Data processing method and device, computer equipment and readable storage medium

Abstract

Embodiments of the invention provide a data processing method and apparatus, a computer device and a readable storage medium. The method comprises the steps of obtaining P first attribute tags of to-be-pushed updated game props, Q second attribute tags of a pushed historical game prop set, and historical interaction data of K candidate objects about historical game props; analyzing historical interaction data corresponding to the candidate object Ki to obtain tag preference degrees corresponding to the Q second attribute tags of the candidate object Ki; according to the label preference degree, the Q second attribute labels and the P first attribute labels, the prop preference degree of the candidate object Ki for the updated game prop is recognized; and according to the prop preference degrees corresponding to the K candidate objects, screening a push object from the K candidate objects, and pushing the updated game prop to the push object. By adopting the method and the device, the pushing accuracy of the updated game props can be improved, and the pushing cost of the updated game props can be reduced.

Description

Technical Field

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

[0002] With the rapid development of the game industry, various types of game applications are emerging in an endless stream, and the competition in the game industry is becoming increasingly fierce. Among them, many game applications provide a variety of game props for virtual objects controlled by game objects. Game props are virtual items used in games. In order to increase the fun of game applications, new game props will be added to game applications. In practice, it is found that different game objects often have different game prop requirements. Therefore, if new game props are pushed to invalid game objects, the push cost will increase. Invalid game objects refer to game objects that do not need new game props. If new game props are pushed to valid game objects, the game objects' interest in game applications will be increased, and then the stickiness between game objects and game applications will be increased. Valid game objects refer to game objects that need new game props. Based on this, how to accurately push new game props to game objects is a problem that needs to be solved urgently. Summary of the invention

[0003] The embodiments of the present application provide a data processing method, an apparatus, a computer device, and a readable storage medium, which can improve the push accuracy of updated game props (i.e., new game props) and reduce the push cost of updated game props.

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

[0005] Obtain P first attribute tags of the updated game props to be pushed, Q second attribute tags of the historical game props in the pushed historical game props set, and historical interaction data of K candidate objects on the historical game props; P and Q are both positive integers, and K is an integer greater than 1;

[0006] For candidate object K among K candidate objects i The corresponding historical interaction data is parsed to obtain the candidate object K i The label preferences corresponding to the Q second attribute labels respectively; i is a positive integer less than or equal to K;

[0007] According to the candidate K i For the label preferences corresponding to the Q second attribute labels, the Q second attribute labels, and the P first attribute labels, identify the candidate objects K i Item preference for updating game items;

[0008] According to the item preferences of the K candidate objects for the updated game items, push objects corresponding to the updated game items are selected from the K candidate objects, and the updated game items are pushed to the push objects.

[0009] On the one hand, an embodiment of the present application provides a data processing device, including:

[0010] An acquisition module, used to acquire P first attribute tags of the updated game props to be pushed, Q second attribute tags of the historical game props in the pushed historical game props set, and historical interaction data of K candidate objects on the historical game props; P and Q are both positive integers, and K is an integer greater than 1;

[0011] The parsing module is used to analyze the candidate object K among the K candidate objects. i The corresponding historical interaction data is parsed to obtain the candidate object K i The label preferences corresponding to the Q second attribute labels respectively; i is a positive integer less than or equal to K;

[0012] The recognition module is used to identify the candidate objects K i For the label preferences corresponding to the Q second attribute labels, the Q second attribute labels, and the P first attribute labels, identify the candidate objects K i Item preference for updating game items;

[0013] The push module is used to select push objects corresponding to the updated game props from the K candidate objects according to the prop preferences of the K candidate objects for the updated game props, and push the updated game props to the push objects.

[0014] Among them, the P first attribute labels and the Q second attribute labels are all attribute sub-labels;

[0015] The identification module includes: a first determining unit, a second determining unit, a searching unit and a third determining unit.

[0016] A first determining unit is used to determine the attribute parent labels to which the P first attribute labels respectively belong according to a universal label network, so as to obtain M first attribute parent labels; the universal label network reflects the subordinate relationship between the attribute sub-label and the attribute parent label; M is a positive integer;

[0017] The second determining unit is used to determine the attribute parent labels to which the Q second attribute labels respectively belong according to the universal label network, and obtain N second attribute parent labels; N is a positive integer;

[0018] A searching unit, used to search for second attribute parent tags that match the M first attribute parent tags respectively from the N second attribute parent tags, to obtain M matching attribute parent tags;

[0019] The third determining unit is used to determine the candidate object K i Based on the label preferences of the second attribute labels belonging to the M matching attribute parent labels and the P first attribute labels, determine the candidate objects K i Item preference for updating game items.

[0020] The third determining unit is specifically used for:

[0021] From the second attribute tags belonging to the M matching attribute parent tags, find the first attribute tag P among the P first attribute tags. j The matching second attribute tag is used as the first attribute tag P j The corresponding matching second attribute tag; j is a positive integer less than or equal to P;

[0022] The candidate object K i According to the label preference of the matching second attribute label, the candidate object K is determined i For the first attribute label P j 's label preference;

[0023] If we obtain the candidate object K i For the label preferences corresponding to the P first attribute labels, the candidate objects K i The label preferences corresponding to the P first attribute labels are summed up to obtain the candidate object K i Item preference for updating game items.

[0024] The parsing module includes: a fourth determining unit, a parsing unit and a fifth determining unit.

[0025] A parsing unit is used to select a candidate object K from K candidate objects. i In the corresponding historical interaction data, the candidate object K is parsed i The interaction categories and interaction times of historical game props under Q second attribute labels;

[0026] The fourth determining unit is used to determine the candidate object K according to the interaction category and the number of interactions. i Interaction preferences for historical game props under Q second attribute labels;

[0027] The fifth determining unit is configured to determine the candidate object K according to the i Corresponding interaction preference, determine the candidate object K i The label preferences corresponding to the Q second attribute labels.

[0028] The fifth determining unit is specifically used for:

[0029] Determine the candidate object Ki The number of historical game props that participate in the interaction and have a second attribute tag x; the second attribute tag x belongs to any one of Q second attribute tags, and x is a positive integer less than or equal to Q;

[0030] If the number of props is equal to 1, then the candidate object K i The interaction preference of the historical game props with the second attribute label x is determined as the candidate object K i Label preference for the second attribute label x;

[0031] If the number of props is greater than 1, then for candidate object K i The interaction preference of historical game props with the second attribute label x is summed to obtain the candidate object K i The label preference for the second attribute label x.

[0032] The fourth determining unit is specifically used for:

[0033] Obtaining the item weight coefficient corresponding to the target historical game item, and obtaining the interaction weight coefficient of the interaction category corresponding to the target historical game item; the target historical game item belongs to any one of the historical game items under the Q second attribute labels;

[0034] If the number of interaction categories corresponding to the target historical game props is equal to 1, the prop weight coefficient corresponding to the target historical game props, the interaction weight coefficient of the interaction category and the number of interactions are multiplied to obtain the candidate object K i Interaction preference for target historical game props;

[0035] If the number of categories is equal to R, then the item weight coefficient corresponding to the target historical game item, the interaction weight coefficient corresponding to the R interaction categories and the number of interactions are multiplied to obtain the interaction preference corresponding to the R interaction categories. The interaction preference corresponding to the R interaction categories is summed to obtain the candidate object K. i The interaction preference for the target historical game path; R is an integer greater than 1.

[0036] The push module includes: a screening unit and a sixth determination unit.

[0037] A screening unit, used to screen candidate objects whose item preference is greater than or equal to a preference threshold from the K candidate objects;

[0038] The sixth determining unit is used to determine the screened candidate objects as push objects corresponding to the updated game props.

[0039] The push module further includes: a generation unit, a first acquisition unit and a push unit.

[0040] A generating unit, used for generating push multimedia data about updating the game props according to the prop description information of the updated game props;

[0041] A first acquisition unit is used to acquire a push strategy corresponding to updating game props;

[0042] The push unit is used to push the multimedia data to the terminal device corresponding to the push object based on the push strategy.

[0043] The acquisition module includes: a first extraction unit, a word segmentation unit, a filtering unit and a second extraction unit.

[0044] A first extraction unit is used to obtain item description information about the updated game item to be pushed, and extract item attribute description text about the updated game item from the item description information about the updated game item;

[0045] The word segmentation unit is used to perform word segmentation on the item attribute description text to obtain initial attribute description words;

[0046] A filtering unit, used for filtering invalid words in the initial attribute description words to obtain attribute description words;

[0047] The second extraction unit is used to obtain the semantic feature vector of the attribute description word, and perform label extraction on the updated game props according to the semantic feature vector to obtain P first attribute labels of the updated game props.

[0048] Wherein, the second extraction unit is specifically used for:

[0049] Obtain F attribute parent labels in the general label network, and attribute sub-labels respectively included in the F attribute parent labels, the general label network is used to reflect the subordinate relationship between the attribute sub-labels and the attribute parent labels; F is a positive integer;

[0050] Call the label extraction model, and determine the attribute parent label to which the attribute description word belongs from the F attribute parent labels according to the semantic feature vector, as the target attribute parent label;

[0051] According to the attribute sub-tags included in the target attribute parent tag, the tag extraction is performed on the updated game props to obtain P first attribute tags of the updated game props.

[0052] Among them, according to the attribute sub-tags included in the target attribute parent tag, the tag extraction is performed on the updated game props to obtain P first attribute tags of the updated game props, including:

[0053] Get the label feature vector corresponding to the attribute sub-label included in the target attribute parent label;

[0054] Obtain feature similarity between the label feature vector and the semantic feature vector;

[0055] Extracting attribute sub-tags whose feature similarity is greater than or equal to a similarity threshold from the attribute sub-tags included in the target attribute parent tag;

[0056] The extracted attribute sub-tags are determined as P first attribute tags for updating the game props.

[0057] The historical game props set includes the historical game props that have been pushed in H game applications; H is a positive integer;

[0058] The acquisition module further includes: a second acquisition unit and a third extraction unit.

[0059] The second acquisition unit is used to acquire historical log data of K candidate objects respectively related to H game applications;

[0060] The third extraction unit is used to extract historical interaction data of the K candidate objects with respect to historical game props in the historical game prop set from the historical log data of the K candidate objects respectively related to H game applications.

[0061] An embodiment of the present application provides a computer device, including: a processor and a memory;

[0062] The processor is connected to the memory, wherein the memory is used to store a computer program. When the computer program is executed by the processor, the computer device executes the method provided in the embodiment of the present application.

[0063] 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.

[0064] In one aspect, an embodiment of the present application provides a computer program product, which includes 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 the processor executes the computer program, so that the computer device executes the method provided in the embodiment of the present application.

[0065] The label preference in the present application reflects the candidate's preference for the game props under the second attribute label, that is, the label preference reflects the prop style that the candidate is interested in, that is, the label preference can reflect the object label (that is, the user label) of the candidate, and the second attribute label reflects the prop style of the historical game props in one dimension or multiple dimensions, such as one or more dimensions including color, theme, environment, etc. Therefore, by determining the label preference of the candidate for the second attribute label of the historical game props based on the candidate's historical interaction data for the historical game props, that is, one second attribute label corresponds to one label preference, a one-to-one correspondence between the user label and the second attribute label (that is, the prop style) can be achieved, and the problem that the real prop style of the candidate cannot be accurately obtained due to the inability to correspond one-to-one between the user label and the prop style of the game prop can be avoided, that is, the label preference can truly reflect the prop style of the candidate. Furthermore, the candidate's item preference for updating game props is determined by the candidate's label preference for the second attribute label, the first attribute label for updating game props, and the second attribute label. The item preference reflects the candidate's preference for updating game props, that is, the item preference can reflect the candidate's need for updating game props. For example, if the candidate's item preference for updating game props is relatively high, it indicates that the candidate has a relatively high need for updating game props; conversely, if the candidate's item preference for updating game props is relatively low, it indicates that the candidate has a relatively low need for updating game props. Therefore, the computer device can screen the push objects corresponding to the updated game props from the K candidate objects according to the prop preferences corresponding to the K candidate objects, and push the updated game props to the push objects. This can improve the push accuracy of the updated game props, that is, accurately push the updated game props to the game objects that really need (or are interested in), improve the game objects' interest and stickiness in the game application, and at the same time, avoid pushing the updated game props to invalid game objects, thereby reducing the push cost of the game props. BRIEF DESCRIPTION OF THE DRAWINGS

[0066] In order to more clearly illustrate the technical solutions in the embodiments of the present application or the related technologies, the drawings required for use in the embodiments or the related technical descriptions are briefly introduced below. 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 paying creative work.

[0067] Figure 1 It is a structural diagram of a data processing system provided in an embodiment of the present application;

[0068] Figure 2This is a schematic diagram of an application scenario of data processing provided by an embodiment of the present application;

[0069] Figure 3 It is a flowchart of a data processing method provided in an embodiment of the present application;

[0070] Figure 4 is a schematic diagram of a universal tag network provided in an embodiment of the present application;

[0071] Figure 5 is a schematic diagram of determining a prop preference for updating a game prop provided by an embodiment of the present application;

[0072] Figure 6 It is a schematic diagram of determining a push object from K candidate objects provided by an embodiment of the present application;

[0073] Figure 7 It is a flowchart of a data processing method provided in an embodiment of the present application;

[0074] Figure 8 It is a schematic diagram of updating game props push provided by an embodiment of the present application;

[0075] Fig. 9 It is a schematic diagram of updating game props push provided by an embodiment of the present application;

[0076] Fig.10 is a structural schematic diagram of a data processing device provided in an embodiment of the present application;

[0077] Fig.11 It is a structural diagram of a computer device provided in an embodiment of the present application. DETAILED DESCRIPTION

[0078] The following will be combined with the drawings in the embodiments of the present application to clearly and completely describe the technical solutions in the embodiments of the present application. Obviously, the described embodiments are only part of the embodiments of the present application, not all of the embodiments. Based on the embodiments in the present application, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of this application.

[0079] See also Figure 1 , Figure 1 Schematic diagram of a data processing system provided in an embodiment of the present application. Figure 1 As shown, the data processing system may include a server 10 and a terminal device cluster. The terminal device cluster may include one or more terminal devices, and the number of terminal devices is not limited here. Figure 1As shown, it may specifically include terminal device 100a, terminal device 100b, terminal device 100c, ..., terminal device 100n. Figure 1 As shown, the terminal device 100a, the terminal device 100b, the terminal device 100c, ..., the terminal device 100n can be respectively connected to the above-mentioned server 10 through the network, so that each terminal device can exchange data with the server 10 through the network connection. Of course, the terminal device 100a, the terminal device 100b, the terminal device 100c, ..., the terminal device 100n can communicate with each other through a direct network connection, that is, point-to-point communication can be achieved between each terminal device; that is, when data needs to be exchanged between each two terminal devices, one terminal device (i.e., the sending terminal device) can send data directly to another terminal device (i.e., the receiving terminal device).

[0080] Each terminal device in the terminal device cluster may include: smart phones, tablet computers, laptop computers, desktop computers, intelligent voice interaction devices, smart home appliances (e.g., smart TVs), wearable devices, vehicle-mounted terminals, and other smart terminals with data processing functions. Figure 1 Each terminal device in the terminal device cluster shown in the figure can be installed with an application with data processing function. When the application is running in each terminal device, it can be respectively connected with the above Figure 1 The server 10 shown in the figure can exchange data, such as the application can specifically include a game application, a game service center application (such as a game store application), a game props push application, etc. Among them, the application in the embodiment of the present application can be integrated into a certain application, and the application can also be an independent application. The embodiment of the present application does not limit the type of application. For ease of understanding, the embodiment of the present application can be in Figure 1 Select one terminal device from the multiple terminal devices shown as the target terminal device. Figure 1 The terminal device 100a shown is used as a target terminal device. An application with a data processing function may be installed in the target terminal device. At this time, the target terminal device can implement data interaction with the server 10 through the application.

[0081] Among them, Figure 1 As shown, the server 10 can be a device that provides background services for applications in terminal devices. The server 10 is 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, CDN (Content Delivery Network), and big data and artificial intelligence platforms.

[0082] It is understandable that based on Figure 1 A data processing system in the present application can be applied to the game props push scenario. The game props in the embodiment of the present application can refer to virtual items used by the virtual character controlled by the game object (such as the game player) during the game, such as game tools (such as defense items, transportation items, combat items, etc.), virtual character food (fruits, vegetables, staple food, etc.), game skins (such as character tops, character pants, character shoes, and character accessories, etc.). Since game props can be optimized or updated (such as releasing new game props), therefore, historical game props can refer to game props that have been pushed in the historical time period, and updated game props can refer to game props that have been optimized or newly added in the current time period. Among them, the game props in the historical game props set can refer to game props that have been pushed in the historical time period in one game application or multiple game applications. In addition, the K candidate objects in this embodiment can refer to the game objects in the game application to which the updated game props belong, and K is an integer greater than 1.

[0083] Among them, the updated game props in the embodiment of the present application may have P first attribute tags, P is a positive integer, and P first attribute tags can be used to reflect any one or more of the prop attributes of the updated game props, such as the prop style, prop type, and prop use. It can be understood that the updated game props can have one or more first attribute tags at the same time, and the first attribute tags can reflect the abstract attribute semantic characteristics of the updated game props, and can more accurately reflect the multi-dimensional prop attributes of the updated game props. Similarly, the historical game props in the pushed historical game props set can have Q second attribute tags, Q is a positive integer, and Q second attribute tags can also be used to reflect any one or more of the prop style, prop type, and prop use of the historical game props. Similarly, the historical game props can have one or more second attribute tags at the same time, and the second attribute tags can reflect the abstract attribute semantic characteristics of the historical game props, and can more accurately reflect the multi-dimensional prop attributes of the historical game props.

[0084] In the specific implementation, in order to improve the activity and return rate of game objects (i.e., candidate objects) in game applications, game applications will conduct operation activities for specific events (such as the release of new versions of game applications, festival activities, etc.), and these operation activities usually launch specific exclusive game props. How to correctly identify which users will be interested in the game props launched by the operation activities (such as historical game props that have been pushed in the historical time period or newly released updated game props) is crucial to the success of the operation activities. Especially in the promotion stage of the operation activities, game applications usually display the promotional copy materials of the game props launched by the operation activities through off-site advertising applications, and deliver them to specific game objects to attract game objects to enter the game and improve the return rate and activity of the game objects. If the game objects that may be interested in the game props launched by the operation activities can be accurately identified, the game props launched by the operation activities can be avoided from being pushed to invalid game objects (i.e., game objects that do not need the game props launched by the operation activities), thereby greatly saving the cost of pushing game props.

[0085] Among them, if the game props launched by the operation activities are historical game props that have been released in the historical time period, we can judge whether a certain game object will like the historical game props based on the historical interaction data of the game objects in the past and the use of these historical game props. However, the game props launched by the operation activities are usually newly released (i.e., updated game props), so there is a cold start problem. The cold start problem means that there is no historical interaction data between the game object and the updated game props in the historical time period, that is, the game objects have never claimed or used the updated game props in the historical time period, so it is impossible to obtain the game object’s preference for the updated game props through the game object’s historical interaction data on the updated game props. Therefore, how to accurately identify the push object of the updated game props is a problem that needs to be solved urgently. Based on this, the embodiment of the present application can determine the push object of the updated game props from the K candidate objects according to the P first attribute tags of the updated game props to be pushed and the Q second attribute tags of the historical game props in the pushed historical game props set, that is, determine the push object that will be interested in the updated game props from the K candidate objects, and push the updated game props to the push object. In this way, the updated game props can be accurately pushed to the game object, reducing the cost of pushing the game props, and at the same time, it can also avoid interfering with some game objects that are not interested.

[0086] For further understanding, please refer to Figure 2 , Figure 2 Schematic diagram of an application scenario of data processing provided by an embodiment of the present application. Figure 2 The terminal device 201a, the terminal device 202a, the terminal device 203a, the terminal device 204a, etc. in the terminal device cluster 20a shown in the figure can be the above Figure 1 The terminal device of the terminal device cluster in the corresponding embodiment is as follows: Figure 2 The terminal device 20c shown may be the above Figure 1 The terminal device 100b in the corresponding embodiment is as follows: Figure 2 The server 20d shown may be the above Figure 1 The server 10 in the corresponding embodiment. The terminal devices in the terminal device cluster 20a and the terminal device 20c respectively establish a network connection with the server 20d, and can exchange data with the server 20d through the network connection. It can be understood that after the game application publishes the updated game props, the management object 20b corresponding to the terminal device 20c can trigger the push operation for the updated game props through the terminal device 20c. Of course, it can also mean that after the game application publishes the updated game props, the push operation for the updated game props is automatically triggered. The terminal device 20c can respond to the push operation for the updated game props, and send the updated game props, the prop description information of the updated game props, the historical game props in the pushed historical game props set, and the prop description information corresponding to the historical game props in the historical game props set to the server 20d, so that the server 20d determines the push object corresponding to the updated game props from the K candidate objects, and pushes the updated game props to the push object. Of course, the server 20d may also store the pushed historical game props set and the prop description information of the historical game props, and the terminal device 20c only needs to send the updated game props and the prop description information of the updated game props to the server 20d. Among them, the prop description information of the updated game props and the prop description information of the historical game props may refer to one or more of the information such as the prop name, prop image, prop description video, prop description audio, and prop introduction text.

[0087] It should be understood that the server 20d can obtain the historical interaction data of the K candidate objects respectively regarding the historical game props through the terminal devices in the terminal device cluster 20a. Specifically, the terminal devices in the terminal device cluster 20a may refer to the terminal devices respectively held by the K candidate objects, and one or more game applications may be installed in the terminal devices in the terminal device cluster 20a. The candidate objects among the K candidate objects can perform game activities through the game applications installed by the terminal devices in the terminal device cluster 20a. Therefore, the terminal devices in the terminal device cluster 20a store the historical interaction data of the candidate objects among the K candidate objects for the historical game props that have been pushed in one or more game applications. The terminal devices in the terminal device cluster 20a can send the historical interaction data of the candidate objects for the historical game props to the server 20d. For example, the terminal device 201a stores the candidate objects K among the K candidate objects. iFor the historical interaction data of historical game props in one or more game applications, the terminal device 201a can i The historical interaction data for one or more historical game items is sent to the server 20d.

[0088] Further, the server 20d can perform label extraction on the updated game props according to the prop description information of the updated game props, and obtain P first attribute labels of the updated game props. Specifically, a general label network is maintained in the server 20d, and the general label network includes F attribute parent labels and attribute sub-labels respectively included in the F attribute parent labels, and the general label network is used to reflect the subordinate relationship between the attribute sub-label and the attribute parent label. The server 20d can determine the P first attribute labels of the updated game props from the attribute sub-labels respectively included in the F attribute parent labels according to the prop description information of the updated game props. Similarly, the server 20d can perform label extraction on the historical game props in the historical game props set according to the prop description information corresponding to the historical game props in the historical game props set, and obtain Q second attribute labels of the historical game props set. Specifically, the server 20d can determine the Q second attribute labels of the historical game props set from the attribute sub-labels respectively included in the F attribute parent labels according to the prop description information of the historical game props.

[0089] Specifically, to obtain candidate object K among K candidate objects i Taking updating the prop preference of a game prop as an example, the server 20d may update the candidate object K i Analyze the historical interaction data of historical game props and obtain candidate objects K i Specifically, the historical interaction data includes purchase data, usage data, sales data, sharing data, etc. Different historical interaction data for historical game props reflect different interest of the candidate objects in the game props. Therefore, the server 20d can select the candidate objects K. i Analyze the historical interaction data of historical game props and determine the candidate object K i The interest in historical game props is used to obtain candidate objects K according to the interest in historical game props. i The server 20d further calculates the label preferences corresponding to the Q second attribute labels based on the candidate object K. i According to the label preferences corresponding to the Q second attribute labels, determine the candidate object K i The server 20d can select the candidate object K iThe label preferences corresponding to the P first attribute labels are summed up to obtain the candidate object K i Update the item preference corresponding to the game items.

[0090] Further, when the server 20d obtains the prop preferences of the K candidate objects for updating the game props, respectively, it can screen the push objects corresponding to the updated game props from the K candidate objects according to the prop preferences of the K candidate objects for updating the game props. For example, the server 20d can determine the candidate objects whose prop preferences are greater than or equal to the preference threshold among the K candidate objects as the push objects corresponding to the updated game objects. Among them, the preference threshold can be determined according to the push service requirements. It can be seen that the embodiment of the present application can determine the push object for updating the game props from the K candidate objects, that is, determine the push object that is interested in updating the game props among the K candidate objects, and push the updated game props to the push object. In this way, the updated game props can be accurately pushed to the game object, reducing the cost of pushing the game props, and at the same time, it can also avoid interfering with some game objects that are not interested.

[0091] For further information, see Figure 3 , Figure 3 is a flow chart of a data processing method provided in an embodiment of the present application. Figure 3 As shown, this method can be Figure 1 can be executed by any terminal device in Figure 1 The server 10 in the embodiment can also be executed by Figure 1 The terminal device and the server in the data processing method are executed together. The devices used to execute the method in this application can be collectively referred to as computer devices. The data processing method may include but is not limited to the following steps:

[0092] S101, obtaining P first attribute tags of updated game props to be pushed, Q second attribute tags of historical game props in a set of pushed historical game props, and historical interaction data of K candidate objects on historical game props.

[0093] Specifically, the updated game props to be pushed may refer to the newly released game props in the game application. Since the updated game props are newly released, it is convenient for the game object to better understand or obtain the game props of interest. The updated game props can be pushed to the game object to improve the exposure and usage rate of the updated game props. The historical game props in the pushed historical game props set may refer to the game props that have been released or pushed to the game object in one or more game applications within the historical time period. The computer device can extract the label of the updated game props to obtain P first attribute labels of the updated game props. Among them, the P first attribute labels can reflect any one or more of the prop style, prop type, prop use, etc. of the updated game props, and P is a positive integer. It can be understood that the updated game props can have one or more first attribute labels at the same time, and the first attribute label can reflect the abstract attribute semantic characteristics of the updated game props, and can more accurately reflect the multi-dimensional prop attributes of the updated game props. Similarly, the historical game props in the pushed historical game props set can have Q second attribute tags, where Q is a positive integer, and the Q second attribute tags can also be used to reflect any one or more of the prop style, prop type, prop purpose, etc. of the historical game props. Similarly, the historical game props can have one or more second attribute tags at the same time, and the second attribute tags can reflect the abstract attribute semantic characteristics of the historical game props, and can more accurately reflect the multi-dimensional prop attributes of the historical game props. Among them, the historical game props in the historical game props set can refer to game props that have been pushed in one game application, or can refer to game props that have been pushed in multiple game applications.

[0094] Specifically, the computer device may also obtain historical interaction data of K candidate objects on historical game props. Among them, the K candidate objects may refer to game objects in the game application corresponding to the updated game props, such as the K candidate objects may refer to all registered game objects in the game application corresponding to the updated game props, the K candidate objects may also refer to high-value game objects in the game application corresponding to the updated game props (such as game objects with a game duration greater than or equal to a duration threshold), and the K candidate objects may also refer to lost game objects in the game application corresponding to the updated game props (such as game objects that have not been active in the game application corresponding to the updated game props for a long period of time). Among them, K is an integer greater than 1. Among them, since the historical game props in the historical game props set may be game props in one or more game applications, the computer device may obtain historical interaction data of K candidate objects on historical game props from one or more game applications. Among them, the historical interaction data may refer to one or more of the data such as the use data, purchase data, sharing data, and selling data of the candidate objects for the historical game props.

[0095] It is understandable that the historical interaction data corresponding to each candidate object may refer to the historical interaction data of the corresponding candidate object for historical game props in one or more game applications. In this way, based on the historical interaction data of the candidate object for historical game props in one or more game applications, the preference of each candidate object for game props can be accurately extracted, and then it can be accurately determined whether each candidate is interested in updating game props, and then the updated game props can be pushed to the interested candidate object, which can improve the push accuracy of the updated game props.

[0096] Optionally, the specific method for the computer to obtain the P first attribute labels of the updated game props to be pushed may include: obtaining the prop description information about the updated game props to be pushed, and extracting the prop attribute description text about the updated game props from the prop description information of the updated game props. Performing word segmentation processing on the prop attribute description text to obtain initial attribute description words, filtering invalid words in the initial attribute description words to obtain attribute description words. Obtaining the semantic feature vector of the attribute description word, and performing label extraction on the updated game props according to the semantic feature vector to obtain the P first attribute labels of the updated game props.

[0097] Specifically, the computer device can obtain the prop description information about the updated game props to be pushed, and the prop description information of the updated game props can include one or more of the prop name, prop image, prop description video, prop description audio, and prop introduction text. Further, the computer device can extract the prop attribute description text about the updated game props from the prop description information of the updated game props. For example, if the prop description information of the updated game props is the prop name or the prop introduction text, the computer device can determine the prop name or the prop introduction text of the updated game props as the prop attribute description text about the updated game props. If the prop description information of the updated game props is the prop image and the prop description video, the computer device can use image recognition technology to perform image recognition on the prop image of the updated game props or the video frame in the prop description video to obtain the prop attribute description text about the updated game props. For example, if the computer device performs image recognition on the prop image of the updated game prop or the video frame in the prop description video and obtains that the color of the updated game prop is yellow, the text "yellow" can be determined as the prop attribute description text of the updated game prop. Of course, if the prop description information of the updated game prop is the prop description audio, the computer device can use audio recognition technology to perform audio recognition on the prop description audio to obtain the prop attribute description text about the updated game prop.

[0098] Furthermore, the computer device can perform word segmentation processing on the prop attribute description text of the updated game props to obtain the initial attribute description words about the updated game props, and filter the invalid words in the initial attribute description words to obtain the attribute description words. Among them, the invalid words in the initial attribute description words may refer to stop words such as "of", "is", "etc.", "in", "about", etc., which cannot reflect the prop attributes such as the prop style, prop type and prop purpose of the updated game props. In this way, the extraction accuracy of the first attribute label can be improved. The computer device can obtain the semantic feature vector of the attribute description words of the updated game props, extract the label of the updated game props according to the semantic feature vector, and obtain P first attribute labels of the updated game props.

[0099] Optionally, the computer device extracts labels from the attribute description words according to the semantic feature vector, and the specific method of obtaining P first attribute labels of the updated game props may include: obtaining F attribute parent labels in the general label network, and the attribute sub-labels respectively included in the F attribute parent labels, and the general label network is used to reflect the subordinate relationship between the attribute sub-labels and the attribute parent labels; F is a positive integer. Call the label extraction model, and determine the attribute parent label to which the attribute description words belong from the F attribute parent labels according to the semantic feature vector as the target attribute parent label. According to the attribute sub-labels belonging to the target attribute parent label, label extraction is performed on the updated game props to obtain P first attribute labels of the updated game props.

[0100] Specifically, the computer device can obtain F attribute parent tags in the general tag network, and the attribute sub-tags respectively included in the F attribute parent tags, where F is a positive integer, and the general tag network is used to reflect the subordinate relationship between the attribute sub-tag and the attribute parent tag. It can be understood that the F attribute parent tags can refer to tag dimensions, such as one or more of "color", "theme", "gender", "environment", "activity", "object" and "festival". The attribute sub-tags respectively included in the F attribute parent tags can refer to tags under the tag dimension. For example, the attribute parent tag "color" can include attribute sub-tags such as "gold", "red", "green", and "colorful"; the attribute parent tag "theme" can include attribute sub-tags such as "science fiction", "fantasy", and "funny". It can be understood that the general tag network can be used to reflect the attribute sub-tags included in each attribute parent tag, and the attribute parent tag to which each attribute sub-tag belongs.

[0101] It should be noted that, for the universality of attribute tags in different game applications, the second attribute tags of different historical game props and the first attribute tags of updated game props all belong to the universal tag network, which reflects the subordinate relationship between the attribute sub-tag and the attribute parent tag, that is, the universal tag network includes multiple attribute parent tags, and the attribute sub-tag of each attribute parent tag. The attribute parent tag is used to reflect the prop style of the rough dimension (such as prop type) of the game props, and the attribute sub-tag is used to reflect the more fine-grained prop style of the game props. For example, the attribute parent tag includes color, theme, character, etc., and the attribute sub-tag of the attribute parent tag color can include specific colors such as red, yellow, blue, and green. The first attribute tag and the second attribute tag can both belong to the attribute sub-tag in the universal tag network. The universal tag network is used to unify the naming of the attribute tags of game props, that is, the second attribute tag of the historical game props and the first attribute tag of the updated game props all belong to the universal tag network. For example, if the colors of historical game props a1 in game application A and game props b1 in game application B are both red, it can be determined that the colors of historical game props a1 in game application A and game props b1 in game application B are both the attribute sub-label "red" in the universal label network. In this way, it can be avoided that the attribute labels extracted from game props in different game applications are greatly different, resulting in the attribute labels of candidate objects in other game applications cannot be applied to the game application that updates the game props, and the push objects interested in the updated game props cannot be determined.

[0102] like Figure 4 As shown, Figure 4 is a schematic diagram of a universal tag network provided in an embodiment of the present application, such as Figure 4 As shown, the general label network includes attribute parent labels such as "color", "theme", "gender", "environment", "activity", "festival", "object", etc., and the attribute parent label can refer to the label dimension. Figure 4As shown, each attribute parent tag includes one or more attribute sub-tags, such as the attribute parent tag "color" includes the attribute sub-tag "gold", the attribute sub-tag "red", the attribute sub-tag "color", the attribute sub-tag "black", ... and other attribute sub-tags. The attribute parent tag "theme" includes the attribute sub-tag "science fiction", the attribute sub-tag "fantasy", the attribute sub-tag "funny", the attribute sub-tag "cute", ... and other attribute sub-tags. The attribute parent tag "gender" includes the attribute sub-tag "male", the attribute sub-tag "female", the attribute sub-tag "neutral", ... and other attribute sub-tags. The attribute parent tag "environment" includes the attribute sub-tag "ocean", the attribute sub-tag "desert", the attribute sub-tag "forest", the attribute sub-tag "snow", ... and other attribute sub-tags. The attribute parent tag "activity" includes the attribute sub-tag "dance", the attribute sub-tag "racing", the attribute sub-tag "boxing", the attribute sub-tag "football", ... and other attribute sub-tags. The attribute parent tag "festival" includes the attribute subtag "New Year", the attribute subtag "New Year's Day", the attribute subtag "Dragon Boat Festival", the attribute subtag "National Day", ... and other attribute subtags. The attribute parent tag "object" includes the attribute subtag "bird", the attribute subtag "fruit", the attribute subtag "vegetable", the attribute subtag "machinery", ... and other attribute subtags. Through the universal tag network, labels can be quickly extracted for updating game props or historical game props, thereby improving the efficiency of extracting game prop labels. At the same time, through the universal tag network, it is also convenient to subsequently determine the prop preferences of candidate objects for updating game props, thereby improving the efficiency of determining prop preferences.

[0103] Further, the computer device can call the label extraction model, and determine the attribute parent label to which the attribute description word belongs from F attribute parent labels according to the semantic feature vector of the attribute description word, as the target attribute parent label. According to the attribute sub-label included in the target attribute parent label, the updated game props are label extracted to obtain P first attribute labels of the updated game props. For example, if the attribute description words of the updated game props include "fantasy", the computer device can determine that the attribute parent label of the attribute description word "fantasy" is "theme" according to the semantic feature vector of the attribute description word "fantasy". Further, the computer device can extract the attribute sub-label matching the attribute description word "fantasy" from the attribute sub-label included in the attribute parent label "theme" as the first attribute label of the updated game props. In this way, by first determining the target attribute parent label to which the attribute description word belongs, determining the first attribute label of the updated game props from the attribute sub-label included in the target attribute parent label, without matching the attribute description word with all the attribute sub-labels, the amount of calculation can be reduced, the P first attribute labels of the updated game props can be quickly extracted, and the efficiency of attribute label extraction can be improved.

[0104] Optionally, the computer device extracts labels for the updated game props according to the attribute sub-labels included in the target attribute parent label, and the specific method of obtaining the P first attribute labels for the updated game props may include: obtaining the label feature vector corresponding to the attribute sub-label included in the target attribute parent label. Obtaining the feature similarity between the label feature vector and the semantic feature vector, and extracting the attribute sub-labels whose feature similarity is greater than or equal to the similarity threshold from the attribute sub-labels included in the target attribute parent label. Determine the extracted attribute sub-labels as the P first attribute labels for the updated game props.

[0105] Specifically, the computer device can obtain the label feature vector corresponding to the attribute sub-label included in the target attribute parent label through the label extraction model, and obtain the feature similarity between the label feature vector and the semantic feature vector. From the attribute sub-label included in the target attribute parent label, extract the attribute sub-label whose feature similarity is greater than or equal to the similarity threshold, and determine the extracted attribute sub-label as the P first attribute labels of the updated game props. For example, if the name of the updated game prop is "Golden Nether Queen Scepter", the attribute description words of the updated game props can include "gold", "Nether", "Queen" and "scepter". Specifically, taking the attribute description word "gold" as an example, the computer device can determine that the attribute parent label to which the attribute description word "gold" belongs is "color", and the general label network reflects that the attribute parent label "color" includes attribute sub-labels such as "gold", "red", "green", and "color". The computer device can obtain the label feature vector of each attribute sub-label in the attribute parent label "color", obtain the semantic feature vector of the attribute description word "gold", and the feature similarity between the label feature vector of each attribute sub-label in the attribute parent label "color". If the feature similarity between the attribute description word "gold" and the attribute sub-tag "gold" is 90%, the feature similarity between the attribute description word "gold" and the attribute sub-tag "color" is 30%, the feature similarity between the attribute description word "gold" and the attribute sub-tag "red" is 10%, etc., the similarity threshold can be 80%, then the attribute sub-tag "gold" can be determined as the first attribute tag for updating game props. In the same way, the attribute sub-tags corresponding to "gold", "queen" and "scepter" can be obtained respectively, and then P first attribute tags for updating game props can be obtained. Among them, the similarity threshold can be set according to the specific situation, and the embodiment of the present application is not limited here. It can be seen that the embodiment of the present application can extract P first attribute tags for updating game props from the attribute sub-tags included in the target attribute parent tag, which can narrow the tag extraction range and thus improve the tag extraction efficiency. Similarly, the computer device can obtain Q second attribute tags of historical game props in the historical game props set.

[0106] Optionally, the computer device can train the binary classifier corresponding to each attribute sub-tag in the general label network, and judge whether the corresponding attribute sub-tag can be used as the first attribute tag for updating the game props based on the binary classifier corresponding to each attribute sub-tag. For example, taking the attribute sub-tag y01 in the general label network as an example, the computer device can train the binary classifier corresponding to the attribute sub-tag y01. After the training is completed, the prop description information of the updated game props can be input into the binary classifier corresponding to the attribute sub-tag y01. Through the binary classifier corresponding to the attribute sub-tag y01, the prop description information of the updated game props is extracted to obtain a semantic feature vector (such as an embedding vector (i.e., embedding) pre-trained by text or image). Further, through the binary classification algorithm (such as logistic regression, decision tree, SVM, neural network, etc.) in the binary classifier corresponding to the attribute sub-tag y01, it is judged whether the attribute sub-tag y01 can be used as the first attribute tag for updating the game props. In this way, the computer device can obtain Q second attribute tags of the historical game props in the historical game props set. Among them, an updated game prop may have one or more attribute parent tags, and may also have one or more attribute child tags.

[0107] Optionally, the historical game props set includes historical game props that have been pushed in H game applications, H is a positive integer, and the specific manner in which the computer device obtains the historical interaction data of the K candidate objects respectively on the historical game props may include: obtaining the historical log data of the K candidate objects respectively on the H game applications. From the historical log data of the K candidate objects respectively on the H game applications, extract the historical interaction data of the K candidate objects respectively on the historical game props in the historical game props set.

[0108] Specifically, the computer device can obtain the historical log data of K candidate objects on H game applications, that is, the historical log data of K candidate objects in H game applications, and extract the historical interaction data of K candidate objects for historical game props in the historical game props set from the historical log data of K candidate objects on H game applications. It can be seen that the computer device can extract the historical interaction data of K candidate objects for historical game props from one or more game applications, so that the problem that the candidate object does not have historical interaction data for historical game props in a game application and the prop preference of the candidate object cannot be obtained can be avoided, and the accuracy of the prop preference extraction of the candidate object can be improved. In this way, the cold start problem of updating game props can be solved. Even if there is no historical record of the candidate object for the historical game props in the game application corresponding to the updated game props, the push object interested in updating the game props can be identified.

[0109] S102: for a candidate object K among the K candidate objects. i The corresponding historical interaction data is parsed to obtain the candidate object K i The label preferences corresponding to the Q second attribute labels.

[0110] For ease of understanding, let K be the candidate among K candidates. i For example, the computer device can perform the following operations on the candidate object K: i The corresponding historical interaction data is parsed to obtain the candidate object K i According to the preference of historical game props, we can then i According to the preference of historical game props, determine the candidate object K i The label preferences corresponding to the Q second attribute labels respectively. Wherein, i is a positive integer less than or equal to K. Similarly, the computer device can obtain the label preferences corresponding to the Q second attribute labels respectively for each candidate object in the K candidate objects.

[0111] Optionally, the computer device obtains the candidate object K i The specific method for determining the label preferences corresponding to the Q second attribute labels may include: selecting candidate object K from among the K candidate objects i In the corresponding historical interaction data, the candidate object K is parsed i According to the interaction categories and interaction times of the historical game props under the Q second attribute labels, determine the candidate objects K i The interaction preference of historical game props under Q second attribute labels. i Corresponding interaction preference, determine the candidate object K i The label preferences corresponding to the Q second attribute labels.

[0112] Specifically, the candidate object K i The corresponding historical interaction data may include candidate objects K i For the historical interaction data of each historical game prop in the historical game prop set, the historical interaction data may include one or more of purchase data, usage data, sharing data, and sales data. The computer device may select the candidate object K i In the corresponding historical interaction data, the candidate object K is parsed i The interaction categories and interaction times of the historical game props under the Q second attribute tags. The interaction categories may include one or more of the purchase category, use category, sharing category, and sale category. It is understandable that the computer device can select the candidate objects K i In the corresponding historical interaction data, the candidate object K is parsedi For each interaction category and interaction count of the historical game props under the second attribute tag. For example, the Q second attribute tags include the second attribute tag x, and the historical game props under the second attribute tag x include the historical game prop S001. The computer device selects the candidate object K i In the corresponding historical interaction data, the candidate object K is parsed i The interaction category and interaction number of the historical game prop S001. i Based on the interaction categories and interaction times of historical game props under Q second attribute labels, determine the candidate objects K i The interaction preference of historical game props under Q second attribute labels.

[0113] It is understandable that the computer device can i For each historical game item’s interaction category and number of interactions, calculate the candidate object K i For example, the computer device can determine the interaction preference of each historical game item according to the candidate object K i Based on the interaction category and interaction times of historical game prop S001, calculate the candidate object K i The interaction preference for the historical game prop S001. Further, the computer device can select the candidate object K i Based on the interaction preferences of historical game props under Q second attribute labels, determine the candidate objects K i For example, the computer device can calculate the label preference corresponding to the Q second attribute labels according to the candidate object K i According to the interaction preference of the historical game prop S001 under the second attribute label x, the candidate object K is determined i The label preference for the second attribute label x.

[0114] Optionally, the computer device may select a candidate object K i The corresponding interaction preference determines the candidate object K i The specific method for determining the label preferences corresponding to the Q second attribute labels may include: determining the candidate object K i The number of historical game props that participate in the interaction and have the second attribute label x; the second attribute label x belongs to any one of the Q second attribute labels, and x is a positive integer less than or equal to Q. If the number of props is equal to 1, then the candidate object K i The interaction preference of the historical game props with the second attribute label x is determined as the candidate object K i The label preference for the second attribute label x. If the number of props is greater than 1, then the candidate object K iThe interaction preference of historical game props with the second attribute label x is summed to obtain the candidate object K i The label preference for the second attribute label x.

[0115] Specifically, the computer device Q second attribute tags include the second attribute tag x. Taking the second attribute tag x as an example, the computer device can determine the candidate object K from the historical game props set. i The historical game props participating in the interaction are used as the historical game props participating in the interaction. Candidate object K i The historical game props involved in the interaction refer to the items that interact with the candidate object K i There are interactive historical game props, such as candidate object K i Purchased historical game props, candidate objects K i Shared historical game props or candidates K i There are historical game props that have been used, etc. Further, the computer device can obtain the number of props of historical game props with the second attribute label x from the participating historical game props. If the number of props is equal to 1, the candidate object K i The interaction preference of the historical game props with the second attribute label x is determined as the candidate object K i The label preference for the second attribute label x. For example, the candidate object K i For the historical game props with the second attribute label x, the interaction preference is 0.4, then the candidate object K i The label preference for the second attribute label x is 0.4.

[0116] Specifically, if the number of props is greater than 1, that is, there are multiple historical game props in the participating historical game props with the second attribute label x, then for the candidate object K i The interaction preference of historical game props with the second attribute label x is summed to obtain the candidate object K i The label preference for the second attribute label x. For example, among the historical game props, three historical game props have the second attribute label x, and the three historical game props include historical game prop S002, historical game prop S003, and historical game prop S004. Candidate object K i The interaction preference for historical game prop S002 is 0.3, and the candidate object K i The interaction preference for historical game prop S003 is 0.4, and the candidate object K i The interaction preference for historical game prop S004 is 0.2, so the candidate object K i The label preference for the second attribute label x is 0.9.

[0117] Optionally, the computer device determines the candidate object K according to the interaction category and the number of interactions. i The specific method for the interaction preference of the historical game props under the Q second attribute labels may include: obtaining the prop weight coefficient corresponding to the target historical game prop, and obtaining the interaction weight coefficient of the interaction category corresponding to the target historical game prop; the target historical game prop belongs to any one of the historical game props under the Q second attribute labels. If the number of the interaction category corresponding to the target historical game prop is equal to 1, then the prop weight coefficient corresponding to the target historical game prop, the interaction weight coefficient of the interaction category and the number of interactions are multiplied to obtain the candidate object K i For the interaction preference of the target historical game props. If the number of categories is equal to R, the prop weight coefficient corresponding to the target historical game props, the interaction weight coefficient corresponding to the R interaction categories and the number of interactions are multiplied to obtain the interaction preference corresponding to the R interaction categories, and the interaction preference corresponding to the R interaction categories is summed to obtain the candidate object K. i The interaction preference for the target historical game path; R is an integer greater than 1.

[0118] Specifically, taking the target historical game prop as an example, the target historical game prop belongs to any one of the historical game props under the Q second attribute tags, and the computer device can obtain the prop feature information of the target historical game prop, which may include one or more of the information such as prop quality, prop rarity, and prop price. The computer device can determine the prop weight coefficient corresponding to the target historical game prop according to the prop feature information corresponding to the target historical game prop. For example, the higher the prop quality of the target historical game prop, the higher the prop weight coefficient of the target historical game prop; of course, the lower the prop quality of the target historical game prop, the lower the prop weight coefficient of the target historical game prop. For example, the higher the prop rarity of the target historical game prop (such as the number of target historical game props released is small), the higher the prop weight coefficient of the target historical game prop; of course, the lower the prop rarity of the target historical game prop, the lower the prop weight coefficient of the target historical game prop. At the same time, the computer device can obtain the interaction weight coefficient of the interaction category corresponding to the target historical game prop, and the interaction weight coefficient of the interaction category can be determined according to the preference degree of the candidate object reflected by the interaction category. For example, the preference degree of the candidate object reflected by the sharing category is higher than the preference degree of the candidate object reflected by the use category, so the interaction weight coefficient of the sharing category is higher than the interaction weight coefficient of the use category.

[0119] Furthermore, the computer device may obtain the number of interaction categories corresponding to the target historical game props. If the number of interaction categories corresponding to the target historical game props is equal to 1, such as the candidate object Ki Only the target historical game props have been purchased, etc. The computer device can multiply the prop weight coefficient corresponding to the target historical game props, the interaction weight coefficient of the interaction category, and the number of interactions (such as the number of purchases) to obtain the candidate object K i The interaction preference for the target historical game props. Specifically, if the number of categories is equal to R, R is an integer greater than 1, such as candidate object K i If the target historical game props are purchased, used, and shared, the computer device can respectively multiply the prop weight coefficient corresponding to the target historical game props, the interaction weight coefficient corresponding to the R interaction categories, and the number of interactions to obtain the interaction preference degrees corresponding to the R interaction categories, and sum the interaction preference degrees corresponding to the R interaction categories to obtain the candidate object K. i Interaction preference for the target historical game props. For example, the R types of interaction categories include purchase categories, usage categories, and sharing categories. The computer device can multiply the item weight coefficients corresponding to the target historical game props, and the interaction weight coefficients and the number of interactions corresponding to the purchase categories to obtain the interaction preference corresponding to the purchase category. Similarly, the computer device can multiply the item weight coefficients corresponding to the target historical game props, and the interaction weight coefficients and the number of interactions corresponding to the usage categories to obtain the interaction preference corresponding to the usage category. Similarly, the computer device can multiply the item weight coefficients corresponding to the target historical game props, and the interaction weight coefficients and the number of interactions corresponding to the sharing categories to obtain the interaction preference corresponding to the sharing category. Further, the computer device can sum the interaction preference corresponding to the purchase category, the interaction preference corresponding to the usage category, and the interaction preference corresponding to the sharing category to obtain the candidate object K. i Interaction preference for the target historical game path.

[0120] Specifically, the historical game props set includes candidate object K i Taking the S historical game props that participate in the interaction and have the second attribute label x as an example, S is a positive integer less than or equal to the total number of participating historical game props. It can be understood that the S historical game props in the participating historical game props all have the second attribute label x, that is, the S historical game props can have the same second attribute label x. For example, historical game prop S01 has second attribute labels such as "gold", "fantasy", and "animal", and historical game prop S02 has second attribute labels such as "gold", "cute", and "plant". It can be seen that historical game props S01 and historical game props S02 both have the second attribute label "gold".

[0121] Optionally, the computer device processes a candidate object K among the K candidate objects. iThe corresponding historical interaction data is parsed to obtain the candidate object K i A specific method for determining the label preferences corresponding to the Q second attribute labels may include: obtaining S historical game props from S historical game props a The corresponding first prop feature information, according to the historical game props S a Corresponding prop feature information, determine the historical game props S a The corresponding prop weight coefficient; a is a positive integer less than or equal to S. For the candidate object K among the K candidate objects i The corresponding historical interaction data is parsed to obtain the candidate object K i For historical game props S a R types of interaction categories, and the number of interactions corresponding to each of the R types of interaction categories; R is a positive integer. a The corresponding prop weight coefficient, R types of interaction categories, and the number of interactions corresponding to the R types of interaction categories are used to determine the candidate object K i For historical game props S a Until the interaction preferences corresponding to the S historical game props are obtained, the candidate object K is determined according to the interaction preferences corresponding to the S historical game props. i The label preference for the second attribute label x.

[0122] Specifically, the computer device may obtain the historical game props S among the S historical game props. a The prop feature information may include one or more of the prop quality, prop rarity, and prop price. The computer device may use the historical game props S a Corresponding prop feature information, determine the historical game props S a The corresponding item weight coefficient. For example, historical game item S a The higher the quality of the props, the higher the historical game props S a The higher the item weight coefficient, the higher the historical game item S a The lower the quality of the props, the lower the historical game props S a The lower the item weight coefficient is. For example, historical game item S a The higher the rarity of the item (such as historical game items S a The number of releases is small), then the historical game props S a The higher the item weight coefficient, the higher the historical game item S a The lower the rarity of the item, the more historical game items S a The lower the item weight coefficient.

[0123] Specifically, the computer device may select a candidate object K among the K candidate objects. i The corresponding historical interaction data is parsed to obtain the candidate object K i For historical game props S a R types of interaction categories, the R types of interaction categories may include one or more of a purchase category, a use category, a share category, and a sell category. At the same time, the computer device may obtain the number of interactions corresponding to the R types of interaction categories, such as the number of purchases corresponding to the purchase category, the number of uses corresponding to the use category, the number of shares corresponding to the share category, etc. For example, the candidate object K i Purchased historical game item S a 2 times, candidate K i Wearing historical game props S a 6 times, candidate K i Wearing historical game props S a Go to battle 3 times, etc. Further, the computer device can be based on the historical game props S a The corresponding prop weight coefficient, R types of interaction categories, and the number of interactions corresponding to the R types of interaction categories are used to determine the candidate object K i For historical game props S a In this way, the computer device can obtain the candidate object K i According to the interaction preference degrees corresponding to the S historical game props, the candidate objects K are determined. i The label preference for the second attribute label x.

[0124] Optionally, the computer device can be used according to the historical game props S a The corresponding prop weight coefficient, R types of interaction categories, and the number of interactions corresponding to the R types of interaction categories are used to determine the candidate object K i For historical game props S a The specific method of determining the interaction preference may include: obtaining the interaction category R among the R interaction categories. t The corresponding interaction weight coefficient; t is a positive integer less than or equal to R. For the interaction category R t Corresponding number of interactions and interaction weight coefficient, historical game props S a The corresponding prop weight coefficients are multiplied to obtain the candidate object K i In the interaction category R t The interaction preference under i For the interaction preferences corresponding to the R interaction categories, the interaction preferences corresponding to the R interaction categories are summed up to obtain the candidate object K i For historical game props S ainteraction preference.

[0125] Specifically, the computer device may obtain an interaction category R among the R interaction categories. t The corresponding interaction feature information may include the interaction category R t The corresponding interactive interest level may be the interest level reflected when the game object implements an interactive operation on the game props. For example, if the game object shares historical game props (i.e., there is an operation of sharing category), it means that the game object has a high interest level in the historical game props, and it can be determined that the interactive interest level corresponding to the sharing category is high. The computer device can t Corresponding interaction feature information, determine the interaction category R t The corresponding interaction weight coefficient. For example, the interaction category R t The higher the corresponding interaction interest, the higher the interaction category R t The higher the corresponding interaction weight coefficient, the higher the interaction category R t The lower the corresponding interaction interest, the higher the interaction category R t In this way, the computer device can obtain the interaction weight coefficients corresponding to the R interaction categories, such as the interaction weight coefficient corresponding to the purchase category is 5, the interaction weight coefficient corresponding to the dress-up category is 1, and the interaction weight coefficient corresponding to the sharing category is 6.

[0126] Furthermore, the computer device may t Corresponding number of interactions and interaction weight coefficient, historical game props S a The corresponding prop weight coefficients are multiplied to obtain the candidate object K i In the interaction category R t Specifically, the computer device obtains the candidate object K i In the interaction category R t When the interaction preference is under , the candidate object K can be constructed i With interaction category R t The mapping relationship between them is obtained, pref[K i ,R t ]. Further, the computer device may i ,R t ] to initialize, pref[K i ,R t ] is set to 0, that is, pref[K i ,R t ]=0. The computer device can use the following formula (1) to determine the candidate object K i In the interaction category R t The interaction preference under .

[0127] pref[K i ,R t ]=pref[K i ,R t ]+count(K i ,S a ,R t )*item_weight(S a )*action_weight(R t ) (1)

[0128] Among them, count(K) in formula (1) i ,S a ,R t ) refers to the candidate object K i For historical game props S a The interaction category R implemented t The corresponding number of interactions, item_weight(S a ) refers to the historical game props S a The corresponding prop weight coefficient, action_weight(R t ) is the interaction category R t The corresponding interaction weight coefficient.

[0129] In this way, the computer device can obtain the candidate object K i For the interaction preferences corresponding to the R interaction categories, if the candidate objects K are obtained i For the interaction preferences corresponding to the R interaction categories, the interaction preferences corresponding to the R interaction categories can be summed to obtain the candidate object K i For historical game props S a interaction preference.

[0130] Optionally, the computer device determines the candidate object K according to the interaction preference corresponding to each of the S historical game props. i The specific method for determining the label preference of the second attribute label x may include: if S is equal to 1, then for the candidate object K i For historical game props S a The interaction preference is normalized to obtain the candidate object K i The label preference of the second attribute label x. If S is greater than 1, the interaction preferences corresponding to the S historical game props are summed to obtain the superposition preference, and the superposition preference is normalized to obtain the candidate object K. i The label preference for the second attribute label x.

[0131] Specifically, if S is equal to 1, that is, the historical game props set contains only historical game props S a With the second attribute label x, then for the candidate object K i For historical game props S a The interaction preference is normalized to obtain the candidate object K i The label preference for the second attribute label x. If S is greater than 1, it means that there are more than one historical game props S in the historical game props set. a has the second attribute label x, and there are other historical game props that also have the second attribute label x, then the interaction preferences corresponding to the S historical game props are summed up to obtain the superposition preference, and the superposition preference is normalized to obtain the candidate object K i The label preference for the second attribute label x. In this way, the label preference obtained after normalization can better reflect the candidate object K i The preference degree for the second attribute label x.

[0132] Specifically, the computer device obtains the candidate object K i For the label preference of the second attribute label x, the candidate object K can be constructed i The mapping relationship between the second attribute label x and pref[K i ,x]. Further, the computer device can perform pref[K i ,x] to initialize, and pref[K i ,x] is set to 0, that is, pref[K i ,x]=0. Specifically, when the computer device obtains the interaction preference corresponding to S historical game props (such as the historical game props S when S is equal to 1), a The interaction preference of S historical game props can be determined as the candidate object K i For the initial label preference of the second attribute label x, the initial label preference is normalized to obtain the candidate object K i Regarding the label preference of the second attribute label x, the computer device may use the following formula (2) to normalize the initial label preference.

[0133]

[0134] Among them, pref[K i ,x]′ refers to the candidate object K i For the initial label preference of the second attribute label x, ∑ x′∈Q pref( i ,′) 2It refers to the sum of the squares of the initial label preferences corresponding to the Q second attribute labels. In this way, the initial label preferences corresponding to the Q second attribute labels are normalized so that the sum of the label preferences corresponding to the Q second attribute labels is 1. In this way, the candidate object K can be intuitively reflected. i The preference level for each second attribute label.

[0135] S103, based on the candidate object K i For the label preferences corresponding to the Q second attribute labels, the Q second attribute labels, and the P first attribute labels, identify the candidate objects K i Item preference for updating game items.

[0136] Specifically, the computer device may i According to the label preferences corresponding to the Q second attribute labels and the Q second attribute labels, determine the candidate objects K i According to the label preferences corresponding to the P first attribute labels, we can then i According to the label preferences corresponding to the P first attribute labels, identify the candidate objects K i Item preference for updating game items.

[0137] Optionally, the P first attribute tags of the updated game props and the Q second attribute tags of the historical game props in the historical game prop set are all attribute sub-tags. i A specific method for updating the prop preference of game props may include: determining the attribute parent tags to which the P first attribute tags belong respectively according to the universal tag network, and obtaining M first attribute parent tags; the universal tag network reflects the subordinate relationship between the attribute child tag and the attribute parent tag; M is a positive integer. According to the universal tag network, determining the attribute parent tags to which the Q second attribute tags belong respectively, and obtaining N second attribute parent tags; N is a positive integer. From the N second attribute parent tags, searching for the second attribute parent tags that match the M first attribute parent tags respectively, and obtaining M matching attribute parent tags. According to the candidate object K i Based on the label preferences of the second attribute labels belonging to the M matching attribute parent labels and the P first attribute labels, determine the candidate objects K i Item preference for updating game items.

[0138] It is understandable that if Figure 4 As shown, Figure 4The general label network in shows the subordinate relationship between the F attribute parent labels and the attribute sub-labels respectively. The P first attribute labels of the updated game props can be attribute sub-labels obtained by extracting labels from the updated game props according to the attribute sub-labels respectively included in the F attribute parent labels in the general label network. Similarly, the Q second attribute labels of the historical game props in the historical game props set can also be attribute sub-labels obtained by extracting labels from the historical game props in the historical game props set according to the attribute sub-labels respectively included in the F attribute parent labels in the general label network.

[0139] Specifically, the computer device can determine the attribute parent tags to which the P first attribute tags of the updated game props belong respectively according to the universal tag network, and obtain M first attribute parent tags, where M is a positive integer. It can be understood that since the P first attribute tags are determined from the attribute child tags respectively included in the F attribute parent tags in the universal tag network, the first attribute tags in the P first attribute tags all have attribute parent tags to which they belong. Similarly, the computer device can determine the attribute parent tags to which the Q second attribute tags belong respectively according to the universal tag network, and obtain N second attribute parent tags, where N is a positive integer. Of course, the computer device can mark the attribute parent tag to which each first attribute tag belongs when determining the P first attribute tags of the updated game props, so that the computer device can directly obtain the attribute parent tags to which the P first attribute tags belong respectively. Similarly, the computer device can also mark the attribute parent tag to which each second attribute tag belongs when determining the O second attribute tags of the historical game props set, so that the computer device can directly obtain the attribute parent tags to which the Q second attribute tags belong respectively.

[0140] Furthermore, the computer device may search for second attribute parent tags that match the M first attribute parent tags from the N second attribute parent tags, and obtain M matching attribute parent tags. The computer device may obtain attribute sub-tags belonging to the M matching attribute parent tags, and select the attribute sub-tags from the candidate objects K. i Among the label preferences of Q second attribute labels, determine the candidate object K i The label preference of the second attribute label belonging to the M matching attribute parent labels. i According to the label preferences of the second attribute labels belonging to the M matching attribute parent labels, the label preferences corresponding to the P first attribute labels are determined, and the candidate objects K are determined according to the label preferences corresponding to the P first attribute labels. i For updating the props preference of game props. It can be seen that by determining M matching attribute parent tags, K candidate objects are obtained. iFor the label preferences of the second attribute labels belonging to M matching attribute parent labels, the label preferences corresponding to the P first attribute labels are determined, without having to determine the labels that match the P first attribute labels from the Q second attribute labels. This can narrow the label matching range, thereby reducing the amount of calculation and improving the efficiency of determining the prop preference.

[0141] Optionally, the computer device may select a candidate object K i Based on the label preferences of the second attribute labels belonging to the M matching attribute parent labels and the P first attribute labels, determine the candidate objects K i A specific method for updating the prop preference of a game prop may include: searching for a second attribute tag belonging to M matching attribute parent tags that matches a first attribute tag P among P first attribute tags. j The matching second attribute tag is used as the first attribute tag P j The corresponding matching second attribute label; j is a positive integer less than or equal to P. i According to the label preference of the matching second attribute label, the candidate object K is determined i For the first attribute label P j The label preference of i For the label preferences corresponding to the P first attribute labels, the candidate objects K i The label preferences corresponding to the P first attribute labels are summed up to obtain the candidate object K i Item preference for updating game items.

[0142] Specifically, the computer device can search for a second attribute tag belonging to M matching attribute parent tags that matches a first attribute tag P among P first attribute tags. j The matching second attribute tag is used as the first attribute tag P j The corresponding matching second attribute tag. It is understandable that the computer device can search for a second attribute tag that matches each of the P first attribute tags from the second attribute tags belonging to the M matching attribute parent tags. Specifically, the computer device determines the second attribute parent tag M that matches the second attribute parent tag M in the M first attribute parent tags from the M matching attribute parent tags. c The target matching attribute parent tag that matches is determined from the second attribute tags belonging to the target matching attribute parent tag. c Thus, from the second attribute tags belonging to the target matching attribute parent tag, determine the second attribute tags that match the second attribute parent tag M. cBy matching each first attribute tag with a second attribute tag, an attribute subtag under an attribute parent tag is realized, and the attribute subtag under the matching attribute parent tag is matched, so that the tag matching range can be further narrowed, and the efficiency of determining the tag preference can be further improved. At the same time, the cold start problem of updating game props can also be solved. Even if the candidate object has no history of updating game props, the prop preference of the candidate object that is interested in updating game props can be identified.

[0143] Further, the computer device can select from the candidate objects K i According to the label preferences corresponding to the Q second attribute labels, determine the candidate object K i According to the label preference of the matching second attribute label, the candidate object K i According to the label preference of the matching second attribute label, the candidate object K is determined i For the first attribute label P j In the same way, if we obtain the candidate object K i For the label preferences corresponding to the P first attribute labels, the computer device can select the candidate objects K i The label preferences corresponding to the P first attribute labels are summed up to obtain the candidate object K i Item preference for updating game items.

[0144] like Figure 5 As shown, Figure 5 is a schematic diagram of determining a prop preference for updating a game prop provided by an embodiment of the present application, such as Figure 5 As shown, taking the name of the updated game prop as "Golden Nether Queen's Scepter" as an example, the computer device can extract the label of the updated game prop based on the name of the updated game prop "Golden Nether Queen's Scepter" and the universal label network, and obtain the P first attribute labels of the updated game prop as "gold", "fantasy" and "female". At the same time, the computer device can determine the attribute parent label corresponding to each first attribute label, that is, the attribute parent label of the first attribute label "gold" is "color", the attribute parent label of the first attribute label "fantasy" is "theme", and the attribute parent label of the first attribute label "gender" is "female", thereby obtaining the label affiliation structure 50a of the updated game prop.

[0145] Similarly, the computer device can extract labels from the historical game props in the historical game props set to obtain Q second attribute labels and the second attribute parent label to which each second attribute label belongs, that is, the second attribute labels "gold", "red", "color", "black", etc. under the second attribute parent label "color", the second attribute labels "science fiction", "fantasy", "funny", "cute", etc. under the second attribute parent label "theme", the second attribute labels "male" and "female" under the second attribute parent label "gender", and the second attribute labels "ocean", "desert", "forest", "snow", etc. under the second attribute parent label "environment". Among them, the computer device can obtain the historical interaction data of the candidate object for the historical game props, parse the historical interaction data, and obtain the label preference of the candidate object for each second attribute label. For details, please refer to the content of the above step S102, and the embodiment of this application will not be repeated here.

[0146] like Figure 5 As shown, a label preference table 50b consisting of the second attribute parent label of historical game props, the second attribute label and the label preference corresponding to each second attribute label is shown. That is, the label preference of the second attribute label "gold" under the second attribute parent label "color" is 0.1, the label preference of the second attribute label "red" is 0.01, the label preference of the second attribute label "color" is 0.1, the label preference of the second attribute label "black" is 0.05, etc. The label preference of the second attribute label "science fiction" under the second attribute parent label "theme" is 0.16, the label preference of the second attribute label "fantasy" is 0.13, the label preference of the second attribute label "funny" is 0.03, the label preference of the second attribute label "cute" is 0.01, etc. The label preference of the second attribute label "male" under the second attribute parent label "gender" is 0.01, and the label preference of the second attribute label "female" is 0.17. The label preference of the second attribute label "ocean" under the second attribute parent label "environment" is 0.09, the label preference of the second attribute label "desert" is 0.06, the label preference of the second attribute label "forest" is 0.08, the label preference of the second attribute label "snow" is 0.03, and so on.

[0147] Further, the computer device may determine, from the second attribute parent tags of the historical game props, a second attribute parent tag set 50c that matches each first attribute parent tag of the updated game props, such as Figure 5As shown, the first attribute parent tag "color" matches the second attribute parent tag "color", the first attribute parent tag "theme" matches the second attribute parent tag "theme", and the first attribute parent tag "gender" matches the second attribute parent tag "gender". The computer device can determine the second attribute tag that matches the first attribute tag from the second attribute tags included in the matching second attribute parent tags as the matching second attribute tag. The tag preference of the matching second attribute tag is determined as the tag preference of the corresponding first attribute tag. Figure 5 As shown, the computer device can determine the label preference of 0.1 of the second attribute label "gold" that matches the first attribute label "gold" as the label preference of the first attribute label "gold". The computer device can determine the label preference of 0.13 of the second attribute label "fantasy" that matches the first attribute label "fantasy" as the label preference of the first attribute label "fantasy". The computer device can determine the label preference of 0.17 of the second attribute label "female" that matches the first attribute label "female" as the label preference of the first attribute label "female". Furthermore, the computer device can obtain a label preference table 50d composed of the first attribute parent label of the updated game props, the first attribute label, and the label preferences corresponding to each first attribute label. Further, the computer device can sum the label preferences of all first attribute labels for the updated game props to obtain the item preference 50e of the candidate object for the updated game props, that is, 0.13+0.17+0.1=0.4. In this way, an attribute subtag under an attribute parent tag can be matched with an attribute subtag under a matching attribute parent tag, which can further narrow the tag matching range and further improve the efficiency of determining tag preference.

[0148] S104, according to the item preferences of the K candidate objects for the updated game items, select a push object corresponding to the updated game item from the K candidate objects, and push the updated game item to the push object.

[0149] Specifically, the computer device may obtain K candidate objects and their prop preferences for updating the game props. For details, please refer to the above-mentioned method for obtaining the candidate objects K. i Method for updating props preferences for game props. When the prop preferences of K candidate objects for updating game props are obtained, the computer device can select the push object corresponding to the updated game prop from the K candidate objects according to the prop preferences corresponding to the K candidate objects, and push the updated game prop to the push object.

[0150] Optionally, the computer device can sort the prop preferences corresponding to the K candidate objects from large to small, obtain the sorted prop preferences, select the candidate objects corresponding to the prop preferences arranged in the top E from the sorted prop preferences, and determine them as the push objects corresponding to the updated game props. Optionally, the computer device also screens the candidate objects whose prop preferences are greater than or equal to the preference threshold from the K candidate objects, and determines the candidate objects obtained by screening as the push objects corresponding to the updated game props. Among them, the preference threshold can be set according to the specific push service requirements. It can be seen that the embodiment of the present application can determine the push object for updating the game props from the K candidate objects, that is, determine the push object that is interested in updating the game props from the K candidate objects, and push the updated game props to the push object. In this way, the updated game props can be accurately pushed for the game object, reducing the cost of pushing the game props, and also avoiding interference with some uninterested game objects.

[0151] Of course, the number of updated game props can also be one or at least two. When the number of updated game props is G, and G is at least two, the P first attribute tags are the attribute tags corresponding to the G updated game props. The computer device can construct a query (i.e., a query engine) for the G updated game props, and each query consists of each first attribute tag in the P first attribute tags and the number of props corresponding to the first attribute tag. Among them, the number of props corresponding to the first attribute tag refers to the number of props with the corresponding first attribute tag in the G updated game props. For example, the query (G, P01) corresponding to the first attribute tag P01 is equal to 3, which means that there are 3 updated game props with the first attribute tag P01 among the G updated game props.

[0152] The composition of the query for each first attribute tag obtained by the computer device can be seen in formula (3).

[0153] query(G, p) = ∑ g∈G tag(g,p) (3)

[0154] Wherein, query(G, p) in formula (3) refers to the query of the first attribute label p among P first attribute labels, ∑ g∈G tag(g, p) refers to summing the number of updated game props with the first attribute tag p among the G updated game props. For example, if the updated game prop g among the G updated game props has the first attribute tag p, then tag(g, p) is equal to 1, otherwise it is equal to 0. Since there may be multiple updated game props with the same first attribute tag among the G updated game props, counting the number of props corresponding to each first attribute tag can improve the accuracy of obtaining the prop preference of each candidate object for the G updated game props.

[0155] For example, the G updated game props include the prop "Golden Nether Queen Scepter" and the prop "Golden Cute Corn Set", the prop "Golden Nether Queen Scepter" has the first attribute tag "gold", the first attribute tag "fantasy" and the first attribute tag "female", and the prop "Golden Cute Corn Set" has the first attribute tag "gold", the first attribute tag "cute" and the first attribute tag "plant". query(G) = {"gold": 2, "fantasy": 1, "female": 1, "cute": 1, "plant": 1}.

[0156] Specifically, the computer device determines the candidate object K among the K candidate objects. i When updating the item preferences of G game items, the candidate object K can be determined according to the following formula (4): i The preference similarity with G updated game props is determined as the candidate object K i Item preferences for G updated game items.

[0157] similarity(K i , G) = ∑ p∈P pref(K i ,p)*query(G,p) (4)

[0158] Among them, pref(K i , p) refers to the label preference of candidate object Ki for the first attribute label p among P first attribute labels, query(G, p) refers to the number of props with the first attribute label p among G updated game props, ∑ p∈P pref(K i , p)*query(G, p) refers to summing up the label preferences corresponding to the P first attribute labels of the candidate object Ki.

[0159] like Figure 6 As shown, Figure 6 is a schematic diagram of determining a push object from K candidate objects provided by an embodiment of the present application, such as Figure 6As shown, taking the determination of G updated game props push objects from K candidate objects as an example, G is a positive integer greater than 1, and the computer device can execute step 60a to extract Q second attribute tags of the historical game props in the historical game props set. The specific tag extraction method can refer to the tag extraction content of the above step S101, and the embodiment of the present application will not be repeated here. At the same time, the computer device can execute step 60b to extract the historical interaction data of the K candidate objects for the historical game props respectively. The historical game props in the historical game props set can refer to the pushed historical game props in one or more game applications. Therefore, the computer device can extract the historical interaction data of the candidate objects for the historical game props from one or more game applications. Similarly, the computer device can execute step 60c to extract P first attribute tags of the G updated game props. The extraction of the P first attribute tags can refer to the tag extraction content of the above step S101, and the embodiment of the present application will not be repeated here.

[0160] Further, the computer device can execute step 60d to determine the label preference of each candidate object in the K candidate objects for each second attribute label in the Q second attribute labels based on the historical interaction data of the K candidate objects for the historical game props. For details, please refer to the content of the above step S102, which will not be repeated in the embodiments of the present application. At the same time, the computer device can execute step 60e to construct a label query engine (i.e., query) for G updated game props. Since there will be multiple updated game props with the same first attribute label among the G updated game props, the number of props corresponding to each first attribute label is counted, which can improve the accuracy of obtaining the prop preference of each candidate object for the G updated game props. Further, the computer device can execute step 60f to calculate the prop interest of each candidate object for the G updated game props. For details, please refer to the content of the above step S103, which will not be repeated in the embodiments of the present application. After obtaining the prop interest degrees corresponding to the K candidate objects, the computer device can execute step 60g to determine G push objects for updated game props from the K candidate objects according to the prop interest degrees corresponding to the K candidate objects. It can be seen that the present application can determine the push objects that are interested in the G updated game props from the K candidate objects, and then accurately push the G updated game props, reduce the game prop push cost, and also avoid interference with some uninterested game objects.

[0161] In the embodiment of the present application, the label preference can be used to reflect the candidate object's preference for the game props under the second attribute label, that is, the label preference reflects the prop style that the candidate object is interested in, that is, the label preference can reflect the object label (that is, the user label) of the candidate object, and the second attribute label reflects the prop style of the historical game props in one dimension or multiple dimensions, such as one or more dimensions including color, theme, environment, etc. Therefore, by determining the label preference of the candidate object for the second attribute label of the historical game props based on the candidate object's historical interaction data for the historical game props, that is, one second attribute label corresponds to one label preference, a one-to-one correspondence between the user label and the second attribute label (that is, the prop style) can be achieved, and the problem that the real prop style of the candidate object cannot be accurately obtained due to the inability to correspond one-to-one between the user label and the prop style of the game prop can be avoided, that is, the label preference can truly reflect the prop style of the candidate object. Furthermore, the candidate's item preference for updating game props is determined by the candidate's label preference for the second attribute label, the first attribute label for updating game props, and the second attribute label. The item preference reflects the candidate's preference for updating game props, that is, the item preference can reflect the candidate's need for updating game props. For example, if the candidate's item preference for updating game props is relatively high, it indicates that the candidate has a relatively high need for updating game props; conversely, if the candidate's item preference for updating game props is relatively low, it indicates that the candidate has a relatively low need for updating game props. Therefore, the computer device can screen the push objects corresponding to the updated game props from the K candidate objects according to the prop preferences corresponding to the K candidate objects, and push the updated game props to the push objects. This can improve the push accuracy of the updated game props, that is, accurately push the updated game props to the game objects that really need (or are interested in), improve the game objects' interest and stickiness in the game application, and at the same time, avoid pushing the updated game props to invalid game objects, thereby reducing the push cost of the game props.

[0162] For further information, see Figure 7 , Figure 7 is a flow chart of a data processing method provided in an embodiment of the present application. Figure 7 As shown, this method can be Figure 1 can be executed by any terminal device in Figure 1 The server 10 in the embodiment can also be executed by Figure 1 The terminal device and the server in the data processing method are executed together. The devices used to execute the method in this application can be collectively referred to as computer devices. The data processing method may include but is not limited to the following steps:

[0163] S201, obtaining P first attribute tags of updated game props to be pushed, Q second attribute tags of historical game props in a set of pushed historical game props, and historical interaction data of K candidate objects on the historical game props.

[0164] S202: for a candidate object K among the K candidate objects. i The corresponding historical interaction data is parsed to obtain the candidate object K i The label preferences corresponding to the Q second attribute labels.

[0165] S203, based on the candidate object K i For the label preferences corresponding to the Q second attribute labels, the Q second attribute labels, and the P first attribute labels, identify the candidate objects K i Item preference for updating game items.

[0166] S204, according to the item preferences of the K candidate objects for the updated game items, select the push objects corresponding to the updated game items from the K candidate objects.

[0167] Specifically, the contents of steps S201 to S204 in the embodiment of the present application can be referred to above. Figure 3 The contents of step S101 to step S104 in the present embodiment of the application will not be repeated here.

[0168] S205: Generate push multimedia data about updating the game props according to the prop description information of the updated game props.

[0169] Specifically, the computer device can generate push multimedia data about updating game props according to the prop description information of the updated game props. Among them, the prop description information of the updated game props can include one or more of the information such as the prop name, prop image, prop description video, prop description audio, and prop introduction text, and the push multimedia data can refer to the prop push image, prop push video, etc. Of course, the computer device can push personalized push multimedia data for each push object based on the object feature information of each push object, so as to improve the click rate of the push object for the push multimedia data, and then improve the push accuracy of the updated game props. Among them, the object feature information can refer to the multimedia type that the push object likes to watch, such as images or videos.

[0170] S206, obtaining a push strategy corresponding to the updated game props, and based on the push strategy, pushing the multimedia data to the terminal device corresponding to the push object.

[0171] Specifically, the computer device may obtain a push strategy corresponding to the updated game props, and the push strategy may include a prop acquisition method for the updated game props, such as a purchase method, a task completion acquisition method, etc. When the prop acquisition method for the updated game props is a purchase method, the push strategy may include bid information for the updated game props, etc. When the prop acquisition method for the updated game props is a task completion acquisition method, the push strategy may include task information required to complete the updated game props, etc. Among them, the push strategy may also include a prop push algorithm, and the computer device may push the push multimedia data corresponding to the updated game props to the terminal device corresponding to the push object in a suitable scenario according to the prop push algorithm. Specifically, the computer device may determine the push scenario that the push object is interested in according to the prop push algorithm, and push the push multimedia data to the terminal device corresponding to the push object in the push scenario. Among them, the push scenario may refer to an information interaction application, or an information browsing web page, or a multimedia data reading application (text application, audio application, video application, etc.), etc.

[0172] Among them, the terminal device corresponding to the push object can display the push multimedia data corresponding to the updated game props on the device display screen, and then the push object can view the push multimedia data corresponding to the updated game props. If the push object is for the push multimedia data corresponding to the updated game props, the push multimedia data corresponding to the updated game props can be clicked. The terminal device corresponding to the push object can respond to the push exclusive click operation on the push multimedia data corresponding to the updated game props, and display the acquisition page of the updated game props. The acquisition page can be a purchase page for the updated game props, or it can refer to a task information page that needs to be completed to obtain the updated game props, etc.

[0173] It can be seen that the embodiment of the present application can determine the push object for updating game props from K candidate objects, that is, determine the push object that is interested in updating game props from the K candidate objects, and push the updated game props to the push object. In this way, the updated game props can be accurately pushed to the game object, reducing the cost of pushing game props, and at the same time, it can also avoid interfering with some uninterested game objects. At the same time, the embodiment of the present application can generate push multimedia data about updating game props, so as to accurately push updated game props to the push object and improve the accuracy of pushing updated game props.

[0174] like Figure 8 As shown, Figure 8 is a schematic diagram of updating game props push provided by an embodiment of the present application, such as Figure 8As shown, on the game advertiser side, after the computer device obtains the newly released updated game props 80a in the game application, it can obtain K candidate objects 80b in the game application. The K candidate objects 80b may refer to all registered game objects in the game application, the K candidate objects may also refer to high-value game objects in the game application (such as game objects whose game duration is greater than or equal to the duration threshold), and the K candidate objects may also refer to lost game objects in the game application (such as game objects that have not been active in the game application for a long period of time). Further, the computer device can filter out the push object 80c of the updated game props 80a from the K candidate objects 80b. For details, please refer to the above Figure 3 The contents of steps S101-S104 in the embodiment of the present application are not repeated here. Further, the computer device can obtain the prop description information of the updated game prop 80a, and the prop description information of the updated game prop 80a can include one or more of the prop name, prop image, prop description video, prop description audio, and prop introduction text. According to the prop description information of the updated game prop 80a and the push object 80c of the updated game prop 80a, the push multimedia data 80d of the updated game prop 80a is generated.

[0175] Specifically, the computer device can push personalized push multimedia data 80d to the push object 80c according to the prop description information of the updated game prop 80a and the object feature information of the push object 80c, so as to improve the click rate of the push object 80c for the push multimedia data 80d, thereby improving the push accuracy of the updated game prop 80a. Further, the computer device can send the push multimedia data 80d to the push platform 80f according to the push strategy 80e of the updated game prop 80a. On the push platform side, the computer device can send the push multimedia data 80d to the terminal device 80g corresponding to the push object 80c through the push platform 80f based on the push strategy 80e. Among them, the push strategy 80e can include the prop acquisition method of the updated game prop 80a, such as the purchase method, the task completion acquisition method, etc. When the prop acquisition method of the updated game prop 80a is the purchase method, the push strategy 80e can include the bid information of the updated game prop 80a, etc. When the prop acquisition method for updating the game props 80a is the task acquisition method, the push strategy 80e may include task information required to complete the updated game props 80a, etc. Among them, the push strategy 80e may also include a prop push algorithm. On the object side, the terminal device 80g can push the push multimedia data 80d corresponding to the updated game props 80a to the push object 80c in a suitable scenario according to the prop push algorithm. It can be seen that the embodiment of the present application determines the push object for updating the game props from K candidate objects, that is, determines the push object that is interested in the updated game props from the K candidate objects, and pushes the updated game props to the push object. In this way, the updated game props can be accurately pushed to the game object, reducing the cost of pushing the game props, and also avoiding interference to some uninterested game objects. At the same time, the embodiment of the present application can generate push multimedia data about updating the game props, so as to accurately push the updated game props to the push object and improve the push accuracy of the updated game props.

[0176] like Fig. 9 As shown, Fig. 9 is a schematic diagram of updating game props push provided by an embodiment of the present application, such as Fig. 9As shown, after the terminal device 90a obtains the updated game props in the game application and the prop description information of the updated game props (which can be the management object of the game application uploaded to the terminal device 90a), the updated game props (the updated game props here can refer to the data packet of the updated game props) and the prop description information of the updated game props can be sent to the server 90b. The server 90b can extract the label of the updated game props to obtain P first attribute labels of the updated game props, and obtain K candidate objects in the game application (which can be object identification information), and obtain Q second attribute labels of the historical game props set that have been pushed in one or more game applications. Further, the server 90b can obtain the historical interaction data corresponding to the K candidate objects for the historical game props, determine the label preference corresponding to the Q second attribute labels of each candidate object based on the historical interaction data, and determine the label preference corresponding to the P first attribute labels according to the label preference corresponding to the Q second attribute labels. For details, please refer to the contents of the above steps S101-S103, and the embodiments of the present application will not be repeated here. Furthermore, the server 90b can determine the item preference of each candidate object for updating the game item according to the tag preferences corresponding to the P first attribute tags, and determine the push object for updating the game item from the K candidate objects according to the item preference of each candidate object for updating the game item. At the same time, the server 90b can generate push video data about updating the game item and send the push video data to the terminal device 90c corresponding to the push object.

[0177] like Fig. 9 As shown, the terminal device 90c can display the push video data corresponding to the updated game props in the display page 90e, and the push object 90d can view the push video data corresponding to the updated game props in the terminal device 90c. If the push object is interested in the push video data corresponding to the updated game props, the push video data corresponding to the updated game props can be clicked. The terminal device 90c can respond to the click operation of the push object 90d, display the display page 90f, display the prop description information of the updated game props in the display page 90f, and display the purchase control for the updated game props. If the push object needs to purchase the updated game props, the purchase control can be clicked to trigger the purchase operation for the updated game props. It can be seen that determining the push object for updated game props from K candidate objects, that is, determining the push object that is interested in the updated game props from the K candidate objects, can achieve accurate push of updated game props, thereby reducing the cost of pushing game props, and at the same time avoid interfering with candidate objects that are not interested in the updated game props, and at the same time generate push multimedia data about the updated game props, so as to achieve accurate push of updated game props to the push object and improve the push accuracy of updated game props.

[0178] In the embodiment of the present application, the label preference can be used to reflect the candidate object's preference for the game props under the second attribute label, that is, the label preference reflects the prop style that the candidate object is interested in, that is, the label preference can reflect the object label (that is, the user label) of the candidate object, and the second attribute label reflects the prop style of the historical game props in one dimension or multiple dimensions, such as one or more dimensions including color, theme, environment, etc. Therefore, by determining the label preference of the candidate object for the second attribute label of the historical game props based on the candidate object's historical interaction data for the historical game props, that is, one second attribute label corresponds to one label preference, a one-to-one correspondence between the user label and the second attribute label (that is, the prop style) can be achieved, and the problem that the real prop style of the candidate object cannot be accurately obtained due to the inability to correspond one-to-one between the user label and the prop style of the game prop can be avoided, that is, the label preference can truly reflect the prop style of the candidate object. Furthermore, the candidate's item preference for updating game props is determined by the candidate's label preference for the second attribute label, the first attribute label for updating game props, and the second attribute label. The item preference reflects the candidate's preference for updating game props, that is, the item preference can reflect the candidate's need for updating game props. For example, if the candidate's item preference for updating game props is relatively high, it indicates that the candidate has a relatively high need for updating game props; conversely, if the candidate's item preference for updating game props is relatively low, it indicates that the candidate has a relatively low need for updating game props. Therefore, the computer device can select the push object corresponding to the updated game props from the K candidate objects according to the prop preferences corresponding to the K candidate objects, and push the updated game props to the push object, so that the push accuracy of the updated game props can be improved, that is, the updated game props can be accurately pushed to the game objects that really need (or are interested in), and the interest and stickiness of the game objects for the game application can be improved. At the same time, the updated game props can be avoided from being pushed to invalid game objects, reducing the push cost of the game props. In addition, the embodiment of the present application can generate push multimedia data about the updated game props, so as to accurately push the updated game props to the push object, and improve the push accuracy of the updated game props.

[0179] For further information, see Fig.10 , Fig.10 It is a structural diagram of a data processing device provided in an embodiment of the present application. The data processing device may include: an acquisition module 11, a parsing module 12, an identification module 13 and a push module 14.

[0180] The acquisition module 11 is used to acquire P first attribute tags of the updated game props to be pushed, Q second attribute tags of the historical game props in the pushed historical game props set, and historical interaction data of K candidate objects on the historical game props respectively; P and Q are both positive integers, and K is an integer greater than 1;

[0181] The parsing module 12 is used to analyze the candidate object K among the K candidate objects. i The corresponding historical interaction data is parsed to obtain the candidate object K i The label preferences corresponding to the Q second attribute labels respectively; i is a positive integer less than or equal to K;

[0182] The recognition module 13 is used to identify the candidate object K i For the label preferences corresponding to the Q second attribute labels, the Q second attribute labels, and the P first attribute labels, identify the candidate objects K i Item preference for updating game items;

[0183] The push module 14 is used to select a push object corresponding to the updated game prop from the K candidate objects according to the prop preferences of the K candidate objects for the updated game prop, and push the updated game prop to the push object.

[0184] Among them, the P first attribute labels and the Q second attribute labels are all attribute sub-labels;

[0185] The identification module 13 includes: a first determining unit 1301 , a second determining unit 1302 , a searching unit 1303 and a third determining unit 1304 .

[0186] The first determining unit 1301 is used to determine the attribute parent labels to which the P first attribute labels respectively belong according to the general label network, and obtain M first attribute parent labels; the general label network reflects the subordinate relationship between the attribute sub-label and the attribute parent label; M is a positive integer;

[0187] The second determining unit 1302 is used to determine the attribute parent labels to which the Q second attribute labels respectively belong according to the universal label network, and obtain N second attribute parent labels; N is a positive integer;

[0188] A searching unit 1303 is used to search for second attribute parent tags that match the M first attribute parent tags respectively from the N second attribute parent tags, and obtain M matching attribute parent tags;

[0189] The third determining unit 1304 is used to determine the candidate object K i Based on the label preferences of the second attribute labels belonging to the M matching attribute parent labels and the P first attribute labels, determine the candidate objects K iItem preference for updating game items.

[0190] The third determining unit 1304 is specifically configured to:

[0191] From the second attribute tags belonging to the M matching attribute parent tags, find the first attribute tag P among the P first attribute tags. j The matching second attribute tag is used as the first attribute tag P j The corresponding matching second attribute tag; j is a positive integer less than or equal to P;

[0192] The candidate object K i According to the label preference of the matching second attribute label, the candidate object K is determined i For the first attribute label P j 's label preference;

[0193] If we obtain the candidate object K i For the label preferences corresponding to the P first attribute labels, the candidate objects K i The label preferences corresponding to the P first attribute labels are summed up to obtain the candidate object K i Item preference for updating game items.

[0194] The historical game props set includes S historical game props with a second attribute tag x, and the second attribute tag x belongs to any one of the Q second attribute tags; S is a positive integer less than or equal to the total number of historical game props in the historical game props set;

[0195] The parsing module 12 includes: a parsing unit 1201 , a fourth determining unit 1202 and a fifth determining unit 1203 .

[0196] The parsing unit 1201 is used to select a candidate object K from the K candidate objects. i In the corresponding historical interaction data, the candidate object K is parsed i The interaction categories and interaction times of historical game props under Q second attribute labels;

[0197] The fourth determining unit 1202 is used to determine the candidate object K according to the interaction category and the number of interactions. i Interaction preferences for historical game props under Q second attribute labels;

[0198] The fifth determining unit 1203 is used to determine the candidate object K i Corresponding interaction preference, determine the candidate object K i The label preferences corresponding to the Q second attribute labels.

[0199] The fifth determining unit 1203 is specifically configured to:

[0200] Determine the candidate object K i The number of historical game props that participate in the interaction and have a second attribute tag x; the second attribute tag x belongs to any one of Q second attribute tags, and x is a positive integer less than or equal to Q;

[0201] If the number of props is equal to 1, then the candidate object K i The interaction preference of the historical game props with the second attribute label x is determined as the candidate object K i Label preference for the second attribute label x;

[0202] If the number of props is greater than 1, then for candidate object K i The interaction preference of historical game props with the second attribute label x is summed to obtain the candidate object K i The label preference for the second attribute label x.

[0203] The fourth determining unit 1202 is specifically configured to:

[0204] Obtaining the item weight coefficient corresponding to the target historical game item, and obtaining the interaction weight coefficient of the interaction category corresponding to the target historical game item; the target historical game item belongs to any one of the historical game items under the Q second attribute labels;

[0205] If the number of interaction categories corresponding to the target historical game props is equal to 1, the prop weight coefficient corresponding to the target historical game props, the interaction weight coefficient of the interaction category and the number of interactions are multiplied to obtain the candidate object K i Interaction preference for target historical game props;

[0206] If the number of categories is equal to R, then the item weight coefficient corresponding to the target historical game item, the interaction weight coefficient corresponding to the R interaction categories and the number of interactions are multiplied to obtain the interaction preference corresponding to the R interaction categories. The interaction preference corresponding to the R interaction categories is summed to obtain the candidate object K. i The interaction preference for the target historical game path; R is an integer greater than 1.

[0207] The push module 14 includes: a screening unit 1401 and a sixth determining unit 1402 .

[0208] A screening unit 1401 is used to screen candidate objects whose prop preference is greater than or equal to a preference threshold from the K candidate objects;

[0209] The sixth determining unit 1402 is used to determine the screened candidate objects as push objects corresponding to the updated game props.

[0210] The push module 14 further includes: a generating unit 1403 , a first acquiring unit 1404 and a pushing unit 1405 .

[0211] The generating unit 1403 is used to generate push multimedia data about the updated game props according to the prop description information of the updated game props;

[0212] The first acquisition unit 1404 is used to acquire a push strategy corresponding to the updated game props;

[0213] The push unit 1405 is used to push the multimedia data to the terminal device corresponding to the push object based on the push strategy.

[0214] The acquisition module 11 includes: a first extraction unit 1101 , a word segmentation unit 1102 , a filtering unit 1103 and a second extraction unit 1104 .

[0215] The first extraction unit 1101 is used to obtain the item description information about the updated game item to be pushed, and extract the item attribute description text about the updated game item from the item description information about the updated game item;

[0216] The word segmentation unit 1102 is used to perform word segmentation processing on the item attribute description text to obtain initial attribute description words;

[0217] A filtering unit 1103 is used to filter invalid words in the initial attribute description words to obtain attribute description words;

[0218] The second extraction unit 1104 is used to obtain the semantic feature vector of the attribute description word, and perform label extraction on the updated game props according to the semantic feature vector to obtain P first attribute labels of the updated game props.

[0219] The second extraction unit 1104 is specifically used for:

[0220] Obtain F attribute parent labels in the general label network, and attribute sub-labels respectively included in the F attribute parent labels, the general label network is used to reflect the subordinate relationship between the attribute sub-labels and the attribute parent labels; F is a positive integer;

[0221] Call the label extraction model, and determine the attribute parent label to which the attribute description word belongs from the F attribute parent labels according to the semantic feature vector, as the target attribute parent label;

[0222] According to the attribute sub-tags included in the target attribute parent tag, the tag extraction is performed on the updated game props to obtain P first attribute tags of the updated game props.

[0223] Among them, according to the attribute sub-tags included in the target attribute parent tag, the tag extraction is performed on the updated game props to obtain P first attribute tags of the updated game props, including:

[0224] Get the label feature vector corresponding to the attribute sub-label included in the target attribute parent label;

[0225] Obtain feature similarity between the label feature vector and the semantic feature vector;

[0226] Extracting attribute sub-tags whose feature similarity is greater than or equal to a similarity threshold from the attribute sub-tags included in the target attribute parent tag;

[0227] The extracted attribute sub-tags are determined as P first attribute tags for updating the game props.

[0228] The historical game props set includes the historical game props that have been pushed in H game applications; H is a positive integer;

[0229] The acquisition module 11 further includes: a second acquisition unit 1105 and a third extraction unit 1106 .

[0230] The second acquisition unit 1105 is used to acquire the historical log data of K candidate objects respectively related to H game applications;

[0231] The third extraction unit 1106 is used to extract historical interaction data of the K candidate objects with respect to the historical game props in the historical game prop set from the historical log data of the K candidate objects respectively related to the H game applications.

[0232] According to one embodiment of the present application, Fig.10 The various modules in the data processing device shown can be respectively or all combined into one or several units to constitute, or some of the units (some) can also be split into at least two smaller sub-units in function, and the same operation can be achieved without affecting the realization of the technical effects of the embodiments of the present application. The above modules are divided based on logical functions. In practical applications, the functions of one module can also be realized by at least two units, or the functions of at least two modules can be realized by one unit. In other embodiments of the present application, the data processing device may also include other units. In practical applications, these functions can also be implemented with the assistance of other units, and can be implemented by the collaboration of at least two units.

[0233] According to one embodiment of the present application, the program can be executed by running on a general-purpose computer device such as a computer including a central processing unit (CPU), a random access memory medium (RAM), a read-only memory medium (ROM), and other processing elements and storage elements. Figure 3 A computer program (including program code) for each step involved in the corresponding method shown in Fig.10 The data processing device shown in and the data processing method of the embodiment of the present application are implemented. The above computer program can be recorded on, for example, a computer readable recording medium, and loaded into the above computer device through the computer readable recording medium, and run therein.

[0234] In the embodiment of the present application, the label preference can be used to reflect the candidate object's preference for the game props under the second attribute label, that is, the label preference reflects the prop style that the candidate object is interested in, that is, the label preference can reflect the object label (that is, the user label) of the candidate object, and the second attribute label reflects the prop style of the historical game props in one dimension or multiple dimensions, such as one or more dimensions including color, theme, environment, etc. Therefore, by determining the label preference of the candidate object for the second attribute label of the historical game props based on the candidate object's historical interaction data for the historical game props, that is, one second attribute label corresponds to one label preference, a one-to-one correspondence between the user label and the second attribute label (that is, the prop style) can be achieved, and the problem that the real prop style of the candidate object cannot be accurately obtained due to the inability to correspond one-to-one between the user label and the prop style of the game prop can be avoided, that is, the label preference can truly reflect the prop style of the candidate object. Furthermore, the candidate's item preference for updating game props is determined by the candidate's label preference for the second attribute label, the first attribute label for updating game props, and the second attribute label. The item preference reflects the candidate's preference for updating game props, that is, the item preference can reflect the candidate's need for updating game props. For example, if the candidate's item preference for updating game props is relatively high, it indicates that the candidate has a relatively high need for updating game props; conversely, if the candidate's item preference for updating game props is relatively low, it indicates that the candidate has a relatively low need for updating game props. Therefore, the computer device can select the push object corresponding to the updated game props from the K candidate objects according to the prop preferences corresponding to the K candidate objects, and push the updated game props to the push object, so that the push accuracy of the updated game props can be improved, that is, the updated game props can be accurately pushed to the game objects that really need (or are interested in), and the interest and stickiness of the game objects for the game application can be improved. At the same time, the updated game props can be avoided from being pushed to invalid game objects, reducing the push cost of the game props. In addition, the embodiment of the present application can generate push multimedia data about the updated game props, so as to accurately push the updated game props to the push object, and improve the push accuracy of the updated game props.

[0235] For further information, see Fig.11 , Fig.11 is a schematic diagram of the structure of a computer device provided in an embodiment of the present application, and the computer device may be a terminal device or a server. Fig.11As 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: a user interface 1003, and at least one communication bus 1002. Among them, the communication bus 1002 is used to realize the connection and communication between these components. Among them, in some embodiments, the user interface 1003 may include a display screen (Display), a keyboard (Keyboard), and the optional user interface 1003 may also include a standard wired interface and a wireless interface. Optionally, the network interface 1004 may 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 it may be a non-volatile memory (non-volatile memory), such as at least one disk storage. Optionally, the memory 1005 may also be at least one storage device located away from the aforementioned processor 1001. As Fig.11 As shown, the memory 1005 as a computer-readable storage medium may include an operating system, a network communication module, a user interface module, and a device control application program.

[0236] In such Fig.11 In the computer device 1000 shown, the network interface 1004 can provide a network communication function; the user interface 1003 is mainly used to provide an input interface for the user; and the processor 1001 can be used to call the device control application stored in the memory 1005 to achieve:

[0237] Obtain P first attribute tags of the updated game props to be pushed, Q second attribute tags of the historical game props in the pushed historical game props set, and historical interaction data of K candidate objects on the historical game props; P and Q are both positive integers, and K is an integer greater than 1;

[0238] For candidate object K among K candidate objects i The corresponding historical interaction data is parsed to obtain the candidate object K i The label preferences corresponding to the Q second attribute labels respectively; i is a positive integer less than or equal to K;

[0239] According to the candidate K i For the label preferences corresponding to the Q second attribute labels, the Q second attribute labels, and the P first attribute labels, identify the candidate objects K i Item preference for updating game items;

[0240] According to the item preferences of the K candidate objects for the updated game items, push objects corresponding to the updated game items are selected from the K candidate objects, and the updated game items are pushed to the push objects.

[0241] It should be understood that the computer device 1000 described in the embodiment of the present application can also execute the above Figure 7 The description of the data processing method in the corresponding embodiment can also be performed as described above. Fig.10 The description of the data processing device in the corresponding embodiment will not be repeated here. In addition, the description of the beneficial effects of the same method will not be repeated here.

[0242] In addition, it should be pointed out here that: the embodiment of the present application also provides a computer-readable storage medium, and the computer-readable storage medium stores a computer program executed by the data processing device mentioned above, and when the processor executes the computer program, it can execute the above-mentioned Figure 3 or Figure 7 The description of the data processing method in the corresponding embodiment will not be repeated here. In addition, the description of the beneficial effects of the same method will not be repeated. 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.

[0243] In addition, it should be noted that: the embodiment of the present application also provides a computer program product, which may include a computer program, and the computer program may be 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 may execute the computer program, so that the computer device performs the above Figure 3 or Figure 7 The description of the data processing method in the corresponding embodiment will not be repeated here. In addition, the description of the beneficial effects of the same method will not be repeated. For technical details not disclosed in the computer program product embodiment involved in this application, please refer to the description of the method embodiment of this application.

[0244] It should be noted that the data collection and processing in this application should be strictly in accordance with the requirements of relevant national laws and regulations when applied in examples, and the informed consent or separate consent of the personal information subject (or with a legal basis) should be obtained, and the subsequent data use and processing should be carried out within the scope of authorization of laws and regulations and personal information subjects. For example, when this application obtains object information such as historical interaction data of K candidate objects for historical game props, object feature information, and historical log data of K candidate objects for H game applications, it is necessary to obtain the informed consent or separate consent of the corresponding candidate objects.

[0245] Those skilled in the art can understand that all or part of the processes in the above-mentioned embodiments can be implemented by instructing related hardware through a computer program, and the computer program can be stored in a computer-readable storage medium, and when the program is executed, it can include the processes of the embodiments of the above-mentioned methods. Among them, the storage medium can be a disk, an optical disk, a read-only memory (ROM) or a random access memory (RAM), etc.

[0246] The above disclosure is only the preferred embodiment of the present application, which 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: Obtain P first attribute tags of the updated game props to be pushed, Q second attribute tags of the historical game props in the pushed historical game props set, and historical interaction data of K candidate objects on the historical game props respectively; P and Q are both positive integers, and K is an integer greater than 1; For the candidate object K among the K candidate objects i The corresponding historical interaction data is parsed to obtain the candidate object K i The label preferences corresponding to the Q second attribute labels respectively; i is a positive integer less than or equal to K; According to the candidate object K i According to the label preferences corresponding to the Q second attribute labels, the Q second attribute labels and the P first attribute labels, the candidate object K is identified. i An item preference for the updated game item; According to the item preferences of the K candidate objects for the updated game item, a push object corresponding to the updated game item is selected from the K candidate objects, and the updated game item is pushed to the push object.

2. The method according to claim 1, characterized in that The P first attribute tags and the Q second attribute tags are all attribute sub-tags; The candidate object K i According to the label preferences corresponding to the Q second attribute labels, the Q second attribute labels and the P first attribute labels, the candidate object K is identified. i The item preference for the updated game item includes: According to the universal label network, the attribute parent labels to which the P first attribute labels respectively belong are determined to obtain M first attribute parent labels; the universal label network reflects the subordinate relationship between the attribute sub-label and the attribute parent label; M is a positive integer; According to the universal label network, determining the attribute parent labels to which the Q second attribute labels respectively belong, and obtaining N second attribute parent labels; N is a positive integer; From the N second attribute parent tags, searching for second attribute parent tags that match the M first attribute parent tags respectively, to obtain M matching attribute parent tags; According to the candidate object K i According to the label preference of the second attribute label belonging to the M matching attribute parent labels and the P first attribute labels, the candidate object K is determined i An item preference for the updated game item.

3. The method according to claim 2, characterized in that The candidate object K i According to the label preference of the second attribute label belonging to the M matching attribute parent labels and the P first attribute labels, the candidate object K is determined i The item preference for the updated game item includes: From the second attribute tags belonging to the M matching attribute parent tags, search for the first attribute tag P among the P first attribute tags. j The second attribute tag matched is used as the first attribute tag P j The corresponding matching second attribute tag; j is a positive integer less than or equal to P; The candidate object K i According to the label preference of the matching second attribute label, the candidate object K is determined i For the first attribute label P j 's label preference; If the candidate object K is obtained i According to the label preferences corresponding to the P first attribute labels, the candidate objects K i The label preferences corresponding to the P first attribute labels are summed to obtain the candidate object K i An item preference for the updated game item.

4. The method according to claim 1, characterized in that: The candidate object K among the K candidate objects i The corresponding historical interaction data is parsed to obtain the candidate object K i The label preferences corresponding to the Q second attribute labels respectively include: From the K candidate objects, candidate object K i In the corresponding historical interaction data, the candidate object K is parsed i Interaction categories and interaction times of historical game props under the Q second attribute tags; According to the interaction category and the number of interactions, the candidate object K is determined i Interaction preferences for the historical game items under the Q second attribute tags; According to the candidate object K i Corresponding interaction preference, determine the candidate object K i The label preferences respectively corresponding to the Q second attribute labels.

5. The method according to claim 4, characterized in that The candidate object K i Corresponding interaction preference, determine the candidate object K i The label preferences corresponding to the Q second attribute labels respectively include: Determine the candidate object K i The number of historical game props that participate in the interaction and have a second attribute tag x; the second attribute tag x belongs to any one of the Q second attribute tags, and x is a positive integer less than or equal to Q; If the number of props is equal to 1, the candidate object K i The interaction preference of the historical game props with the second attribute label x is determined as the candidate object K i The label preference for the second attribute label x; If the number of props is greater than 1, then the candidate object K i The interaction preference of the historical game props with the second attribute label x is summed up to obtain the candidate object K i The label preference for the second attribute label x.

6. The method according to claim 4, characterized in that The candidate object K is determined according to the interaction category and the number of interactions. i The interaction preferences of the historical game props under the Q second attribute tags include: Obtaining a prop weight coefficient corresponding to a target historical game prop, and obtaining an interaction weight coefficient of an interaction category corresponding to the target historical game prop; the target historical game prop belongs to any one of the historical game props under the Q second attribute labels; If the number of interaction categories corresponding to the target historical game props is equal to 1, the prop weight coefficient corresponding to the target historical game props, the interaction weight coefficient of the interaction category and the number of interactions are multiplied to obtain the candidate object K i Interaction preference for the target historical game item; If the number of categories is equal to R, then the item weight coefficient corresponding to the target historical game item, the interaction weight coefficient corresponding to the R interaction categories, and the number of interactions are multiplied to obtain the interaction preference corresponding to the R interaction categories, and the interaction preference corresponding to the R interaction categories is summed to obtain the candidate object K. i The interaction preference for the target historical game path; R is an integer greater than 1.

7. The method according to claim 1, characterized in that The step of selecting the push object corresponding to the updated game prop from the K candidate objects according to the prop preference degrees respectively corresponding to the K candidate objects includes: From the K candidate objects, select candidate objects whose item preference is greater than or equal to the preference threshold; The screened candidate objects are determined as push objects corresponding to the updated game props.

8. The method according to claim 1, characterized in that The pushing the updated game props to the push object includes: Generate push multimedia data about the updated game props according to the prop description information of the updated game props; Obtaining the push strategy corresponding to the updated game props; Based on the push strategy, the push multimedia data is pushed to the terminal device corresponding to the push object.

9. The method according to claim 1, characterized in that: The step of obtaining P first attribute tags of the updated game props to be pushed includes: Acquire item description information about the updated game item to be pushed, and extract item attribute description text about the updated game item from the item description information of the updated game item; Performing word segmentation processing on the item attribute description text to obtain initial attribute description words; Filtering invalid words in the initial attribute description words to obtain attribute description words; The semantic feature vector of the attribute description word is obtained, and a label is extracted for the updated game prop according to the semantic feature vector to obtain P first attribute labels of the updated game prop.

10. The method according to claim 9, characterized in that The step of extracting labels from the updated game props according to the semantic feature vector to obtain P first attribute labels of the updated game props includes: Obtain F attribute parent tags in a general tag network, and attribute sub-tags respectively included in the F attribute parent tags, wherein the general tag network is used to reflect the subordinate relationship between the attribute sub-tags and the attribute parent tags; F is a positive integer; Calling a label extraction model to determine, according to the semantic feature vector, the attribute parent label to which the attribute description term belongs from the F attribute parent labels as the target attribute parent label; According to the attribute sub-tags included in the target attribute parent tag, the tag extraction is performed on the updated game prop to obtain P first attribute tags of the updated game prop.

11. The method according to claim 10, characterized in that The step of extracting a tag from the updated game item according to the attribute subtags included in the target attribute parent tag to obtain P first attribute tags of the updated game item includes: Obtaining a label feature vector corresponding to an attribute sub-label included in the target attribute parent label; Obtaining feature similarity between the label feature vector and the semantic feature vector; Extracting, from the attribute subtags included in the target attribute parent tag, the attribute subtags whose feature similarity is greater than or equal to a similarity threshold; The extracted attribute sub-tags are determined as the P first attribute tags of the updated game props.

12. The method according to claim 1, characterized in that The historical game props set includes historical game props that have been pushed in H game applications; H is a positive integer; The obtaining of the historical interaction data of the K candidate objects respectively with respect to the historical game props includes: Obtain historical log data of the K candidate objects respectively related to the H game applications; From the historical log data of the K candidate objects respectively related to the H game applications, historical interaction data of the K candidate objects respectively with respect to the historical game props in the historical game prop set are extracted.

13. A data processing device, characterized in that: include: An acquisition module, used to acquire P first attribute tags of the updated game props to be pushed, Q second attribute tags of the historical game props in the pushed historical game props set, and historical interaction data of K candidate objects respectively about the historical game props; P and Q are both positive integers, and K is an integer greater than 1; A parsing module is used to analyze the candidate object K among the K candidate objects. i The corresponding historical interaction data is parsed to obtain the candidate object K i The label preferences corresponding to the Q second attribute labels respectively; i is a positive integer less than or equal to K; A recognition module is used to identify the candidate object K i According to the label preferences corresponding to the Q second attribute labels, the Q second attribute labels and the P first attribute labels, the candidate object K is identified. i An item preference for the updated game item; The push module is used to select a push object corresponding to the updated game prop from the K candidate objects according to the prop preferences of the K candidate objects for the updated game prop, and push the updated game prop to the push object.

14. A computer device, characterized in that: include: Processor and memory; The processor is connected to the memory, wherein 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 12.

15. 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 12.

16. 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 that a computer device having the processor executes the method according to any one of claims 1 to 12.

Images