Live broadcast service data processing method and device, equipment and medium

Abstract

The application relates to the network live broadcast technical field and discloses a live broadcast service data processing method and device, equipment and a medium. The method comprises the following steps: acquiring a data definition template corresponding to a live broadcast service, wherein the data definition template contains data items distributed in multiple data sources and required by the live broadcast service and operation items used for performing predefined operations on the data items, the data items contain attribute data items of anchor users, and the operation items comprise attribute labels; according to the operation items, performing the predefined operations corresponding to the operation items on the corresponding data items in the multiple data sources, determining a data set corresponding to the attribute labels in the operation items, and the data set comprises a data item subset corresponding to multiple anchor users; and pushing the data set corresponding to the live broadcast service to a terminal device. The application realizes standardized operations of the data items of multiple live broadcast services calling multiple data sources through the data definition template, avoids data entropy increase, improves system efficiency and saves system overhead.

Description

Technical Field

[0001] The present application relates to the field of network live broadcast technology, and in particular to a live broadcast service data processing method and its corresponding device, computer equipment and computer-readable storage medium. Background Art

[0002] Online live streaming platforms that provide online live streaming services usually embed multiple live streaming services in the access website or application of their platform. Each live streaming service usually corresponds to an access page, which is used to provide users with data sets corresponding to the same type of service, so that terminal devices can parse specific data according to the data sets for interface display.

[0003] On the same platform, data comes from multiple sources, and data access protocols are diverse. Different data originates from different databases or tables and may require different protocols for access. Furthermore, live streaming services are also diverse, requiring different data items and processing results. Consequently, as the platform develops, according to the law of increasing entropy, platform data becomes increasingly disorganized, reducing platform operational efficiency and data access efficiency, impacting the user experience.

[0004] The deeper problem is that the complex and disordered data will make it much more difficult for new live broadcast businesses to obtain the data they need, thereby further increasing the difficulty of developing new businesses on the platform.

[0005] In view of the above aspects, it is necessary to improve the processing method of data required for the live broadcast business in the live broadcast platform. Summary of the Invention

[0006] The primary purpose of this application is to solve at least one of the above problems and provide a live broadcast business data processing method and its corresponding device, computer equipment and computer-readable storage medium.

[0007] In order to meet the various objectives of this application, this application adopts the following technical solutions:

[0008] A live broadcast service data processing method proposed to meet one of the purposes of this application includes the following steps:

[0009] Obtaining a data definition template corresponding to a live broadcast service, wherein the data definition template includes data items distributed across multiple data sources and required for invoking the live broadcast service, and operation items for performing predefined operations on the data items. The data items include attribute data items of the anchor user, and the operation items include one or more attribute tags.

[0010] According to the operation item in the data definition template, a predefined operation corresponding to the operation item is performed on the corresponding data item in the plurality of data sources, and a data set corresponding to the attribute label in the operation item is determined, the data set including a subset of data items corresponding to a plurality of anchor users;

[0011] In response to a data calling instruction of a terminal device for a live broadcast service, the terminal device is pushed a data set corresponding to the live broadcast service.

[0012] In a deepened embodiment, according to the operation item in the data definition template, a predefined operation corresponding to the operation item is performed on the corresponding data item in the plurality of data sources, and a data set corresponding to the attribute label in the operation item is determined, including the following steps:

[0013] The data definition template is parsed to determine the data items and operation items in each data source;

[0014] The data items in each data source are obtained by applying a distributed lock call, and a subset of data items of each anchor user is determined according to the data items associated with the same anchor user;

[0015] An intelligent classification model pre-trained to a converged state is used to determine the attribute label of each anchor user according to each subset of data items;

[0016] An anchor user data item subset having an attribute label matching the attribute label in the operation item is screened out;

[0017] Each subset of data items of each anchor user is formatted to construct a data set in a standardized format.

[0018] In an extended embodiment, the intelligent classification model is pre-iteratively trained to a converged state, and the training process includes the following steps:

[0019] A single training sample is called from a training data set, the training sample including attribute data corresponding to a plurality of attribute data items of an anchor user;

[0020] The attribute data in the training sample is vectorized to obtain a sample vector;

[0021] The sample vector is input into the intelligent classification model for semantic extraction and classification mapping to obtain an attribute label predicted by the classification;

[0022] The loss value of the attribute label predicted by the intelligent classification model is calculated according to the attribute label corresponding to the training sample;

[0023] The intelligent classification model is gradient updated according to the loss value, or iteratively trained until the model reaches a converged state.

[0024] In some extended embodiments, after the step of filtering out a subset of anchor user data items whose attribute tags match the attribute tags in the operation items, the following steps are included:

[0025] Obtaining popularity data of each attribute tag corresponding to each anchor user, wherein the popularity data is determined based on user behavior data statistics of the anchor user's live broadcast room;

[0026] The data item subsets of each anchor user are sorted according to the popularity data so that the subsequently generated data sets maintain the corresponding sorting.

[0027] In some extended embodiments, before the step of obtaining the popularity data of each attribute tag corresponding to each anchor user, the following steps are included:

[0028] Obtaining user behavior data associated with each live broadcast user, wherein the user behavior data corresponds to descriptive data generated by corresponding access events of the live broadcast room of the live broadcast user being followed by a user, receiving a gift from a user, or being entered by a user;

[0029] The user behavior data is counted to obtain the user popularity corresponding to each anchor user, where the user popularity is a weighted statistical result of the number of multiple access events;

[0030] For each attribute tag in the preset attribute tag system, the user popularity of the anchor user carrying the attribute tag is accumulated to obtain the accumulated popularity corresponding to the attribute tag;

[0031] Normalization is performed based on the accumulated heat of each attribute tag in the attribute tag system to obtain heat data corresponding to each attribute tag.

[0032] In some extended embodiments, after the step of obtaining the data definition template corresponding to the live broadcast service, the following steps are included:

[0033] According to the timing task trigger, the specified data items in the data definition template corresponding to the live broadcast business are dispatched from their corresponding data sources to the secondary cache for calling;

[0034] The data set obtained by performing operations on the data items of the data source in the second-level cache is stored in the first-level cache.

[0035] In some extended embodiments, before the step of obtaining the data definition template corresponding to the live broadcast service, the following steps are included:

[0036] Run the data source adaptation service and open the interface to the external data source to enable the external data source to access and participate in providing the data of the said data set for the live broadcast business.

[0037] A live broadcast service data processing apparatus provided for adapting to one of the purposes of the present application, comprising a template calling module, a data processing module, and a data pushing module, wherein: the template calling module is configured to obtain a data definition template corresponding to a live broadcast service, the data definition template contains data items distributed in multiple data sources required to be called by the live broadcast service and operation items for performing predefined operations on the data items, the data items contain attribute data items of anchor users, and the operation items include one or more attribute labels; the data processing module is configured to perform the predefined operations corresponding to the operation items on the corresponding data items in the multiple data sources according to the operation items in the data definition template, and determine a data set corresponding to the attribute labels in the operation items, the data set including data item subsets corresponding to multiple anchor users; and the data pushing module is configured to push the corresponding data set of the live broadcast service to a terminal device in response to a data calling instruction of the live broadcast service of the terminal device.

[0038] In a deepened partial embodiment, the data processing module comprises: a template analysis unit configured to analyze the data definition template to determine the data items and operation items in each data source; a subset processing unit configured to apply a distributed lock call to obtain the data items in the each data source, and determine data item subsets of each anchor user according to the data items associated with the same anchor user; a label determination unit configured to determine attribute labels of each anchor user according to each data item subset by using an intelligent classification model pre-trained to a convergence state; an anchor screening unit configured to screen anchor user data item subsets with attribute labels matching the attribute labels in the operation items; and a format unification unit configured to format the data item subsets of each anchor user to construct a data set in a standardized format.

[0039] In an extended partial embodiment, the intelligent classification model is pre-iteratively trained to a convergence state in a training module, and the training module comprises: a sample calling unit configured to call a single training sample from a training data set, the training sample including attribute data corresponding to multiple attribute data items of an anchor user; a vector encoding unit configured to vectorize the attribute data in the training sample to obtain a sample vector; a classification mapping unit configured to input the sample vector into the intelligent classification model to perform semantic extraction and classification mapping, and obtain an attribute label predicted by classification; a loss calculation unit configured to calculate a loss value of the attribute label predicted by the intelligent classification model according to the attribute label corresponding to the training sample; and an iteration decision unit configured to implement gradient update on the intelligent classification model according to the loss value, or continue to iteratively train until the model reaches a convergence state.

[0040] In some extended embodiments, the data processing module includes the following units that run after the anchor screening unit: a popularity calling unit, used to obtain popularity data of each attribute label corresponding to each anchor user, and the popularity data is determined based on user behavior data statistics of the anchor user's live broadcast room; a sorting processing unit, used to sort a subset of data items of each anchor user according to the popularity data, so that the subsequently generated data sets maintain the corresponding sorting.

[0041] In some extended embodiments, the data processing module includes the following units that run before the popularity calling unit: a data description unit, used to obtain user behavior data associated with each anchor user, and the user behavior data corresponds to the description data generated by the corresponding access events of the anchor user's live broadcast room being followed by users, receiving gifts from users, and being entered by users; a data statistics unit, used to count the user behavior data and obtain the user popularity corresponding to each anchor user, and the user popularity is a weighted statistical result of the number of multiple access events; a popularity accumulation unit, used to correspond to each attribute tag in a preset attribute tag system, accumulate the user popularity of the anchor user carrying the attribute tag, and obtain the accumulated popularity corresponding to the attribute tag; a popularity determination unit, used to normalize according to the accumulated popularity of each attribute tag in the attribute tag system, and obtain the popularity data corresponding to each attribute tag.

[0042] In some extended embodiments, the live broadcast business data processing device of the present application also includes the following modules that run after the template calling module: a secondary cache module, which is used to dispatch the specified data items in the data definition template corresponding to the live broadcast business from its corresponding data source to the secondary cache for calling according to the timed task trigger; a primary cache module, which is used to store the data set obtained by performing operations based on the data items of the data source in the secondary cache in the primary cache.

[0043] In some extended embodiments, the live broadcast service data processing device of the present application also includes a data adaptation module that runs before the template calling module, which is used to run the data source adaptation service and open an interface to the external data source to enable external data source access and participate in providing the data of the data set for the live broadcast service.

[0044] A computer device provided to meet one of the purposes of this application includes a central processing unit and a memory, wherein the central processing unit is used to call and run a computer program stored in the memory to execute the steps of the live broadcast business data processing method described in this application.

[0045] A computer-readable storage medium is provided to meet another purpose of the present application, which stores a computer program implemented according to the live broadcast business data processing method in the form of computer-readable instructions. When the computer program is called and executed by a computer, the steps included in the method are executed.

[0046] A computer program product provided to meet another purpose of the present application includes a computer program / instruction, which, when executed by a processor, implements the steps of the method described in any embodiment of the present application.

[0047] Compared with the prior art, this application has many technical advantages, including but not limited to the following aspects:

[0048] First, this application utilizes the correspondence between live broadcast services and data definition templates to achieve standardized representation of data required for different live broadcast services through data items and operation items in the data definition templates. The data items include attribute data items corresponding to each anchor user, and the operation items include one or more attribute tags. Thus, the operation items can play the role of predefined operations. Subsequently, relevant data items corresponding to the live broadcast service are obtained according to the data definition template. Based on the association between the attribute data items and the attribute tags, a data set containing a subset of data items of multiple anchor users is determined. That is, a subset of data items corresponding to anchor users who have a certain predefined operational correspondence with the attribute tags. Therefore, the data item subset plays the role of associating descriptions of anchor users. When a terminal device needs to load the corresponding page of the live broadcast service and initiates a request to the server, the data set can be pushed to the corresponding terminal device, so that the corresponding terminal device displays key information of each anchor user based on the subset of data items of the anchor user in the data set.

[0049] Secondly, in this application, the data definition template realizes the centralized definition of data items of multiple data sources on the one hand, and on the other hand, it is pre-associated with the live broadcast business, thereby realizing the mapping and sorting of the correspondence between the data source and the live broadcast business. For the live broadcast platform, developers can maintain the orderly call of data items of multiple data sources by maintaining the data definition template, thereby serving the corresponding live broadcast business and processing the data. The live broadcast business and the data items required by it are always managed in an orderly manner, and there is no need to obtain data items separately. This can not only effectively avoid the situation where each data source repeatedly obtains the same data items required by different live broadcast businesses, minimizing the impact of the law of increasing entropy, but also make it more convenient for developers to maintain the data logic between the data items required to be called by each live broadcast business.

[0050] In addition, since this application sorts out the relationship between the live broadcast business and the data it needs to call, and optimizes the resource usage of various links such as data storage and access, it can save the storage and running load of the service cluster, thereby saving platform system overhead and platform deployment costs. BRIEF DESCRIPTION OF THE DRAWINGS

[0051] The above and / or additional aspects and advantages of the present application will become apparent and easily understood from the following description of the embodiments in conjunction with the accompanying drawings, in which:

[0052] Figure 1 This is a flowchart of a typical embodiment of the live broadcast service data processing method of the present application;

[0053] Figure 2 and Figure 3 Both are exemplary graphical user interfaces, showing the interface effects of the live broadcast services corresponding to "Guess What You Like" and "Popular" respectively;

[0054] Figure 4 This is a flowchart of a process for obtaining a data set corresponding to a live broadcast service according to a data definition template in an embodiment of the present application;

[0055] Figure 5 This is a flowchart of an exemplary intelligent classification model training process in an embodiment of the present application;

[0056] Figure 6 A flowchart of a process for determining heat data of each attribute tag in an embodiment of the present application;

[0057] Figure 7 This is a principle block diagram of the live broadcast service data processing device of this application;

[0058] Figure 8 A schematic diagram of the structure of a computer device used in this application; DETAILED DESCRIPTION

[0059] The following describes in detail embodiments of the present application, examples of which are shown in the accompanying drawings, wherein the same or similar reference numerals throughout represent the same or similar elements or elements having the same or similar functions. The embodiments described below with reference to the accompanying drawings are exemplary and are only used to explain the present application, and are not to be construed as limiting the present application.

[0060] It will be understood by those skilled in the art that, unless expressly stated otherwise, the singular forms "a", "an", "said" and "the" used herein may also include the plural forms. It should be further understood that the term "comprising" used in the specification of the present application refers to the presence of the features, integers, steps, operations, elements and / or components, but does not exclude the presence or addition of one or more other features, integers, steps, operations, elements, components and / or groups thereof. It should be understood that when we refer to an element as being "connected" or "coupled" to another element, it may be directly connected or coupled to the other element, or there may be intermediate elements. In addition, "connected" or "coupled" as used herein may include wireless connections or wireless couplings. The term "and / or" used herein includes all or any units and all combinations of one or more associated listed items.

[0061] It will be understood by those skilled in the art that, unless otherwise defined, all terms (including technical and scientific terms) used herein have the same meaning as commonly understood by those skilled in the art to which this application belongs. It should also be understood that terms such as those defined in common dictionaries should be understood to have meanings consistent with their meanings in the context of the prior art and will not be interpreted in an idealized or overly formal sense unless specifically defined as herein.

[0062] Those skilled in the art will appreciate that the term "client", "terminal", "terminal device" as used herein encompasses devices that are solely wireless signal receivers, devices that are wireless signal receivers with no transmit capability, and devices that are both receivers and transmitters capable of bi-directional communication over a bi-directional communication link. Such devices can include cellular or other communication devices with or without a multi-line display, a plurality of push-to-talk buttons, and / or a numeric keypad. Such devices can also include Personal Communications Service (PCS) devices, Personal Digital Assistants (PDAs), cellular telephone / PDA combinations, Internet / Intranet access devices, and / or other devices that are capable of receiving wireless signals. The term "client", "terminal", "terminal device" as used herein can be portable, transportable, installed in a vehicle (aeronautical, maritime, and / or land-based), or adapted to be operated locally and / or in a distributed manner on Earth and / or in any other location in space. The term "client", "terminal", "terminal device" as used herein can also be a communication terminal, an Internet access terminal, a music / video playing terminal, such as a PDA, a Mobile Internet Device (MID), and / or a mobile phone with music / video playing function, a smart television, a set-top box, and / or the like.

[0063] The term "server", "client", "service node", and the like as used herein refers to hardware that has the equivalent capability of a personal computer, and is essentially an electronic device with a central processing unit (including an arithmetic unit and a controller), a memory, an input device, and an output device, and the like necessary components disclosed by the Von Neumann principle. A computer program is stored in the memory, the central processing unit calls the program stored in the external memory into the memory to run, executes the instructions in the program, and interacts with the input and output devices, thereby completing a specific function.

[0064] It should be noted that the concept of "server" referred to in this application can also be extended to server clusters. Based on the network deployment principles understood by those skilled in the art, the servers described should be logically divided. In physical space, these servers can be independent of each other but callable through interfaces, or integrated into a single physical computer or a computer cluster. Those skilled in the art should understand this flexibility and should not use it to constrain the implementation of the network deployment method of this application.

[0065] Unless expressly specified, one or more technical features of the present application can be deployed on a server for implementation and accessed by a client through a remote call to obtain an online service interface provided by the server, or can be directly deployed and run on a client for implementation.

[0066] Unless expressly specified otherwise, the neural network models referenced or may be referenced in this application may be deployed on a remote server and remotely called on the client, or may be deployed and directly called on a client with sufficient device capabilities. In some embodiments, when it runs on the client, its corresponding intelligence may be obtained through transfer learning to reduce the requirements for the client's hardware operating resources and avoid excessive occupation of the client's hardware operating resources.

[0067] Unless explicitly specified, the various data involved in this application can be stored remotely on a server or on a local terminal device, as long as they are suitable for being called by the technical solution of this application.

[0068] Those skilled in the art should be aware that although the various methods of this application are described based on the same concept and thus exhibit commonality, unless otherwise specified, these methods can be independently executed. Similarly, the various embodiments disclosed in this application are all based on the same inventive concept. Therefore, concepts with the same expression, as well as concepts that are appropriately transformed for convenience despite different expression, should be understood as equivalent.

[0069] Unless expressly stated to be mutually exclusive, the various embodiments disclosed in this application may be cross-combined with the relevant technical features of the various embodiments to flexibly construct new embodiments, as long as such combination does not deviate from the creative spirit of this application and can meet the needs of the prior art or resolve certain deficiencies in the prior art. Those skilled in the art should be aware of such flexibility.

[0070] A live broadcast business data processing method of the present application can be programmed as a computer program product and deployed in a computer device for execution. The method can be executed by accessing an interface opened after the computer program product is running and conducting human-computer interaction with the process of the computer program product through a graphical user interface.

[0071] See also Figure 1 In a typical embodiment, the live broadcast service data processing method of the present application includes the following steps:

[0072] Step S1100: Obtain a data definition template corresponding to the live broadcast service. The data definition template includes data items distributed across multiple data sources required for the live broadcast service and operation items for performing predefined operations on these data items. The data items include attribute data items of the anchor user, and the operation items include one or more attribute tags:

[0073] In the online live broadcast platform, the collection, storage, and call of data items from various data sources inside and outside the platform are unified and scheduled, and the corresponding data items are displayed through the corresponding page code of the live broadcast business implemented by the development team.

[0074] The data source includes a variety of databases suitable for database engines, such as HBase, MySQL, and MongoDB. For large-scale online live broadcast platforms, the database is usually a database that supports distributed storage. Each database may contain multiple pre-defined data tables, each data table is used to store multiple data items, and data connections can be achieved across databases and data tables through data item correspondences, such as through the UID of the device or user, or by associating with the same attribute tags. Therefore, in the online platform, the personal information related to an anchor user may be distributed in multiple data items in multiple data tables of multiple data sources, which can be a relatively complex tree or network relationship. The corresponding part of the data items of the personal information is an attribute data item, which is used to store the attribute data corresponding to the corresponding anchor user, for example, storing the age, gender, live broadcast theme description information, personal hobby keywords, etc. of the anchor user.

[0075] The live broadcast service usually corresponds to a page that users of the live broadcast platform can access. For example, the user accesses the website or application of the live broadcast platform on the terminal device. Figure 2 When the "Guess You Like" page is displayed, a corresponding live broadcast service is called. At this time, this live broadcast service is used to recommend video covers of multiple anchor users based on the user's personal preferences, so that the user can quickly enter the live broadcast room of one of the anchor users. Figure 3 Similarly, when the "popular" page is shown, a video cover of a currently popular anchor user is pushed to the user, making it convenient for the user to enter one of the live broadcast rooms.

[0076] Online live streaming platforms often offer multiple websites and applications that provide various services to platform users. Furthermore, a single website or application often provides multiple live streaming services. The underlying data for all of these live streaming services is provided by data sources maintained or accessible by the platform. Therefore, building a bridge between live streaming services and data sources, and maintaining the relationships between live streaming services and complexly distributed data items, helps improve development efficiency and data maintenance effectiveness. Therefore, a data maintenance platform can be used to provide customized services for the development of various live streaming services. Through this customized service, the development team can create and edit the data items required for a particular live streaming service, forming a corresponding data definition template. Later, when implementing the corresponding live streaming service, this data definition template can be called to obtain the desired data set.

[0077] The data definition template in this application not only includes the designation of the data items required for the live broadcast business, but also includes operation items for indicating the implementation of specified predefined operations on these data items. The data items may be distributed across multiple different data sources, and the operation items may specify the information contained therein through various pre-agreement methods. In one example of this application, the operation items may be parsed by this application as predicting the corresponding attribute labels of the data items associated with each anchor user by including one or more attribute labels, and then using the mapping relationship between the attribute labels of the operation items and the predicted attribute labels, a subset of the data items of some anchor users is matched to construct a data set. According to this example, it can be seen that the predefined operations have been agreed upon in advance and can be parsed as a matching operation on the data items.

[0078] As mentioned above, among the data items associated with each anchor user, a part of them are attribute data items, which are particularly suitable for participating in the operations defined by the operation items, while the rest can be other types of data items, which can serve the needs of the live broadcast business background, including participating in the operations defined by the operation items, and are not limited by this application.

[0079] When the online live broadcast platform needs to prepare data sets for each live broadcast business, it obtains the data definition template corresponding to each live broadcast business according to the predetermined trigger mechanism, and then performs a pre-protocol analysis on the data definition template to determine the data items and operation items therein for future use.

[0080] Step S1200: Based on the operation items in the data definition template, predefined operations corresponding to the operation items are performed on the corresponding data items in the multiple data sources to determine a data set corresponding to the attribute tags in the operation items, wherein the data set includes a subset of data items corresponding to multiple anchor users:

[0081] As mentioned above, an operation item can represent a predefined operation that is parsed to perform a match on the data items in the data definition template by including one or more attribute tags. Therefore, based on the operation items in the data definition template corresponding to the live broadcast business, that is, one or more attribute tags therein, it is analyzed whether the data items matching each anchor user are associated with the one or more attribute tags. When such an association exists, the data item subset consisting of the data items corresponding to the anchor user is filtered out, and finally the filtered data item subsets are constituted into a data set.

[0082] The matching relationship between the operation item and the attribute label associated with the host user's data item can be confirmed by analyzing whether the host user's data item contains the same attribute label specified by the operation item. It can also be confirmed by using a neural network model pre-trained to a convergent state to perform classification mapping on each host user's data item to predict the attribute label, and then analyzing whether the predicted attribute label corresponds to the attribute label in the operation item. The method of implementing this matching operation is relatively flexible. Those skilled in the art can flexibly choose other means to replace it based on the principles and examples disclosed in this application, without affecting the embodiment of the inventive spirit of this application.

[0083] After the data set is determined, the data set contains a subset of data items corresponding to multiple anchor users. The data item subset includes one or more attribute data items. The attribute data items are generally data items suitable for being directly displayed on the page corresponding to the live broadcast business.

[0084] In order to facilitate the use of a unified code to parse and call the data set, the data set can be formatted according to a pre-set unified format.

[0085] Step S1300: respond to the terminal device's data call instruction for the live broadcast service, and push the corresponding data set of the live broadcast service to the terminal device:

[0086] After completing the preparation of the data sets for each live broadcast service through the aforementioned steps, the data sets are in a callable state, and steps S1100 and S1200 can be executed cyclically to achieve dynamic update of the data sets.

[0087] When a large number of platform users log in to the website or application of the live broadcast platform on their respective terminal devices and need to load or preload the page corresponding to the live broadcast service, they can initiate a live broadcast service loading request to the server. In response to the request, the server calls the corresponding data set based on the correspondence between the live broadcast service and the data set, and pushes it to the terminal device that triggered the request, thereby displaying the anchor user information described by each data item subset in the data set in the graphical user interface of the terminal device, including the anchor user's personal information, etc. Figure 2 and Figure 3 The page shown can display the video covers corresponding to multiple anchor users.

[0088] When the data set is encapsulated in a unified format, as long as the background code corresponding to the page in the terminal device parses the data set according to the unified format, the same subset of anchor user data items can be obtained, regardless of the application category and website category used to load the page. Therefore, for multiple networks and multiple applications developed by the online live broadcast platform, as long as they comply with the same format protocol, they can use the data set in the unified format according to their respective business logic to make personalized data presentations. As a result, during the code implementation process of the live broadcast business page corresponding to the terminal device, there is no need to directly face the underlying data or directly operate the underlying data source, and efficient data access and call operations can be achieved.

[0089] According to this exemplary embodiment, it is not difficult to understand that compared with the prior art, this application has many technical advantages, including but not limited to the following aspects:

[0090] First, this application utilizes the correspondence between live broadcast services and data definition templates to achieve standardized representation of data required for different live broadcast services through data items and operation items in the data definition templates. The data items include attribute data items corresponding to each anchor user, and the operation items include one or more attribute tags. Thus, the operation items can play the role of predefined operations. Subsequently, relevant data items corresponding to the live broadcast service are obtained according to the data definition template. Based on the association between the attribute data items and the attribute tags, a data set containing a subset of data items of multiple anchor users is determined. That is, a subset of data items corresponding to anchor users who have a certain predefined operational correspondence with the attribute tags. Therefore, the data item subset plays the role of associating descriptions of anchor users. When a terminal device needs to load the corresponding page of the live broadcast service and initiates a request to the server, the data set can be pushed to the corresponding terminal device, so that the corresponding terminal device displays key information of each anchor user based on the subset of data items of the anchor user in the data set.

[0091] Secondly, in this application, the data definition template realizes the centralized definition of data items of multiple data sources on the one hand, and on the other hand, it is pre-associated with the live broadcast business, thereby realizing the mapping and sorting of the correspondence between the data source and the live broadcast business. For the live broadcast platform, developers can maintain the orderly call of data items of multiple data sources by maintaining the data definition template, thereby serving the corresponding live broadcast business and processing the data. The live broadcast business and the data items required by it are always managed in an orderly manner, and there is no need to obtain data items separately. This can not only effectively avoid the situation where each data source repeatedly obtains the same data items required by different live broadcast businesses, minimizing the impact of the law of increasing entropy, but also make it more convenient for developers to maintain the data logic between the data items required to be called by each live broadcast business.

[0092] In addition, since this application sorts out the relationship between the live broadcast business and the data it needs to call, and optimizes the resource usage of various links such as data storage and access, it can save the storage and running load of the service cluster, thereby saving platform system overhead and platform deployment costs.

[0093] See also Figure 4 In some further embodiments, step S1200, performing predefined operations corresponding to the operation items on corresponding data items in multiple data sources according to the operation items in the data definition template, and determining data sets corresponding to the attribute tags in the operation items, includes the following steps:

[0094] Step S1210: Parse the data definition template to determine the data items and operation items in each data source.

[0095] For the data definition template obtained based on the live broadcast business call, it can be parsed accordingly according to the pre-agreement to determine the operation items and the data items distributed in each data source. For the server, the data items parsed from the data definition template and their corresponding data tables, data fields, etc. in the data source are all pre-defined and clear.

[0096] Step S1220: Apply a distributed lock call to obtain the data items in each data source, and determine a data item subset of each anchor user based on the data items being associated with the same anchor user:

[0097] In order to call specific data items, after determining the data items required for the live broadcast business, the corresponding data items can be called from various data sources, that is, the corresponding data can be obtained. As mentioned above, since the data stored in the data items are usually stored in association with the anchor user, a data item subset can be constructed for each anchor user with the anchor user as the clustering unit. This subset is composed of the data in the data items corresponding to the anchor user. When implementing the call to the data items in the data source, considering the risk of data access conflicts, distributed locks can be applied to the accessed data items for atomic operations, thereby ensuring that data disorder will not be caused during the data call process.

[0098] Step S1230: Using the intelligent classification model that has been pre-trained to a convergent state, determine the attribute labels of each corresponding anchor user based on each data item subset:

[0099] This embodiment utilizes a pre-trained intelligent classification model to perform classification mapping on a subset of data items for each anchor user, thereby predicting the attribute labels corresponding to each anchor user. The training process of the intelligent classification model will be disclosed in subsequent embodiments of this application and will not be discussed here.

[0100] The intelligent classification model, after being trained to a convergent state, acquires the representation learning ability to accurately extract deep semantic features from a subset of the anchor user's data items. On this basis, by classifying and mapping the deep semantic features, the various attribute labels belonging to the anchor user can be obtained. For example, in a scenario example, the attribute labels may be "popular anchor", "dance anchor", "sports anchor", "knowledge anchor", etc. The attribute labels predicted by the intelligent classification model can be confirmed to be one or more, for example, allowing an anchor user to be both a "popular anchor" and a "knowledge anchor" at the same time, and it can be flexibly implemented depending on whether the attribute labels are a multi-level system.

[0101] The input to the intelligent classification model depends on the data items used as model training samples during pre-training. In one embodiment, because the attribute data items correspond to the personal information of the anchor user, only the attribute data items corresponding to the anchor user can be used to construct the input. Of course, those skilled in the art can also use other data items of the anchor user to participate in the construction of the intelligent classification model input.

[0102] Step S1240: Filter out a subset of anchor user data items whose attribute tags match the attribute tags in the operation items:

[0103] For an anchor user, when the attribute label of the operation item is in one-to-one relationship with the attribute label predicted by the intelligent classification model and is completely the same, it is considered that the matching relationship defined by the operation item is realized, and the data item subset of the anchor user can be screened out for constructing the data set.

[0104] For an anchor user, when there are multiple attribute labels in the operation item and the intelligent classification model only predicts a single attribute label, and the predicted attribute label is one of the multiple attribute labels in the operation item, it is considered that the matching relationship defined by the operation item is realized, and the data item subset of the anchor user can be screened out for constructing the data set.

[0105] For an anchor user, when there is a single attribute label in the operation item and the intelligent classification model predicts multiple attribute labels, if one of the multiple attribute labels is the same as the single attribute label in the operation item, it is considered that the matching relationship defined by the operation item is realized, and the data item subset of the anchor user can be screened out for constructing the data set.

[0106] For an anchor user, when the operation item contains multiple attribute labels and the intelligent classification model also predicts multiple attribute labels, the set relationship can be used to confirm whether the two achieve matching, and a specific example is to investigate whether the multiple attribute labels in the operation item constitute a subset of the predicted multiple attribute labels. If so, it is considered that the two achieve the matching relationship defined by the operation item, and the data item subset of the anchor user can be screened out for constructing the data set.

[0107] It can be seen that the matching of the data item subset of the anchor user whether it realizes the matching relationship defined by the operation item has flexible implementation, and those skilled in the art can flexibly implement according to the various examples herein, without affecting the embodiment of the inventive spirit of the present application.

[0108] Step S1250, formatting the data item subset of each anchor user to construct a data set in a standardized format:

[0109] After the foregoing steps are processed, the data item subsets of each anchor user that meet the requirements of the live broadcast service are matched, and these data item subsets can be constructed into a data set in a standardized format. In order to realize the uniformity of the standardized format, the data item subset corresponding to each anchor user can be packaged in a unified format according to a preset format protocol, so that subsequent unified analysis can be performed according to the format protocol, and then the packaged data item subset is constructed into the data set.

[0110] In this embodiment, it is revealed that the data definition template implements predefined operations by encapsulating attribute tags in its operation items, thereby indicating that a subset of data items called from the data source is intelligently classified and mapped to predict corresponding attribute tags. Then, based on the matching operation indicated by the predefined operation of the operation item, a subset of data items of the anchor user required for the live broadcast business is selected according to whether the predicted attribute tags match the attribute tags of the operation items, and is constructed into a data set in a unified format. Accordingly, the function of the data definition template is enriched, so that the data definition template can not only realize the sorting of data items from multiple sources, but also reflect more complex business logic, instructing the server to perform intelligent analysis and matching on the data item subsets of each anchor user, thereby obtaining a data set that accurately corresponds to the data requirements of the live broadcast business, eliminating the tediousness of developers repeatedly developing complex business logic, while providing richer technical support, and greatly improving the efficiency of software engineering development.

[0111] See also Figure 5 In some extended embodiments, the intelligent classification model is pre-trained iteratively to a convergence state, and the training process includes the following steps:

[0112] Step S2100: Call a single training sample from the training data set, where the training sample includes attribute data corresponding to multiple attribute data items of a host user:

[0113] Before training the intelligent classification model of the present application, a training data set is prepared. The training data set includes a large number of training samples. The number of training samples is limited to a number suitable for training the model to a convergent state.

[0114] The training samples are composed of multiple attribute data items collected from each livestream user. The specific data items can be flexibly determined by those skilled in the art based on the desired predictive capabilities of the model. Each training sample is pre-assigned an attribute label to monitor the model's output.

[0115] Step S2200: vectorize the attribute data in the training sample to obtain a sample vector:

[0116] For each training sample that is to be input into the intelligent classification model to implement model training, its various related attribute data are vectorized by querying the vocabulary to obtain an embedding vector as the sample vector.

[0117] Step S2300: Input the sample vector into the intelligent classification model for semantic extraction and classification mapping to obtain attribute labels predicted by classification:

[0118] The sample vector is input into the intelligent classification model for processing. The intelligent classification model has a built-in text feature extraction model for extracting deep semantic feature information from the sample vector and realizing representation learning of the training sample. The text feature extraction model can be implemented using neural network models such as Bert, TextCNN, LSTM, etc. After the text feature extraction model extracts deep semantic feature information from the sample vector, a corresponding comprehensive vector is obtained. The comprehensive vector is further input into the fully connected layer in the model for full connection, thereby mapping it to a preset classification space, and the classifier is used to classify and predict the corresponding attribute label. The classification space contains multiple classifications corresponding to pre-given attribute labels. When the classification probability of each classification is predicted for a comprehensive vector, the attribute label corresponding to the classification with the largest classification probability is the attribute label predicted by the model for the training sample.

[0119] Step S2400: Calculate the loss value of the attribute label predicted by the intelligent classification model based on the attribute label corresponding to the training sample:

[0120] For the attribute labels predicted by the model, the attribute labels corresponding to the training samples can be called to calculate the model loss value. Since the model is implemented using a classifier, the cross entropy function can be used as the loss function to calculate the loss value.

[0121] Step S2500: perform gradient update on the intelligent classification model according to the loss value, or continue iterative training until the model reaches a convergence state:

[0122] After determining the loss value, the loss value can be compared with a preset threshold value used to identify whether the model has reached a convergence state. When the loss value reaches the preset threshold value, the model is considered to have been trained to a convergence state, and the model training can be terminated, and the model is used to predict the attribute labels corresponding to the subset of data items of the anchor user. When the loss value does not reach the preset threshold value, the model has not converged. Therefore, the model is gradient updated according to the loss value, and the weights of the model are corrected through back propagation to make the model closer to convergence. Then, the next training sample is called from the training data set, and the steps of this embodiment are repeated to continue iterative training of the model until the model is trained to a convergence state.

[0123] In this embodiment, the training process of the intelligent classification model of the present application is provided. According to the training process, the data items originally distributed in multiple data sources are used to train the intelligent classification model after being centrally called by the data definition model, so that the intelligent classification model obtains the ability to determine the attribute label corresponding to the anchor user according to the data item subset of the anchor user, thereby serving the prediction of the attribute label of each data item subset, forming a data cleaning mechanism for the data items specified by the data definition template. Therefore, the data item subset can be optimized through the data definition template, thereby improving the acquisition efficiency of the data set corresponding to the live streaming service. Moreover, due to the artificial intelligence factor of the intelligent classification model, the accuracy of matching the data item subset of the anchor user for the live streaming service can be improved.

[0124] In some embodiments, after the step S1240 of screening out the data item subset of the anchor user whose attribute label matches the attribute label in the operation item, the following steps are included:

[0125] Step S1245, obtaining the heat data of each attribute label corresponding to each anchor user, wherein the heat data is determined according to the user behavior data of the live streaming room of the anchor user:

[0126] For each attribute label in the attribute label system used in the present application, heat data representing the accessed heat of each attribute label can be formed in advance, and the popularity of the corresponding attribute label can be reflected through the heat data. The heat data of each attribute label can be determined by counting the number of times the anchor user carrying the attribute label is accessed, and the number of times the anchor user is accessed by the user can be determined by counting the user behavior data of the anchor user. The subsequent embodiments of the present application will further illustrate this, which is temporarily described here.

[0127] On the basis of the heat data of each attribute label, the heat data corresponding to the attribute label of each anchor user is associated, so as to be used for sorting.

[0128] Step S1246, sorting the data item subset of each anchor user according to the heat data, so that the subsequently generated data set maintains the corresponding order:

[0129] Then, according to the heat data, the data item subset of each anchor user is sorted from high to low, so that the higher the heat, the higher the sorting. Therefore, in the data set constructed according to these data item subsets, the heat sorting of each corresponding anchor user is actually completed.

[0130] This embodiment further improves the processing of the sorting relationship of each anchor user in the data set required for the live broadcast service, so that after the live broadcast service obtains the said data set and parses and displays it, the various anchor users can be displayed in an orderly manner in the graphical user interface of the terminal device, so that the arrangement order of the anchor users is associated with the attribute tags specified by the live broadcast service, thereby highlighting the functional focus of the live broadcast service.

[0131] See also Figure 6 In some extended embodiments, before step S1245, the step of obtaining the popularity data of each attribute tag corresponding to each anchor user, the following steps are included:

[0132] Step S1241: Obtain user behavior data associated with each live broadcast user, wherein the user behavior data corresponds to descriptive data generated by corresponding access events such as the live broadcast room of the live broadcast user being followed by a user, receiving a gift from a user, or being entered by a user:

[0133] When a live streaming platform user visits a live streaming room, corresponding user behavior data will be generated. For example, when a platform user enters the live streaming room, sends a gift to the live streaming user, or shares the live streaming room, a corresponding access event will be triggered, and corresponding user behavior records will be generated in the background and uploaded to the server. Therefore, data mining can be performed on this user behavior data to obtain more semantic information, namely the aforementioned popularity data, which serves to optimize the data set corresponding to the data definition template of this application.

[0134] Step S1242: Count the user behavior data to obtain the user popularity corresponding to each anchor user. The user popularity is a weighted statistical result of the number of multiple access events:

[0135] After obtaining the user behavior data, statistics can be performed on each anchor user one by one based on the number of access events corresponding to these user behavior data to calculate the user popularity corresponding to each anchor user. In the recommended implementation, a preset weighted formula is used to count the number of different access events. For example, for the statistical number of three different access events of the anchor user being followed by users, being given gifts by users, and being entered by users, different weights are associated with each of them. The weights and statistical numbers are multiplied and added together to obtain the weighted statistical results as the popularity of an anchor user. An example of a weighted statistical formula is as follows:

[0136] S i =α×Like+β×Present+γ×Enter

[0137] Among them, S iIt represents the user popularity of a single anchor user, Like, Present, and Enter represent different access events respectively, and α, β, and γ represent the weights corresponding to different access events respectively. The values ​​can be determined by technical personnel in this field as needed.

[0138] Step S1243: For each attribute tag in the preset attribute tag system, accumulate the user popularity of the anchor user carrying the attribute tag to obtain the accumulated popularity corresponding to the attribute tag:

[0139] Since each anchor user can be mapped to one or more attribute tags in the preset attribute tag system, the user popularity of each anchor user is regarded as the user popularity obtained by the anchor user corresponding to each attribute tag carried by the anchor user. Based on this, for each attribute tag in the attribute tag system, the user popularity corresponding to each anchor user can be counted to obtain the cumulative popularity corresponding to each attribute tag. That is, for each attribute tag in the attribute tag system, the corresponding cumulative popularity Sum j , which is the sum of the user popularity obtained from each anchor user.

[0140] Step S1244: normalize the accumulated heat of each attribute tag in the attribute tag system to obtain heat data corresponding to each attribute tag:

[0141] After the above steps, the cumulative heat Sum corresponding to each attribute label in the attribute label system has been determined j Based on this, the following formula can be used for normalization to provide a unified measure for the cumulative heat of each attribute label, thereby determining the heat data corresponding to each attribute label: j :

[0142]

[0143] That is, for an attribute tag, its popularity data is the product of its corresponding cumulative popularity divided by the sum of the cumulative popularity of all attribute tags in the attribute tag system. In this way, the popularity data of each attribute tag can be compared and ranked based on the same metric.

[0144] In order to optimize the data set determined according to the data definition template, this embodiment pre-determines popularity data for each attribute tag in the attribute tag system based on the user behavior data corresponding to the anchor user. The determination of the popularity data refers to the popularity of the anchor user carrying the attribute tag. Therefore, in the data set optimized for the live broadcast business, each selected anchor user is sorted according to the attribute tag. The sorting result can better represent the traffic effect of the anchor user, thereby improving the matching accuracy of the data required for the live broadcast business.

[0145] In an extended embodiment, after the step S1100 of obtaining the data definition template corresponding to the live service, the following step is included:

[0146] In step S1101, the specified data items in the data definition template corresponding to the live service are scheduled from their corresponding data sources to the secondary cache for calling according to the timing task trigger:

[0147] The server can preset a timing task, for example, a timer Schedule technology of Sping is used to trigger the update task at a timing, in response to the timing trigger task, each data item specified in the data definition template corresponding to the live service is scheduled from its corresponding multiple data sources to the secondary cache, so that more efficient operations are performed on these data items in the secondary cache, for example, the matching operation of step S1200 is performed. In a recommended embodiment, the secondary cache can be implemented by using Redis technology, and the data items of each anchor user are stored by using key-value pairs, for example, the UID of the anchor user is used as the key field, and each data item thereof is used as the value field for storage.

[0148] In step S1102, the data set obtained by performing operations on the data items of the data source in the secondary cache is stored in the primary cache:

[0149] In order to further improve the data scheduling efficiency and shorten the data access response time, for the data set obtained according to step S1200, it can be stored in the primary cache which is more efficient than the secondary cache, and subsequent calling and updating of the data set can be operated in the primary cache. In a recommended alternative embodiment, Guava Cache technology can be used to support the implementation of the primary cache to achieve the effect of efficient access.

[0150] In this embodiment, considering that the distributed stored data items occupy more memory space and the data logic is more bottom layer, they are placed in the secondary cache for operation, and considering that the data set constructed according to the data items is relatively high frequency calling, it is placed in the primary cache for operation, which comprehensively balances the implementation efficiency of the entire business logic in different stages, can improve the system running efficiency and robustness, ensures that each terminal device efficiently accesses each data set corresponding to each live service, and can ensure timely updating of the data set, and realizes mutual decoupling of the bottom layer operations corresponding to the data items in various different data sources.

[0151] In an extended embodiment, before the step S1100 of obtaining the data definition template corresponding to the live service, the following step is included:

[0152] Step S1000: Run the data source adaptation service and open an interface to an external data source to enable the external data source to access and participate in providing the data set for the live broadcast service:

[0153] In order to expand the compatibility of the live broadcast platform with data sources, a data source adaptation service can be run in the server. This service supports different external data source access protocols, such as PB, HTTP, YYP, etc., and opens corresponding mutual access interfaces to external data sources, so that the data source can support mutual access to different database engines, so that different data sources can pre-call these mutual access interfaces to access the system that implements the technical solution of this application, so as to facilitate the data definition template of this application to obtain its data items.

[0154] This embodiment further enhances the data acquisition capability of the data definition template, enabling the live broadcast service to further call external data through its data definition template, thereby enriching the information content of the live broadcast service and improving the accuracy of data matching through external data, and also greatly enhancing the compatibility of the system that implements the technical solution of this application.

[0155] See also Figure 7 A live broadcast service data processing device provided to meet one of the purposes of the present application includes a template calling module 1100, a data processing module 1200, and a data push module 1300, wherein: the template calling module 1100 is used to obtain a data definition template corresponding to the live broadcast service, the data definition template includes data items distributed in multiple data sources required to be called by the live broadcast service and operation items for performing predefined operations on these data items, the data items include attribute data items of the anchor user, and the operation items include one or more attribute tags; the data processing module 1200 is used to perform predefined operations corresponding to the operation items on the corresponding data items in multiple data sources according to the operation items in the data definition template, and determine a data set corresponding to the attribute tags in the operation items, the data set including a subset of data items corresponding to multiple anchor users; the data push module 1300 is used to respond to a terminal device's data call instruction for the live broadcast service and push the data set corresponding to the live broadcast service to the terminal device.

[0156] In an embodiment, the data processing module 1200 comprises: a template parsing unit configured to parse the data definition template to determine data items and operation items in each data source; a subset processing unit configured to apply a distributed lock call to obtain the data items in each data source, and determine a data item subset of each anchor user according to the data items associated with the same anchor user; a label determining unit configured to determine attribute labels of each anchor user according to each data item subset by using an intelligent classification model pre-trained to a convergence state; an anchor screening unit configured to screen a data item subset of an anchor user whose attribute label matches the attribute label in the operation item; and a format unification unit configured to format each data item subset of each anchor user to construct a data set in a standardized format.

[0157] In an embodiment, the intelligent classification model is pre-iteratively trained to a convergence state in a training module, and the training module comprises: a sample calling unit configured to call a single training sample from a training data set, the training sample comprising attribute data corresponding to a plurality of attribute data items of an anchor user; a vector encoding unit configured to vectorize the attribute data in the training sample to obtain a sample vector; a classification mapping unit configured to input the sample vector into the intelligent classification model to perform semantic extraction and classification mapping to obtain an attribute label predicted by classification; a loss calculation unit configured to calculate a loss value of the attribute label predicted by the intelligent classification model according to the attribute label corresponding to the training sample; and an iteration decision unit configured to perform gradient update on the intelligent classification model according to the loss value, or continue to iteratively train the model until the model reaches a convergence state.

[0158] In an embodiment, the data processing module 1200 comprises, after the anchor screening unit, the following units: a hotness calling unit configured to obtain hotness data of each attribute label corresponding to each anchor user, the hotness data being determined according to user behavior data of a live broadcast room of the anchor user; and a sorting processing unit configured to sort the data item subset of each anchor user according to the hotness data, so that the data set generated subsequently maintains a corresponding order.

[0159] In some extended embodiments, the data processing module 1200 includes the following units that run before the popularity calling unit: a data description unit, used to obtain user behavior data associated with each anchor user, and the user behavior data corresponds to the description data generated by the corresponding access events of the anchor user's live broadcast room being followed by users, receiving gifts from users, and being entered by users; a data statistics unit, used to count the user behavior data and obtain the user popularity corresponding to each anchor user, and the user popularity is a weighted statistical result of the number of multiple access events; a popularity accumulation unit, used to correspond to each attribute tag in a preset attribute tag system, accumulate the user popularity of the anchor user carrying the attribute tag, and obtain the accumulated popularity corresponding to the attribute tag; a popularity determination unit, used to normalize according to the accumulated popularity of each attribute tag in the attribute tag system, and obtain the popularity data corresponding to each attribute tag.

[0160] In some extended embodiments, the live broadcast business data processing device of the present application also includes the following modules that run after the template calling module 1100: a secondary cache module, which is used to dispatch the specified data items in the data definition template corresponding to the live broadcast business from its corresponding data source to the secondary cache for calling according to the timed task trigger; a primary cache module, which is used to store the data set obtained by performing operations based on the data items of the data source in the secondary cache in the primary cache.

[0161] In some extended embodiments, the live broadcast service data processing device of the present application also includes a data adaptation module that runs before the template calling module 1100, which is used to run the data source adaptation service and open an interface to the external data source to enable external data source access and participate in providing the data of the data set for the live broadcast service.

[0162] In order to solve the above technical problems, the embodiment of the present application also provides a computer device. Figure 8 As shown, a schematic diagram of the internal structure of a computer device. The computer device includes a processor, a computer-readable storage medium, a memory, and a network interface connected via a system bus. Among them, the computer-readable storage medium of the computer device stores an operating system, a database, and computer-readable instructions, and the database may store a control information sequence, and when the computer-readable instructions are executed by the processor, the processor may implement a live broadcast business data processing method. The processor of the computer device is used to provide computing and control capabilities to support the operation of the entire computer device. The memory of the computer device may store computer-readable instructions, and when the computer-readable instructions are executed by the processor, the processor may execute the live broadcast business data processing method of the present application. The network interface of the computer device is used to connect and communicate with the terminal. Those skilled in the art will understand that Figure 8The structure shown in the figure is only a block diagram of a part of the structure related to the solution of the present application, and does not constitute a limitation on the computer device to which the solution of the present application is applied. The specific computer device may include more or fewer components than shown in the figure, or combine certain components, or have a different component arrangement.

[0163] In this embodiment, the processor is used to execute Figure 7 The memory stores the program code and various data required to execute the modules or submodules. The network interface is used to transmit data between user terminals or servers. The memory in this embodiment stores the program code and data required to execute all modules / submodules in the live broadcast service data processing device of this application. The server can call the server's program code and data to execute the functions of all submodules.

[0164] The present application also provides a storage medium storing computer-readable instructions. When the computer-readable instructions are executed by one or more processors, the one or more processors execute the steps of the live broadcast business data processing method of any embodiment of the present application.

[0165] The present application also provides a computer program product, comprising a computer program / instruction, which implements the steps of the method described in any embodiment of the present application when executed by one or more processors.

[0166] Those skilled in the art will appreciate that all or part of the processes in the above-mentioned embodiments of the present application can be implemented by instructing the relevant hardware through a computer program. The computer program can be stored in a computer-readable storage medium. When the program is executed, it can include the processes of the above-mentioned embodiments of the method. The aforementioned storage medium can be a computer-readable storage medium such as a magnetic disk, an optical disk, a read-only memory (ROM), or a random access memory (RAM).

[0167] To sum up, this application implements standardized operations for multiple live broadcast services to call data items from multiple data sources through data definition templates, avoids data entropy increase, improves system efficiency, and saves system overhead.

[0168] Those skilled in the art will appreciate that the steps, measures, and schemes in the various operations, methods, and processes discussed in this application may be interchanged, modified, combined, or deleted. Furthermore, other steps, measures, and schemes in the various operations, methods, and processes discussed in this application may also be interchanged, modified, rearranged, decomposed, combined, or deleted. Furthermore, steps, measures, and schemes in the prior art that are similar to those disclosed in this application may also be interchanged, modified, rearranged, decomposed, combined, or deleted.

[0169] The above description is only part of the implementation methods of the present application. It should be pointed out that for ordinary technicians in this technical field, several improvements and modifications can be made without departing from the principles of the present application. These improvements and modifications should also be regarded as the scope of protection of the present application.

Claims

1. A method for processing live broadcast service data, characterized in that: The server used in the live broadcast platform includes the following steps: According to a predetermined trigger mechanism, a data definition template corresponding to the live broadcast service is obtained. The data definition template includes data items distributed across multiple data sources and required for the live broadcast service, and operation items for performing predefined operations on the data items. The data items include attribute data items of the anchor user, and the operation items include one or more attribute tags. According to the operation items in the data definition template, predefined operations corresponding to the operation items are performed on corresponding data items in multiple data sources to determine a data set corresponding to the attribute tags in the operation items, wherein the data set includes a subset of data items corresponding to multiple anchor users; In response to the terminal device's data call instruction for the live broadcast service, the corresponding data set of the live broadcast service is pushed to the terminal device.

2. The live broadcast service data processing method according to claim 1, characterized in that: According to the operation items in the data definition template, predefined operations corresponding to the operation items are performed on corresponding data items in multiple data sources to determine data sets corresponding to the attribute labels in the operation items, including the following steps: Parsing the data definition template to determine the data items and operation items in each data source; Applying a distributed lock call to obtain the data items in each data source, and determining a subset of data items of each anchor user based on the data items being associated with the same anchor user; Using an intelligent classification model that has been pre-trained to a convergent state, the attribute labels of each anchor user are determined based on each subset of data items; Filtering out a subset of anchor user data items whose attribute tags match the attribute tags in the operation items; Format the data item subsets of each anchor user to construct a data set in a standardized format.

3. The live broadcast service data processing method according to claim 2, characterized in that: The intelligent classification model is pre-trained iteratively to a convergent state, and the training process includes the following steps: Calling a single training sample from a training data set, where the training sample includes attribute data corresponding to multiple attribute data items of a host user; Vectorizing the attribute data in the training sample to obtain a sample vector; Inputting the sample vector into the intelligent classification model for semantic extraction and classification mapping to obtain attribute labels predicted by classification; Calculating the loss value of the attribute label predicted by the intelligent classification model based on the attribute label corresponding to the training sample; The intelligent classification model is gradient updated according to the loss value, or iterative training is continued until the model reaches a convergence state.

4. The live broadcast service data processing method according to claim 2, characterized in that: After the step of filtering out a subset of anchor user data items whose attribute tags match the attribute tags in the operation items, the following steps are included: Obtaining popularity data of each attribute tag corresponding to each anchor user, wherein the popularity data is determined based on user behavior data statistics of the anchor user's live broadcast room; The data item subsets of each anchor user are sorted according to the popularity data so that the subsequently generated data sets maintain the corresponding sorting.

5. The live broadcast service data processing method according to claim 4, characterized in that: Before the step of obtaining the popularity data of each attribute tag corresponding to each anchor user, the following steps are included: Obtaining user behavior data associated with each live broadcast user, wherein the user behavior data corresponds to descriptive data generated by corresponding access events of the live broadcast room of the live broadcast user being followed by a user, receiving a gift from a user, or being entered by a user; The user behavior data is counted to obtain the user popularity corresponding to each anchor user, where the user popularity is a weighted statistical result of the number of multiple access events; For each attribute tag in the preset attribute tag system, the user popularity of the anchor user carrying the attribute tag is accumulated to obtain the accumulated popularity corresponding to the attribute tag; Normalization is performed based on the accumulated heat of each attribute tag in the attribute tag system to obtain heat data corresponding to each attribute tag.

6. The live broadcast service data processing method according to claim 1, characterized in that: After obtaining the data definition template corresponding to the live broadcast service, the following steps are included: According to the timing task trigger, the specified data items in the data definition template corresponding to the live broadcast business are dispatched from their corresponding data sources to the secondary cache for calling; The data set obtained by performing operations on the data items of the data source in the second-level cache is stored in the first-level cache.

7. The live broadcast service data processing method according to claim 1, characterized in that: Before obtaining the data definition template corresponding to the live broadcast service, the following steps are included: Run the data source adaptation service and open the interface to the external data source to enable the external data source to access and participate in providing the data of the said data set for the live broadcast business.

8. A live broadcast service data processing device, characterized in that: Servers used in online live streaming platforms include: A template calling module is used to obtain a data definition template corresponding to the live broadcast service according to a predetermined trigger mechanism. The data definition template includes data items distributed across multiple data sources required for the live broadcast service and operation items for performing predefined operations on these data items. The data items include attribute data items of the anchor user, and the operation items include one or more attribute tags. A data processing module is configured to perform predefined operations corresponding to the operation items in the data definition template on the corresponding data items in the multiple data sources, and determine a data set corresponding to the attribute tags in the operation items, wherein the data set includes a subset of data items corresponding to multiple anchor users; The data push module is used to respond to the terminal device's data call instruction for the live broadcast service and push the corresponding data set of the live broadcast service to the terminal device.

9. A computer device comprising a central processing unit and a memory, characterized in that: The central processing unit is configured to call and run a computer program stored in the memory to execute the steps of the method according to any one of claims 1 to 7.

10. A computer-readable storage medium, characterized in that It stores a computer program implemented according to the method described in any one of claims 1 to 7 in the form of computer-readable instructions, and when the computer program is called and executed by a computer, the steps included in the corresponding method are executed.

Images