresponding to the media domain type A and the interest weight data B2 corresponding to the media domain type B. For example, when the interest weight data A2 is greater than the weight threshold (such as 0.6), the media data with which the object has interacted in the media domain type A (target media domain type) can be feature fused to obtain a target feature vector, and the media data under the media domain type A in the candidate media data set can be feature extracted to obtain a set of candidate feature vectors. After performing similarity calculation (such as cosine similarity calculation) between the target feature vector and each candidate feature vector in the candidate feature vector set, the media data corresponding to the candidate feature vectors whose similarity is greater than or equal to the similarity threshold (such as 0.7) can be integrated into a media push list for the object. If there are multiple media field types whose interest weight data are all greater than the weight threshold, the number of media data of multiple media field types can be allocated proportionally based on the total number of media data that the media push list can accommodate. For example, if the interest weight data A2 and the interest weight data B2 are both 0.7, the same number of media data (such as 10) can be selected from media field type A and media field type B and put into the media push list.

[0148] Step S103: If a key interactive operation of the object on media data of at least two media domain types is detected, normalizing the interest weight data corresponding to the at least two media domain types to obtain weight ratios corresponding to the at least two media domain types;

[0149] Specifically, the computer device can continuously detect the interactive operations of the object (such as automatically collecting through JavaScript scripts in the terminal device page). When it detects the key interactive operations of the object on the media data under at least two media field types (such as the object repeatedly watching a certain segment in the media data B1), the interest weight data corresponding to the at least two media field types are normalized to obtain the weight ratios corresponding to the at least two media field types. For example, the object watched media data A1 belonging to media field type A and media data B1 belonging to media field type B, and frequently fast-forwarded while watching media data A1 or clicked the "not interested" feedback button after watching, indicating that the object had a low interest in watching the film of media field type A. Or if the object clicked the "like" feedback button after watching media data B1, indicating that the object had a high interest in watching the film of media field type B, the business server 100 can normalize the interest weight data corresponding to media field type A and media field type B respectively to obtain the weight ratio A3 corresponding to media field type A and the weight ratio B3 corresponding to media field type B.

[0150] Step S104: adjusting at least two weight ratios based on the key interactive operation, and updating the media push list according to the at least two adjusted weight ratios.

[0151] Specifically, the computer device can adjust the weight ratio A3 and the weight ratio B3 based on the key interactive operations of the object on the media data A1 and the media data B1. For example, if the ratio adjustment values corresponding to the feedback buttons "Like" and "Not Interested" are both 0.1, then after the object clicks the feedback button "Not Interested" for the media data A1, the weight ratio A3 can be reduced by 0.1, and the weight ratio B3 can be increased by 0.1 at the same time; or after the object clicks the feedback button "Like" for the media data B1, the weight ratio B3 can be increased by 0.1, and the weight ratio A3 can be reduced by 0.1 at the same time. In this way, the computer device can update the above-mentioned media push list based on the adjusted weight ratio A3 and weight ratio B3, that is, based on the increased weight ratio B3 and the decreased weight ratio A3, reduce the number of media data under the media field type A in the media push list, and increase the number of media data under the media field type B at the same time, to obtain an updated media push list for the object, thereby replacing the media push list with the updated media push list.

[0152] It can be understood that the embodiment of the present application can generate the object's interest weight data for each media field type by obtaining the log data for the object, and based on the object's interactive operations on media data under at least two media field types in the log data within the target time period, interest matching is performed in the candidate media data set according to the interest weight data corresponding to the at least two media field types, and a media push list for the object can be obtained. When a key interactive operation of the object for media data under at least two media field types is detected, the interest weight data corresponding to the at least two media field types are normalized to obtain the weight ratios corresponding to the at least two media field types. After adjusting the at least two weight ratios through the key interactive operation, the media push list can be updated according to the at least two adjusted weight ratios. Since the interest weight data is generated based on the object's interactive operations on media data under at least two media field types within the target time period, when the object's interactive operations change, the interest weight data will also change accordingly. It can be seen that the interest weight data can accurately reflect the object's interest habits, so media data that meets the object's viewing interests can be more accurately selected based on the interest weight data. Through the key interactive operations of the object, the weight ratios corresponding to at least two media field types can be adjusted in real time, and the media push list can be updated according to the at least two adjusted weight ratios. Therefore, the changes in the object's viewing interests can be obtained in real time, and then the media push list can be updated in real time based on the changes in the object's viewing interests, so as to ensure that the real-time updated media push list is sufficiently matched with the object's viewing interests, so as to improve the object's viewing experience, and also reduce the traffic consumption caused by the object's frequent switching of media data due to lack of interest.

[0153] See Figure 4 , Figure 4 This is a flow diagram of a data processing method provided in an embodiment of the present application. Figure 2 The data processing method can be executed by a computer device, which can be Figure 1 The service server 100 or terminal device 200 shown may also be a system composed of the service server 100 and the terminal device 200; the following description will be made using the computer device as the service server 100 as an example. The data processing method may include at least the following steps S201 to S207:

[0154] Step S201, obtaining log data for an object; based on the log data, the object is assigned to the media domain type S in the target time period. i generating a first dimension parameter based on a behavioral relationship between the interactive behavior of the media data under the target time period and the interactive behavior of the object with respect to historical media data sets under at least two media domain types within the target time period;

[0155] Specifically, the computer device can be used to perform comprehensive tracking on the web page, application, or smart TV side of the media platform. On the web page side, a script (such as a JavaScript script) can be embedded into each page through the log server in the computer device, so that the log server can automatically collect page browsing log data. When the object loads the page, the terminal device can initiate an http (HyperText Transfer Protocol) request to the log server's specific URL (Uniform Resource Locator), thereby passing the collected page browsing log data (such as the page URL, loading time, dwell time, etc.) as parameters. The log server then responds to the http request and records the data. On the application side, an SDK (Software Development Kit) is used to provide an interface to the computer device, encapsulating the differences in the underlying components. Thus, the computer device can record the object's click interaction log data, such as the button clicked, the time of the operation, etc., by calling the SDK.

[0156] By analyzing the complete behavior chain of an object, such as the entire process from the object's discovery of film and television content to its viewing, the computer device can attach necessary related information when collecting behavior log data. This information is generated at the entry page of the object entering the platform and is carried in each subsequent operation of the object through URL parameters or other means, thereby connecting the different behavior log data of the object. By integrating the above-collected page browsing log data, click interaction log data, and behavior log data, log data for the object can be obtained. This log data can be effectively stored using a distributed file system (such as HDFS (Hadoop Distributed File System)) combined with a column-based storage database.

[0157] For media data in different media domain types, the computer device can obtain media data from multiple media platforms, or crawl media data from film and television information websites, forums, etc., without any restrictions. By preprocessing the media data, such as correcting or deleting erroneous data through data cleaning, unifying the data format through standardization, and removing duplicate data through deduplication, the computer device can construct a knowledge graph between the media data, such as classifying different media data according to basic information such as director and plot summary, to obtain historical media data sets under at least two media domain types, wherein the at least two media domain types include media domain type S i , i is a positive integer.

[0158] The computer device can target the media domain type S within the target time period based on the objects in the log data. i The first dimension parameter is generated by analyzing the interaction behavior of the media data under the target time period and the interaction behavior of the object for the historical media data set under at least two media field types in the target time period. The first dimension parameter can be a behavior impact parameter. The specific process of generating the behavior impact parameter can be: obtaining the interaction behavior of the object in the log data for the media field type S in the target time period i The first time decay factor, the first behavior type relevance and the first behavior interaction depth of the media data under the first time decay factor, the first behavior type relevance and the first behavior interaction depth are generated for the media domain type S i The invention relates to a method for generating a first interactive behavior data of the media data under the target time period; obtaining a second time decay factor, a second behavior type relevance, and a second behavior interaction depth of the historical media data set of the object under at least two media field types in the log data within the target time period; generating second interactive behavior data for at least two media field types based on the second time decay factor, the second behavior type relevance, and the second behavior interaction depth; and determining the ratio of the first interactive behavior data to the second interactive behavior data as a behavior influence parameter.

[0159] For example, the first time attenuation factor can be expressed as Used to characterize the object for the media domain type h (ie media domain type S i ), that is, the time t at which the interaction behavior i occurs. i The closer to the current timestamp T, ω i The closer it is to 1, the greater the influence of the interaction behavior i; the first behavior type correlation r i It is used to characterize the matching degree between the media data involved in the interactive behavior i of the object and the media domain type h, with a value of (0, 1); the first behavior is the interaction depth d i It is used to characterize the influence of the interaction behavior i of the object, and the value is (0, 1). If the interaction behavior i is the play behavior, then the first behavior that plays for more than 5 minutes is the interaction depth d i The value can be 0.7. Where i is the interaction behavior of object u for media data of media domain type h within the current timestamp T (e.g., the past hour).

[0160] The computer device can be based on the interactive behavior set I of the object u related to the media domain type h within the current time stamp T (such as the past 1 hour). u,h,T The first time attenuation factor ω in i , first behavior type relevance r i Interaction depth d with the first behavior i , generating the first interactive behavior data The computer device can also collect the interaction behaviors related to the historical media data set (including all media domain types) u,T The second time decay factor ω in j , the second behavior type relevance r j The interaction depth d of the second behavior j Generate second interaction behavior data for at least two media field types Where j is the interaction behavior of subject u with respect to media data of at least two media domain types within the current timestamp T (e.g., the past hour). The generation process of its second interaction behavior data is the same as that of the first interaction behavior data. Therefore, the ratio of the first interaction behavior data to the second interaction behavior data can be determined as the behavior influence parameter. By comprehensively considering time decay and the influence of interaction behavior, the current interest tendency of subject u can be more accurately captured.

[0161] Optionally, the first dimension parameter may also be an intensity dimension parameter. Then, the specific process of generating the intensity dimension parameter may be: obtaining the object in the log data for the media domain type S within the target time period; i The first interaction intensity data are summed up to obtain the first interaction intensity data; the first interaction intensity data are used to characterize the influence of the interactive behavior of the object, and different interactive behaviors correspond to different first single interaction intensity data; the second interaction intensity data of the object in the log data for the historical media data set within the target time period are obtained, and the second interaction intensity data are summed up to obtain the second interaction intensity data; the ratio of the first interaction intensity data to the second interaction intensity data is determined as the intensity dimension parameter.

[0162] For example, the computer device can obtain the interaction intensity set O of the object u to the media data under the media domain type h within the current timestamp T from the log data. u,h,T , and then from the interaction intensity set O u,h,T Get the first single interaction strength data S o , where o represents a single interactive behavior triggered by an object for media data under media domain type h, then S o Represents the interaction intensity corresponding to a single interaction behavior o, and different interaction behaviors may correspond to different first single interaction intensity data S o (For example, the first single interaction strength data S o 3. The first single interaction strength data S of the comment o 2, etc.). Set the interaction strength set O u,h,T Multiple first single interaction strength data S in oPerform summation to obtain the first interaction strength data It is understandable that the computer device can also obtain the interaction intensity set O of the object u to the media data under all media domain types within the current timestamp T from the log data. uT , and then from the interaction intensity set O uT Get the second single interaction strength data S p , where p represents a single interactive behavior triggered by an object for media data under all media domain types, then S p Represents the interaction intensity corresponding to a single interaction behavior p. By transforming the interaction intensity set O u,T Multiple second single interaction strength data S in p Perform summation to obtain the second interaction strength data The ratio of the first interaction intensity data to the second interaction intensity data is thus determined as the intensity dimension parameter. By comprehensively considering the intensity dimensions of different interactive behaviors, the degree of interest of subject u in media domain type h can be further refined. Optionally, the first dimension parameter can include both the behavior impact parameter and the intensity dimension parameter.

[0163] Step S202: Based on the objects in the log data within the target time period, the media domain type S i The second dimension parameter is generated by the relationship between the browsing time of the media data under the target time period and the browsing time of the historical media data set of the object in the target time period;

[0164] Specifically, the computer device can obtain the objects in the log data for the media domain type S within the target time period. i The method comprises the following steps: obtaining multiple first single browsing durations of the media data under the target time period, summing up the multiple first single browsing durations to obtain the first browsing duration; obtaining multiple second single browsing durations of the object in the log data for the historical media data set within the target time period, summing up the multiple second single browsing durations to obtain the second browsing duration; and determining the ratio of the first browsing duration to the second browsing duration as the second dimension parameter.

[0165] For example, the computer device can obtain the total time set K of the media data of the media domain type h that the object u watches within the current timestamp T from the log data. u,h,T , and then the total duration set K u,h,T Multiple first single browsing durations l k Perform sum processing to get the first browsing time Where k represents a single interaction behavior triggered by an object for media data of media domain type h, then l kRepresents the single browsing duration corresponding to a single interactive behavior k. Similarly, by obtaining the total duration set K of media data of all media domain types viewed by object u within the current timestamp T u,T , and the total duration set K u,T Multiple second single browsing durations in l l By summing the results, we can get the second browsing time. Where l represents a single interaction behavior triggered by the object for media data under all media domain types, then l l The second dimension parameter is the ratio of the first browsing duration to the second browsing duration, which represents the single browsing duration corresponding to the single interaction behavior l. This second dimension parameter can more accurately reflect the preference of subject u for media domain type h during the target time period.

[0166] Step S203: generating a third dimension parameter based on the frequency relationship between the object's interaction frequency with media data of the media domain type Si within the target time period in the log data and the object's interaction frequency with the historical media dataset within the target time period;

[0167] Specifically, the computer device can obtain the objects in the log data for the media domain type S within the target time period. i The first interaction number of the media data under the target time period, and the second interaction number of the object in the log data for the historical media data set within the target time period; the ratio of the first interaction number to the second interaction number is determined as the third dimension parameter.

[0168] For example, the computer device can obtain the interaction behavior set M between the object u and the media data under the media domain type h within the current timestamp T from the log data. u,h,T , based on the interactive behavior set M u,h,T Each interaction behavior m in gets the first interaction number (i.e., executing the interactive behavior set M u,h,T The total number of all interaction behaviors in ), where m represents the interaction behavior triggered by the object for the media data under the media domain type h, and the number 1 represents the statistical number of times when an interaction behavior m is executed. Similarly, the set M of interaction behaviors of object u with the media data under all media domain types within the current timestamp T (such as the past week) can be obtained from the log data. u,T , based on the interactive behavior set M u,T Each interaction behavior n gets the second interaction number (i.e., executing the interactive behavior set M u,TThe total number of all interactive behaviors in the data), where n represents the interactive behavior triggered by the object for media data under all media field types, and the number 1 represents the statistical number of times the interactive behavior n is performed once, so the ratio of the first number of interactions to the second number of interactions can be determined as the third dimension parameter.

[0169] Step S204: Generate an object for the media domain type S according to the first dimension parameter, the second dimension parameter and the third dimension parameter. i Interest weight data;

[0170] Specifically, the computer device may obtain at least two adjustment parameters, and perform weighted summation on the first dimension parameter, the second dimension parameter, and the third dimension parameter based on the at least two adjustment parameters to obtain the object's weighted sum for the media domain type S. i interest weight data, wherein at least two adjustment parameters are used to determine the degree of influence of first dimension parameters, second dimension parameters and third dimension parameters on the interest weight data, and the first dimension parameters may include behavior influence parameters related to real-time behavior and intensity dimension parameters for characterizing interaction intensity.

[0171] Taking the example that the first dimension parameter includes both the behavior influence parameter and the intensity dimension parameter, the computer device can obtain at least two adjustment parameters from the log data, namely, the adjustment parameter α for adjusting the behavior influence parameter, the adjustment parameter β for adjusting the second dimension parameter, the adjustment parameter γ for adjusting the third dimension parameter, and the adjustment parameter δ for adjusting the intensity dimension parameter. By weighting the behavior influence parameter, the second dimension parameter, the third dimension parameter, and the intensity dimension parameter with at least two adjustment parameters, the interest weight data W of the object u for the media field type h within the current timestamp T can be obtained. u,h,T , the process can be shown as formula (1):

[0172]

[0173] It can be understood that the computer device can dynamically adjust at least two adjustment parameters according to the data volume of the log data of the object. The specific process can be: if it is detected that the data volume of the log data is less than the data volume threshold, then the adjustment parameters of the at least two adjustment parameters respectively associated with the behavior influence parameter and the second dimension parameter are reduced, and the adjustment parameters of the at least two adjustment parameters respectively associated with the third dimension parameter and the intensity dimension parameter are increased; if it is detected that the data volume of the log data is greater than or equal to the data volume threshold, then the adjustment parameters of the at least two adjustment parameters respectively associated with the behavior influence parameter and the second dimension parameter are increased, and the adjustment parameters of the at least two adjustment parameters respectively associated with the third dimension parameter and the intensity dimension parameter are reduced.

[0174] For example, the initial value of adjustment parameter α can be 0.4, the initial value of adjustment parameter β can be 0.25, the initial value of adjustment parameter γ can be 0.2, and the initial value of adjustment parameter δ can be 0.15. When subject u begins to watch media data under multiple media domain types including media domain type h, that is, in the cold start phase, and the computer device detects that the data volume of subject u's log data is less than a data volume threshold, adjustment parameter α and adjustment parameter β can be both reduced by 0.1, and adjustment parameter γ and adjustment parameter δ can be both increased by 0.1. This will increase the emphasis on judging the media domain types that subject u may be interested in based on the interaction frequency and interaction intensity of subject u at the current timestamp T. When subject u's log data gradually increases to a value greater than or equal to the data volume threshold, adjustment parameter α and adjustment parameter β can be both increased by 0.1, and adjustment parameter γ and adjustment parameter δ can be both reduced by 0.1. This will better reflect subject u's real-time interactive behavior and historical viewing preferences at the current timestamp T.

[0175] Step S205 , performing interest matching in the candidate media data set based on the interest weight data corresponding to at least two media field types, and obtaining a media push list for the target;

[0176] For details, please refer to Figure 5 , Figure 5 This is a data processing scenario provided by the embodiment of the present application. Figure 2 ,like Figure 5As shown, at least two media domain types include a target media domain type. The computer device can, based on a collaborative filtering method, perform interest matching between objects in the candidate media data set according to the interest weight data corresponding to the at least two media domain types, and obtain a media push list for the object. The specific process can be: if the interest weight data related to the target media domain type corresponding to the object is greater than the weight threshold, then the interest weight data related to the target media domain type corresponding to at least two historical objects are determined as target interest weight data, and the historical objects corresponding to the target interest weight data greater than or equal to the weight threshold are determined as candidate objects; the first historical interaction behavior of the candidate object with respect to the candidate media data set is obtained, and the second historical interaction behavior of the object in the log data is obtained; the object similarity between the candidate object and the object is determined based on the first historical interaction behavior and the second historical interaction behavior; if the object similarity is greater than the similarity threshold, a media push list for the object is generated based on the media data with which the candidate object has interacted and with which the object has not interacted. If there are multiple media field types whose interest weight data are all greater than the weight threshold (such as 0.6), the number of media data of multiple media field types can be allocated proportionally based on the total number of media data that the media push list can accommodate. For example, if the interest weight data A2 and the interest weight data B2 are both 0.7, the same number of media data (such as 10) can be selected from media field type A and media field type B and put into the media push list.

[0177] For example, when the computer device detects that the interest weight data u1 related to object u and comedy is greater than a weight threshold (such as 0.7), the interest weight data a1 related to object a and comedy (such as 0.8) and the interest weight data b1 related to object b and comedy (such as 0.8) can be determined as target interest weight data. Since the interest weight data a1 and the interest weight data b1 are both greater than the weight threshold, objects a and b can be determined as candidate objects. The computer device can construct an object-media data scoring matrix between objects u, a, and b and the candidate media datasets, as shown in Table 1:

[0178] Table 1

[0179]

[0180] The computer device can obtain the media data scoring vector (4, 3, 8) of object u, the media data scoring vector (5, 2, 8) of object a, and the media data scoring vector (2, 6, 4) of object b through Table 1, and use this to calculate the similarity between object u and object a and object b respectively. For example, when the similarity threshold is 0.9, and the cosine similarity between object u and object a is calculated, Similarly, the cosine similarity between object u and object b is approximately 0.82, which determines that object u and object a are objects with high similarity. A media push list for object u can be generated based on media data such as comedy films and science fiction films that object a has interacted with but object u has not interacted with.

[0181] It can be understood that the computer device can also use a content-based recommendation method to generate a media push list for the object. The specific process can be: if the interest weight data related to the target media field type corresponding to the object is greater than the weight threshold, then the media data under the target media field type that the object has interacted with is subjected to feature fusion to obtain a target feature vector; each media data under the target media field type in the candidate media data set is subjected to feature extraction to obtain a set of candidate feature vectors; the target feature vector is calculated to have a similarity with each candidate feature vector in the candidate feature vector set, and a media push list for the object is generated based on the media data corresponding to the candidate feature vector whose similarity is greater than or equal to the similarity threshold.

[0182] For example, when the interest weight data u1 related to object u and comedy is greater than a weight threshold, the computer device may perform multi-dimensional feature extraction on the comedy media data with which object u has interacted, and then perform feature fusion on the extracted features. For example, by extracting the video frame features, text features, and audio features of the media data through a multimodal model, and then assigning feature weights to the video frame features, text features, and audio features, respectively, and performing weighted summation, the multi-dimensional features are concatenated into high-dimensional fused features to obtain a target feature vector. By performing feature extraction on each piece of comedy media data in the candidate media dataset, a candidate feature vector set consisting of the candidate feature vectors corresponding to each piece of media data can be obtained.

[0183] By calculating the similarity between the target feature vector and each candidate feature vector, a similarity calculation result can be obtained. For example, the dot product result of the target feature vector and each candidate feature vector is calculated by cosine similarity. After normalizing the modulus of the target feature vector and the modulus of each candidate feature vector, the product of the modulus of the target feature vector and the modulus of each candidate feature vector is calculated, thereby calculating the ratio of the dot product result to the product of the modulus, and using the ratio as the cosine similarity between the target feature vector and each candidate feature vector, that is, the similarity calculation result. The computer device can combine the media data corresponding to the candidate feature vectors whose similarity calculation results are greater than or equal to the similarity threshold into a media push list for object u.

[0184] It can be understood that the computer device can integrate the collaborative filtering method and the content-based recommendation method into a hybrid recommendation method through list generation parameters (such as two normalized generation ratios). For example, a generation ratio of 0.5 can be assigned to each method to generate a media push list, that is, the number of media data generated by the collaborative filtering method in the media push list is the same as the number of media data generated by the content-based recommendation method. When it is detected that the number of historical objects is lower than the object number threshold (such as 10), the generation ratio of collaborative filtering can be reduced and the generation ratio of content-based recommendation can be increased simultaneously; when it is detected that the number of media data in the candidate media data set is less than the number threshold, the generation ratio of content-based recommendation can be reduced and the generation ratio of collaborative filtering can be increased simultaneously. The computer device can also regularly collect feedback data from the object on the media push list, such as the number of clicks, viewing time, and ratings of the recommended media data by the object, and optimize the hybrid recommendation method based on these feedback data.

[0185] Step S206: If a key interactive operation of the object on media data of at least two media domain types is detected, normalizing the interest weight data corresponding to the at least two media domain types to obtain weight ratios corresponding to the at least two media domain types;

[0186] Specifically, computer devices can use distributed stream processing frameworks (such as Apache Flink) to build real-time data processing platforms. When an object generates an interactive operation on media data, such as clicking on the media data's details page, starting to watch, pausing, fast-forwarding, and other operations, the data generated by the interactive operation is transmitted to the Flink cluster in real time via the message queue Kafka. The Flink cluster can then parse and process this data in real time, quickly extracting key information. To achieve fast query and data reading, computer devices can use Redis as a cache database to build an object interaction operation cache and a media data cache. This way, the object's real-time interactive operations are processed by the Flink cluster and immediately updated to the Redis cache.

[0187] In this way, the computer device can obtain the interactive operations of object u from the Redis cache database in real time. When a key interactive operation of object u on media data belonging to comedy is detected (such as repeated viewing of a certain clip in a comedy film or TV show), the interest weight data u2 (such as 0.5) corresponding to comedy and the interest weight data u3 (such as 0.3) corresponding to tragedy can be scaled (that is, normalized to a sum of 1) to obtain the weight ratio corresponding to comedy 0.5 / (0.5+0.3)=0.625, and the weight ratio corresponding to tragedy is 0.375.

[0188] Step S207: Adjust at least two weight ratios based on the key interaction operation, perform interest matching in the candidate media data set according to the at least two adjusted weight ratios, obtain an updated media push list for the object, and replace the media push list with the updated media push list.

[0189] For details, please refer to Figure 6 , Figure 6 This is a data processing scenario provided by the embodiment of the present application. Figure 3 .like Figure 6 As shown, the computer device can adjust at least two weight ratios according to the type of the key interactive operation (positive interactive operation or negative interactive operation), and then update the media push list according to the at least two adjusted weight ratios. The specific process can be: if the key interactive operation is a positive interactive operation, the weight ratio of the key media field type is increased, and the other weight ratios except the weight ratio of the key media field type in at least two weight ratios are simultaneously reduced; the key media field type refers to the media field type to which the media data operated by the key interactive operation belongs; based on at least two adjusted weight ratios, the number of media data belonging to the key media field type in the media push list is increased, and the number of media data belonging to other media field types in the media push list is simultaneously reduced; other media field types refer to media field types except the key media field type in at least two media field types; if the key interactive operation is a negative interactive operation, the weight ratio of the key media field type is reduced, and the other weight ratios are simultaneously increased; based on at least two adjusted weight ratios, the number of media data belonging to the key media field type in the media push list is reduced, and the number of media data belonging to other media field types in the media push list is simultaneously increased. For example, a positive interactive operation may be repeated viewing of a certain segment in the media data, and a negative interactive operation may be frequent fast-forwarding of the media data.

[0190] For example, at least two weight ratios may include a weight ratio c1 corresponding to media domain type c, a weight ratio d1 corresponding to media domain type d, and so on. When the key interactive operation is a forward interactive operation, the key media domain type may be media domain type c, and the forward interactive operation may be the object's repeated viewing of a certain segment of media data under media domain type c. In this case, the computer device may respond to the forward interactive operation by increasing weight ratio c1 by 0.1 and simultaneously decreasing all other weight ratios except weight ratio c1, with the sum of the decreased weight ratios being 0.1. For example, if the other weight ratios may be weight ratio d1 and weight ratio e1, then both weight ratio d1 and weight ratio e1 may be decreased by 0.05. Thus, the computer device may increase the amount of media data under media domain type c in the media push list by 10%, and simultaneously decrease the amount of media data of other media domain types except media domain type c in the media push list by 10%. When the key interactive operation is a negative interactive operation, the key media domain type may be media domain type d, and the negative interactive operation may be the frequent fast-forwarding of the object for media data under media domain type d. The computer device may respond to the negative interactive operation of the object by reducing the weight ratio d1 by 0.2, and simultaneously increasing all other weight ratios except the weight ratio d1, and the sum of the increased weight ratios is 0.2. This may reduce the amount of media data under media domain type d in the media push list by 20%, and simultaneously increase the amount of media data of other media domain types except media domain type d in the media push list by 20%.

[0191] It can be understood that the key interactive operation can also be a trigger operation of the object on the feedback control or a sharing operation on the social platform. If the key interactive operation is a trigger operation on the feedback control, the proportion adjustment value is obtained according to the feedback intention indicated by the feedback control, and the weight ratio of the key media field type in at least two weight ratios is adjusted in a first proportion based on the proportion adjustment value, and the other weight ratios in at least two weight ratios except the weight ratio of the key media field type are adjusted in a second proportion; the key media field type refers to the media field type to which the media data operated by the key interactive operation belongs; if the key interactive operation is a sharing operation on a social platform, the platform priority of the shared social platform is obtained, and the weight ratio of the key media field type is increased based on the proportion mapped by the platform priority, and other weight ratios are reduced simultaneously.

[0192] Specifically, when the key interactive operation is a trigger operation for a feedback control, the key media domain type may be media domain type e, and the trigger operation may be a positive review given by the object based on the rating control of the media data of media domain type e, or a click operation of the object on the "like" control of the media data of media domain type e. The computer device may obtain a proportional adjustment value (such as 0.2) corresponding to the positive review, thereby increasing the weight ratio e1 corresponding to media domain type e by 0.2, and simultaneously reducing the weight ratios of other media domain types in the media push list except media domain type e, and the sum of the reduced weight ratios is 0.2. When the key interactive operation is a sharing operation on a social platform, the key media domain type can be media domain type f. The computer device can obtain the platform priority of the social platform. For example, the platform priority of social platform p1 is the highest level (the mapping ratio is increased by 0.2), and the platform priority of social platform p2 is the second highest level (the mapping ratio is increased by 0.1). When the object shares media data under media domain type f to social platform p1, the weight ratio f1 corresponding to media domain type f can be increased by 0.2, and the weight ratios of other media domain types other than media domain type f in the media push list can be reduced simultaneously, and the sum of the reduced weight ratios is 0.2. When updating the media push list, the computer device can also use an incremental update algorithm to increase the ranking of media data under the key media domain type in the media push list, and simultaneously adjust the ranking of media data of other media domain types, so as to quickly complete the update of the media push list. The object can then obtain more media data belonging to the key media domain type in the updated media push list in a more timely manner.

[0193] Optionally, the computer device may also re-perform interest matching in the candidate media data set based on the interest weight data corresponding to at least two adjusted weight ratios to obtain an updated media push list, and replace the media push list with the updated media push list. The process can be found in the specific description of interest matching in the above step S205 and will not be repeated here.

[0194] It can be understood that the computer device can obtain the device information of the object (such as device model, operating system information and resolution, etc.) through the terminal device interface of the object, and obtain the geographical area where the object is located based on the positioning function of the terminal device. It can also obtain the network delay of the object by regularly sending heartbeat packets to the terminal device and measuring the response time, and then obtain at least two participating objects containing the object in the geographical area. For the playback network status of the target media data (such as the media data under the above-mentioned key media field type) (such as the number of playback freezes, loading time, screen switching quality, etc.), when it is detected that the number of participating objects with a playback network status of network freeze is greater than the object number threshold (such as 50), the clarity recommended version of the target media data in the media push list can be adjusted, that is, low-bitrate, smooth-playing media data can be pushed first.

[0195] It can be understood that the embodiment of the present application can generate the object's interest weight data for each media field type by obtaining the log data for the object, and based on the object's interactive operations on media data under at least two media field types in the log data within the target time period, interest matching is performed in the candidate media data set according to the interest weight data corresponding to the at least two media field types, and a media push list for the object can be obtained. When a key interactive operation of the object for media data under at least two media field types is detected, the interest weight data corresponding to the at least two media field types are normalized to obtain the weight ratios corresponding to the at least two media field types. After adjusting the at least two weight ratios through the key interactive operation, the media push list can be updated according to the at least two adjusted weight ratios. Since the interest weight data is generated based on the object's interactive operations on media data under at least two media field types within the target time period, when the object's interactive operations change, the interest weight data will also change accordingly. It can be seen that the interest weight data can accurately reflect the object's interest habits, so media data that meets the object's viewing interests can be more accurately selected based on the interest weight data. Through the key interactive operations of the object, the weight ratios corresponding to at least two media field types can be adjusted in real time, and the media push list can be updated according to the at least two adjusted weight ratios. Therefore, the changes in the object's viewing interests can be obtained in real time, and then the media push list can be updated in real time based on the changes in the object's viewing interests, so as to ensure that the real-time updated media push list is sufficiently matched with the object's viewing interests, so as to improve the object's viewing experience, and also reduce the traffic consumption caused by the object's frequent switching of media data due to lack of interest.

[0196] See Figure 7 , Figure 7 This is a structural diagram of a data processing device provided in an embodiment of the present application. Figure 7As shown, the data processing device includes a transceiver module 1100 , a data generation module 1200 , a list generation module 1300 , a ratio generation module 1400 and a list update module 1500 .

[0197] The transceiver module 1100 is used to obtain log data for an object;

[0198] A data generation module 1200 is configured to generate interest weight data for each media field type based on the object's interactive operations on media data of at least two media field types within a target time period in the log data;

[0199] List generation module 1300 is configured to perform interest matching in a candidate media data set based on interest weight data corresponding to at least two media domain types, and obtain a media push list for an object; in the media push list, the media domain type corresponding to the data with the largest interest weight has the largest proportion of media data;

[0200] Ratio generation module 1400 is configured to, upon detecting a key interaction operation of an object on media data of at least two media domain types, normalize the interest weight data corresponding to the at least two media domain types to obtain weight ratios corresponding to the at least two media domain types;

[0201] The list updating module 1500 is configured to adjust at least two weight ratios based on the key interaction operation, and update the media push list according to the at least two adjusted weight ratios.

[0202] In one possible implementation, the at least two media domain types include a media domain type S i , i is a positive integer; the data generation module 1200 is used to generate the object's interest weight data for each media field type based on the object's interactive operations on media data under at least two media field types within a target time period in the log data, specifically for performing the following operations:

[0203] Based on the objects in the log data within the target time period for the media domain type S i generating a first dimension parameter based on a behavioral relationship between the interactive behavior of the media data under the target time period and the interactive behavior of the object with respect to historical media data sets under at least two media domain types within the target time period;

[0204] Based on the objects in the log data within the target time period for the media domain type S i The second dimension parameter is generated by the relationship between the browsing time of the media data under the target time period and the browsing time of the historical media data set of the object in the target time period;

[0205] Based on the objects in the log data within the target time period for the media domain type S i The frequency relationship between the interaction frequency of the media data under the target time period and the interaction frequency of the object with respect to the historical media data set in the target time period is used to generate a third dimension parameter;

[0206] According to the first dimension parameter, the second dimension parameter and the third dimension parameter, an object is generated for the media domain type S i Interest weight data.

[0207] In a possible implementation, the first dimension parameter includes a behavior impact parameter; the data generation module 1200 is used to generate a behavior impact parameter based on the behavior of the object in the log data within the target time period for the media domain type S. i The behavioral relationship between the interactive behavior of the media data under the target time period and the interactive behavior of the object for the historical media data sets under at least two media field types during the target time period is used to generate the first dimension parameter, specifically for performing the following operations:

[0208] Get the objects in the log data for the media field type S within the target time period i The first time decay factor, first behavior type relevance and first behavior interaction depth of the media data under the first time decay factor is used to characterize the object for the media field type S i The first behavior type relevance is used to characterize the media data involved in the object's interactive behavior and the media domain type S i The first behavior interaction depth is used to characterize the influence of the object's interaction behavior;

[0209] Based on the first time attenuation factor, the first behavior type relevance and the first behavior interaction depth, a target for the media domain type S is generated. i The first interactive behavior data of the media data;

[0210] Obtaining a second time decay factor, a second behavior type relevance, and a second behavior interaction depth of an object in the log data for a historical media dataset under at least two media field types within a target time period;

[0211] generating second interactive behavior data for at least two media domain types based on a second time decay factor, a second behavior type relevance, and a second behavior interaction depth;

[0212] A ratio of the first interactive behavior data to the second interactive behavior data is determined as a behavior influence parameter.

[0213] In a possible implementation, the first dimension parameter includes an intensity dimension parameter; the data generation module 1200 is used to generate a data parameter based on the object in the log data within the target time period for the media domain type S i The behavioral relationship between the interactive behavior of the media data under the target time period and the interactive behavior of the object for the historical media data sets under at least two media field types during the target time period is used to generate the first dimension parameter, specifically for performing the following operations:

[0214] Get the objects in the log data for the media field type S within the target time period i a plurality of first single interaction intensity data of the media data under the image, summing the plurality of first single interaction intensity data to obtain first interaction intensity data; the first single interaction intensity data is used to characterize the influence degree of the interactive behavior of the object, and different interactive behaviors correspond to different first single interaction intensity data;

[0215] Acquire multiple second single interaction intensity data of the object in the log data with respect to the historical media dataset within the target time period, and sum the multiple second single interaction intensity data to obtain second interaction intensity data;

[0216] A ratio of the first interaction intensity data to the second interaction intensity data is determined as an intensity dimension parameter.

[0217] In a possible implementation, the data generation module 1200 is used to generate a data set based on the media domain type S in the target time period of the object in the log data. i The duration relationship between the browsing time of the media data under the target time period and the browsing time of the historical media data set of the object in the target time period is used to generate the second dimension parameter, specifically for performing the following operations:

[0218] Get the objects in the log data for the media field type S within the target time period i multiple first single browsing durations of the media data under the above conditions, summing the multiple first single browsing durations to obtain a first browsing duration;

[0219] Acquire multiple second single browsing durations of the object in the log data for the historical media dataset within the target time period, and sum the multiple second single browsing durations to obtain a second browsing duration;

[0220] The ratio of the first browsing time to the second browsing time is determined as the second dimension parameter.

[0221] In a possible implementation, the data generation module 1200 is used to generate a data set based on the media domain type S in the target time period of the object in the log data. iThe frequency relationship between the interaction frequency of the media data under the target time period and the interaction frequency of the object for the historical media data set in the target time period is used to generate the third dimension parameter, specifically for the following operations:

[0222] Get the objects in the log data for the media field type S within the target time period i The first number of interactions of the media data under the target time period, and the second number of interactions of the object in the log data with respect to the historical media data set within the target time period;

[0223] The ratio of the first interaction number to the second interaction number is determined as the third dimension parameter.

[0224] In a possible implementation, the data generation module 1200 is configured to generate the object's interest weight data for each media field type based on the first dimension parameter, the second dimension parameter, and the third dimension parameter, and specifically perform the following operations:

[0225] Obtain at least two adjustment parameters;

[0226] Based on at least two adjustment parameters, the first dimension parameter, the second dimension parameter and the third dimension parameter are weighted and summed to obtain the object for the media field type S i interest weight data; at least two adjustment parameters are used to determine the degree of influence of the first dimension parameter, the second dimension parameter and the third dimension parameter on the interest weight data.

[0227] In a possible implementation, the first dimension parameter includes a behavior impact parameter related to the real-time behavior and an intensity dimension parameter for characterizing interaction intensity; the data generation module 1200 is further configured to perform the following operations:

[0228] If it is detected that the data volume of the log data is less than the data volume threshold, then reducing the adjustment parameters associated with the behavior impact parameter and the second dimension parameter respectively among the at least two adjustment parameters, and simultaneously increasing the adjustment parameters associated with the third dimension parameter and the intensity dimension parameter respectively among the at least two adjustment parameters;

[0229] If it is detected that the data volume of the log data is greater than or equal to the data volume threshold, the adjustment parameters of at least two adjustment parameters respectively associated with the behavior impact parameter and the second dimension parameter are increased, and the adjustment parameters of at least two adjustment parameters respectively associated with the third dimension parameter and the intensity dimension parameter are reduced simultaneously.

[0230] In one possible implementation, at least two media domain types include a target media domain type; and the list generation module 1300 is configured to perform interest matching in the candidate media data set based on the interest weight data corresponding to the at least two media domain types, and to obtain a media push list for the target, specifically performing the following operations:

[0231] If the interest weight data related to the target media field type corresponding to the object is greater than the weight threshold, the interest weight data related to the target media field type corresponding to at least two historical objects are determined as target interest weight data, and the historical objects corresponding to the target interest weight data greater than or equal to the weight threshold are determined as candidate objects;

[0232] Obtaining a first historical interaction behavior of the candidate object with respect to the candidate media dataset, and obtaining a second historical interaction behavior of the object in the log data;

[0233] determining an object similarity between the candidate object and the object based on the first historical interaction behavior and the second historical interaction behavior;

[0234] If the object similarity is greater than the similarity threshold, a media push list for the object is generated based on the media data that the candidate object has interacted with and the object has not interacted with.

[0235] In one possible implementation, at least two media domain types include a target media domain type; and the list generation module 1300 is configured to perform interest matching in the candidate media data set based on the interest weight data corresponding to the at least two media domain types, and to obtain a media push list for the target, specifically performing the following operations:

[0236] If the interest weight data corresponding to the object and related to the target media domain type is greater than the weight threshold, then feature fusion is performed on the media data under the target media domain type that the object has interacted with to obtain a target feature vector;

[0237] Perform feature extraction on each media data under the target media domain type in the candidate media data set to obtain a set of candidate feature vectors;

[0238] The target feature vector is similar to each candidate feature vector in the candidate feature vector set, and a media push list for the object is generated based on the media data corresponding to the candidate feature vectors whose similarity is greater than or equal to the similarity threshold.

[0239] In one possible implementation, the list updating module 1500 is configured to adjust at least two weight ratios based on the key interaction operation, and when updating the media push list according to the at least two adjusted weight ratios, specifically to perform the following operations:

[0240] If the key interactive operation is a forward interactive operation, the weight ratio of the key media domain type is increased, and the weight ratios other than the weight ratio of the key media domain type in at least two weight ratios are simultaneously reduced; the key media domain type refers to the media domain type to which the media data operated by the key interactive operation belongs;

[0241] Based on at least two adjusted weight ratios, the amount of media data belonging to the key media domain type in the media push list is increased, and the amount of media data belonging to other media domain types in the media push list is simultaneously reduced; the other media domain types refer to media domain types other than the key media domain type in the at least two media domain types;

[0242] If the key interaction operation is a negative interaction operation, the weight ratio of the key media field type will be reduced, and the weight ratios of other types will be increased simultaneously;

[0243] Based on at least two adjusted weight ratios, the amount of media data belonging to the key media domain type in the media push list is reduced, and the amount of media data belonging to other media domain types in the media push list is increased simultaneously.

[0244] In a possible implementation, when the list updating module 1500 is configured to adjust at least two weight ratios based on the key interaction operation, it is specifically configured to perform the following operations:

[0245] If the key interactive operation is a trigger operation for a feedback control, obtaining a ratio adjustment value based on the feedback intent indicated by the feedback control, performing a first ratio adjustment on the weight ratio of the key media domain type in the at least two weight ratios based on the ratio adjustment value, and performing a second ratio adjustment on the other weight ratios in the at least two weight ratios except the weight ratio of the key media domain type; the key media domain type refers to the media domain type to which the media data operated by the key interactive operation belongs;

[0246] If the key interactive operation is a sharing operation on a social platform, the platform priority of the shared social platform is obtained, and the weight ratio of the key media field type is increased based on the proportion mapped by the platform priority, while the other weight ratios are reduced simultaneously.

[0247] In a possible implementation, when the list updating module 1500 is configured to update the media push list according to at least two adjusted weight ratios, it is specifically configured to perform the following operations:

[0248] Interest matching is performed in the candidate media data set according to at least two adjusted weight ratios to obtain an updated media push list for the object, and the media push list is replaced with the updated media push list.

[0249] In a possible implementation, the list updating module 1500 is further configured to perform the following operations:

[0250] Obtaining playback network status of target media data for at least two participating objects within a geographical area; the at least two participating objects include an object, and the media push list includes the target media data;

[0251] If it is detected that the number of participating objects whose playback network status is a network freeze state is greater than the object quantity threshold, the definition recommended version for the target media data in the media push list is adjusted.

[0252] In the embodiments of the present application, the term "module" or "unit" refers to a computer program or a part of a computer program that has a predetermined function and works together with other related parts to achieve a predetermined goal, and can be implemented in whole or in part by using software, hardware (such as processing circuits or memories) or a combination thereof. Similarly, a processor (or multiple processors or memories) can be used to implement one or more modules or units. In addition, each module or unit can be part of an overall module or unit that includes the function of the module or unit.

[0253] The embodiment of the present application can generate the object's interest weight data for each media field type by obtaining log data for the object, and based on the object's interactive operations on media data under at least two media field types in the log data within the target time period, interest matching is performed in the candidate media data set according to the interest weight data corresponding to the at least two media field types, and a media push list for the object can be obtained. When a key interactive operation of the object for media data under at least two media field types is detected, the interest weight data corresponding to the at least two media field types are normalized to obtain the weight ratios corresponding to the at least two media field types. After adjusting the at least two weight ratios through the key interactive operation, the media push list can be updated according to the at least two adjusted weight ratios. Since the interest weight data is generated based on the object's interactive operations on media data under at least two media field types within the target time period, when the object's interactive operations change, the interest weight data will also change accordingly. It can be seen that the interest weight data can accurately reflect the object's interest habits, so media data that meets the object's viewing interests can be more accurately selected based on the interest weight data. Through the key interactive operations of the object, the weight ratios corresponding to at least two media field types can be adjusted in real time, and the media push list can be updated according to the at least two adjusted weight ratios. Therefore, the changes in the object's viewing interests can be obtained in real time, and then the media push list can be updated in real time based on the changes in the object's viewing interests, so as to ensure that the real-time updated media push list is sufficiently matched with the object's viewing interests, so as to improve the object's viewing experience, and also reduce the traffic consumption caused by the object's frequent switching of media data due to lack of interest.

[0254] See Figure 8 , Figure 8 This is a schematic diagram of the structure of a computer device provided in an embodiment of the present application. Figure 8 As shown, the computer device 1000 may include: a processor 1001, a network interface 1004 and a memory 1005. In addition, the above-mentioned computer device 1000 may also include: an object interface 1003, and at least one communication bus 1002. The communication bus 1002 is used to realize the connection and communication between these components. The object interface 1003 may include a display screen (Display), a keyboard (Keyboard), and the object interface 1003 may optionally include a standard wired interface and a wireless interface. The network interface 1004 may optionally include a standard wired interface and a wireless interface (such as a WI-FI interface). The memory 1005 may be a high-speed RAM memory, or a non-volatile memory (non-volatile memory), such as at least one disk memory. The memory 1005 may optionally also be at least one storage device located away from the aforementioned processor 1001. As Figure 8 As shown, the memory 1005 as a computer-readable storage medium may include an operating system, a network communication module, an object interface module, and a device control application program.

[0255] In such Figure 8 In the computer device 1000 shown, the network interface 1004 can provide a network communication element; the object interface 1003 is mainly used to provide an input interface for the object; and the processor 1001 can be used to call the device control application stored in the memory 1005 to implement the above Figure 3 or Figure 4 The methods indicated in the examples.

[0256] It should be understood that the computer device 1000 described in the embodiment of the present application can execute the above Figure 3 and Figure 4 The description of the data processing method in any corresponding embodiment will not be repeated here. In addition, the description of the beneficial effects of adopting the same method will not be repeated here either.

[0257] In addition, it should be noted that: the embodiment of the present application also provides a computer-readable storage medium, and the computer-readable storage medium stores a computer program, and when the processor executes the computer program, it can execute the above Figure 3 and Figure 4The description of the above-mentioned data processing method in any corresponding embodiment will not be repeated here. In addition, the description of the beneficial effects of adopting the same method will not be repeated here. For technical details not disclosed in the computer-readable storage medium embodiment involved in this application, please refer to the description of the method embodiment of this application.

[0258] The computer-readable storage medium may be the data processing device provided in any of the aforementioned embodiments or the internal storage unit of the computer device, such as the hard disk or memory of the computer device. The computer-readable storage medium may also be an external storage device of the computer device, such as a plug-in hard disk, a smart media card (SMC), a secure digital (SD) card, a flash card, etc. equipped on the computer device. Furthermore, the computer-readable storage medium may also include both the internal storage unit of the computer device and an external storage device. The computer-readable storage medium is used to store the computer program and other programs and data required by the computer device. The computer-readable storage medium may also be used to temporarily store data that has been displayed or is about to be displayed.

[0259] In addition, it should be noted that the present application also provides a computer program product, which includes a computer program stored in a computer-readable storage medium. The processor of the computer device reads the computer program from the computer-readable storage medium, and the processor executes the computer program, so that the computer device performs the above-mentioned Figure 3 and Figure 4 The method provided by any corresponding embodiment.

[0260] The terms "first", "second", etc. in the description, claims, and drawings of the embodiments of the present application are used to distinguish different objects, rather than to describe a specific order. In addition, the terms "comprising" and any variations thereof are intended to cover non-exclusive inclusions. For example, a process, method, apparatus, product, or device comprising a series of steps or units is not limited to the listed steps or modules, but may optionally include steps or modules not listed, or may optionally include other step units inherent to these processes, methods, apparatuses, products, or devices.

[0261] Those skilled in the art will appreciate that the units and algorithm steps of each example described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, computer software, or a combination of the two. To clearly illustrate the interchangeability of hardware and software, the above description has generally described the components and steps of each example in terms of network elements. Whether these network elements are implemented in hardware or software depends on the specific application and design constraints of the technical solution. Professional and technical personnel may use different methods to implement the described network elements for each specific application, but such implementation should not be considered to be beyond the scope of this application.

[0262] The methods and related devices provided in the embodiments of the present application are described with reference to the method flow charts and / or structural diagrams provided in the embodiments of the present application. Specifically, each process and / or block in the method flow charts and / or structural diagrams, as well as the combination of processes and / or blocks in the flow charts and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable device to generate a machine, so that the instructions executed by the processor of the computer or other programmable device generate instructions for implementing the steps in the process. Figure 1 Schematic diagram of one or more processes and / or structures Figure 1 These computer program instructions can also be stored in a computer-readable memory that can guide a computer or other programmable device to work in a specific way, so that the instructions stored in the computer-readable memory produce a product including the instruction device, which implements the function specified in the process. Figure 1 Schematic diagram of one or more processes and / or structures Figure 1 These computer program instructions can also be loaded onto a computer or other programmable device so that a series of operation steps are executed on the computer or other programmable device to produce a computer-implemented process, thereby providing instructions for implementing the process in the computer or other programmable device. Figure 1 The flow or flows and / or structures illustrate the steps of the functions specified in one block or multiple blocks.

[0263] The steps in the method of the embodiment of the present application can be adjusted in order, combined and deleted according to actual needs.

[0264] The modules in the device of the embodiment of the present application can be merged, divided and deleted according to actual needs.

[0265] The above disclosure is only a preferred embodiment of the present application, and certainly cannot be used to limit the scope of rights of the present application. Therefore, equivalent changes made according to the claims of the present application are still within the scope covered by the present application.

Claims

1. A data processing method, characterized in that: include: Get log data for an object; generating interest weight data of the object for each media field type based on interactive operations of the object on media data under at least two media field types within a target time period in the log data; Based on the interest weight data corresponding to the at least two media field types, interest matching is performed in the candidate media data set to obtain a media push list for the object; The media data in the media push list corresponding to the media field type with the largest interest weight accounts for the largest proportion; If a key interactive operation of the object on the media data under the at least two media field types is detected, normalizing the interest weight data corresponding to the at least two media field types to obtain weight ratios corresponding to the at least two media field types; At least two weight ratios are adjusted based on the key interaction operation, and the media push list is updated according to the at least two adjusted weight ratios.

2. The method according to claim 1, characterized in that The at least two media domain types include a media domain type S i , i is a positive integer; generating interest weight data of the object for each media field type based on the interactive operation of the object on media data under at least two media field types within a target time period in the log data, including: Based on the object in the log data for the media domain type S within the target time period i generating a first dimension parameter based on a behavioral relationship between the interactive behavior of the media data under the target time period and the interactive behavior of the object with respect to the historical media data sets under the at least two media domain types within the target time period; Based on the object in the log data for the media domain type S within the target time period i generating a second dimension parameter based on a duration relationship between the browsing duration of the media data under the target time period and the browsing duration of the historical media data set by the object within the target time period; Based on the object in the log data for the media domain type S within the target time period i generating a third dimension parameter based on a frequency relationship between the interaction frequency of the media data under the target time period and the interaction frequency of the object with respect to the historical media data set within the target time period; According to the first dimension parameter, the second dimension parameter and the third dimension parameter, the object is generated for the media field type S i Interest weight data.

3. The method according to claim 2, characterized in that The first dimension parameter includes a behavior impact parameter; the object in the log data is targeted at the media domain type S within the target time period. i The first dimension parameter is generated based on the behavioral relationship between the interactive behavior of the media data under the target time period and the interactive behavior of the object with respect to the historical media data sets under the at least two media domain types within the target time period, including: Obtain the object in the log data for the media domain type S within the target time period i The first time decay factor, the first behavior type relevance and the first behavior interaction depth of the media data under the first time decay factor are used to characterize the object for the media domain type S i The first behavior type relevance is used to characterize the media data involved in the interactive behavior of the object and the media domain type S i The first behavior interaction depth is used to represent the influence of the interactive behavior of the object; Based on the first time decay factor, the first behavior type relevance and the first behavior interaction depth, a target for the media domain type S is generated. i First interactive behavior data of the media data under; Obtaining a second time decay factor, a second behavior type relevance, and a second behavior interaction depth for a historical media dataset of at least two media field types for the object in the log data within the target time period; generating second interaction behavior data for the at least two media domain types based on the second time decay factor, the second behavior type relevance, and the second behavior interaction depth; The ratio of the first interactive behavior data to the second interactive behavior data is determined as the behavior influence parameter.

4. The method according to claim 2, characterized in that The first dimension parameter includes an intensity dimension parameter; the object in the log data is based on the target time period for the media field type S i The first dimension parameter is generated based on the behavioral relationship between the interactive behavior of the media data under the target time period and the interactive behavior of the object with respect to the historical media data sets under the at least two media domain types within the target time period, including: Obtain the object in the log data for the media domain type S within the target time period i a plurality of first single interaction intensity data of the media data under the image, summing the plurality of first single interaction intensity data to obtain first interaction intensity data; the first single interaction intensity data is used to characterize the influence degree of the interactive behavior of the object, and different interactive behaviors correspond to different first single interaction intensity data; Acquire a plurality of second single interaction intensity data of the object for the historical media dataset within the target time period in the log data, and sum the plurality of second single interaction intensity data to obtain second interaction intensity data; A ratio of the first interaction intensity data to the second interaction intensity data is determined as the intensity dimension parameter.

5. The method according to claim 2, characterized in that The object in the log data is used for the media domain type S in the target time period. i The second dimension parameter is generated by the browsing time relationship between the browsing time of the media data under the target time period and the browsing time of the historical media data set by the object within the target time period, including: Obtain the object in the log data for the media domain type S within the target time period i a plurality of first single browsing durations of the media data under the control of the user, and summing the plurality of first single browsing durations to obtain a first browsing duration; Acquire multiple second single browsing durations of the object in the log data for the historical media dataset within the target time period, and sum the multiple second single browsing durations to obtain a second browsing duration; The ratio of the first browsing time to the second browsing time is determined as a second dimension parameter.

6. The method according to claim 2, characterized in that The object in the log data is used for the media domain type S in the target time period. i The frequency relationship between the interaction frequency of the media data under the target time period and the interaction frequency of the object with respect to the historical media data set within the target time period is used to generate a third dimension parameter, including: Obtain the object in the log data for the media domain type S within the target time period i a first interaction number of the media data under the target time period, and a second interaction number of the object in the log data with respect to the historical media data set within the target time period; The ratio of the first interaction times to the second interaction times is determined as a third dimension parameter.

7. The method according to claim 2, characterized in that Generating the object's interest weight data for each media field type according to the first dimension parameter, the second dimension parameter, and the third dimension parameter includes: Obtain at least two adjustment parameters; Based on the at least two adjustment parameters, the first dimension parameter, the second dimension parameter and the third dimension parameter are weightedly summed to obtain the object for the media domain type S i interest weight data; the at least two adjustment parameters are used to determine the degree of influence of the first dimension parameter, the second dimension parameter and the third dimension parameter on the interest weight data.

8. The method according to claim 7, characterized in that The first dimension parameters include a behavior impact parameter related to the real-time behavior and an intensity dimension parameter for characterizing interaction intensity; the method further includes: If it is detected that the data volume of the log data is less than the data volume threshold, reducing the adjustment parameters of the at least two adjustment parameters respectively associated with the behavior impact parameter and the second dimension parameter, and synchronously increasing the adjustment parameters of the at least two adjustment parameters respectively associated with the third dimension parameter and the intensity dimension parameter; If it is detected that the data volume of the log data is greater than or equal to the data volume threshold, the adjustment parameters of the at least two adjustment parameters respectively associated with the behavior impact parameter and the second dimension parameter are increased, and the adjustment parameters of the at least two adjustment parameters respectively associated with the third dimension parameter and the intensity dimension parameter are reduced simultaneously.

9. The method according to claim 1, characterized in that The at least two media domain types include a target media domain type; performing interest matching in a candidate media data set based on interest weight data corresponding to the at least two media domain types to obtain a media push list for the object includes: If the interest weight data related to the target media field type corresponding to the object is greater than a weight threshold, the interest weight data related to the target media field type corresponding to at least two historical objects are determined as target interest weight data, and the historical objects corresponding to the target interest weight data greater than or equal to the weight threshold are determined as candidate objects; Obtaining a first historical interaction behavior of the candidate object with respect to the candidate media data set, and obtaining a second historical interaction behavior of the object in the log data; determining an object similarity between the candidate object and the object according to the first historical interaction behavior and the second historical interaction behavior; If the object similarity is greater than a similarity threshold, a media push list for the object is generated based on the media data with which the candidate object has interacted and with which the object has not interacted.

10. The method according to claim 1, characterized in that The at least two media domain types include a target media domain type; performing interest matching in a candidate media data set based on interest weight data corresponding to the at least two media domain types to obtain a media push list for the object includes: If the interest weight data corresponding to the object and related to the target media field type is greater than a weight threshold, performing feature fusion on the media data under the target media field type with which the object has interacted to obtain a target feature vector; Performing feature extraction on each media data under the target media domain type in the candidate media data set to obtain a set of candidate feature vectors; A similarity calculation is performed between the target feature vector and each candidate feature vector in the candidate feature vector set, and a media push list for the object is generated based on media data corresponding to candidate feature vectors having similarities greater than or equal to a similarity threshold.

11. The method according to claim 1, wherein The adjusting at least two weight ratios based on the key interaction operation, and updating the media push list according to the at least two adjusted weight ratios, includes: If the key interactive operation is a forward interactive operation, then the weight ratio of the key media domain type is increased, and the weight ratios other than the weight ratio of the key media domain type in at least two weight ratios are simultaneously reduced; the key media domain type refers to the media domain type to which the media data operated by the key interactive operation belongs; Based on the at least two adjusted weight ratios, the amount of media data in the media push list belonging to the key media domain type is increased, and the amount of media data in the media push list belonging to other media domain types is simultaneously reduced; the other media domain types refer to media domain types other than the key media domain type in the at least two media domain types; If the key interactive operation is a negative interactive operation, the weight ratio of the key media field type is reduced, and the other weight ratios are increased simultaneously; Based on at least two adjusted weight ratios, the amount of media data belonging to the key media domain type in the media push list is reduced, and the amount of media data belonging to other media domain types in the media push list is increased simultaneously.

12. The method according to claim 1, characterized in that The adjusting of at least two weight ratios based on the key interaction operation includes: If the key interactive operation is a trigger operation for a feedback control, obtaining a ratio adjustment value according to the feedback intent indicated by the feedback control, performing a first ratio adjustment on the weight ratio of the key media domain type in at least two weight ratios based on the ratio adjustment value, and performing a second ratio adjustment on the other weight ratios in the at least two weight ratios except the weight ratio of the key media domain type; the key media domain type refers to the media domain type to which the media data operated by the key interactive operation belongs; If the key interactive operation is a sharing operation on a social platform, the platform priority of the shared social platform is obtained, and the weight ratio of the key media field type is increased based on the proportion mapped by the platform priority, while the other weight ratios are reduced simultaneously.

13. The method according to claim 1, wherein The updating of the media push list according to the at least two adjusted weight ratios includes: Interest matching is performed in the candidate media dataset according to at least two adjusted weight ratios to obtain an updated media push list for the object, and the media push list is replaced with the updated media push list.

14. The method according to claim 1, wherein Also includes: Obtaining playback network status of target media data for at least two participants within a geographical area; The at least two participating objects include the object, and the media push list includes the target media data; If it is detected that the number of participating objects whose playback network status is a network freeze state is greater than the object quantity threshold, the definition recommended version for the target media data in the media push list is adjusted.

15. A data processing device, characterized in that: include: The transceiver module is used to obtain log data for the object; a data generating module, configured to generate interest weight data of the object for each media field type based on interactive operations of the object on media data of at least two media field types within a target time period in the log data; A list generation module is configured to perform interest matching in the candidate media data set based on the interest weight data corresponding to the at least two media field types, and obtain a media push list for the object; The media data in the media push list corresponding to the media field type with the largest interest weight accounts for the largest proportion; a ratio generating module configured to, upon detecting a key interactive operation of the object on the media data of the at least two media domain types, normalize the interest weight data corresponding to the at least two media domain types to obtain weight ratios corresponding to the at least two media domain types; The list updating module is configured to adjust at least two weight ratios based on the key interaction operation, and update the media push list according to the at least two adjusted weight ratios.

16. A computer device, characterized in that: include: processor, memory, and network interface; The processor is connected to the memory and the network interface, wherein the network interface is used to provide a data communication function, the memory is used to store a computer program, and the processor is used to call the computer program so that the computer device executes the method according to any one of claims 1 to 14.

17. A computer-readable storage medium, characterized in that The computer-readable storage medium stores a computer program, which is suitable for being loaded and executed by a processor, so that a computer device having the processor executes the method according to any one of claims 1 to 14.

18. A computer program product, characterized in that The computer program product comprises a computer program, which is stored in a computer-readable storage medium and is suitable for being read and executed by a processor, so as to enable a computer device having a processor to perform the method according to any one of claims 1 to 14.

Images