CN114470791A

CN114470791A - Game item recommendation method and device, computer equipment, storage medium and product

Info

Publication number: CN114470791A
Application number: CN202210135377.2A
Authority: CN
Inventors: 陶冶; 江鑫; 刘阳; 邹丰富; 李鹏飞; 徐广根
Original assignee: Tencent Technology Shenzhen Co Ltd
Current assignee: Tencent Technology Shenzhen Co Ltd
Priority date: 2022-02-14
Filing date: 2022-02-14
Publication date: 2022-05-13

Abstract

The application provides a game prop recommendation method, a game prop recommendation device, computer equipment, a storage medium and a product, and relates to the technical fields of computer networks, games and the like. Dividing an object into a plurality of levels having different game skills by determining a target level of the target object based on game skill information of the target object; determining a first object set based on the target hierarchy and the object characteristics of the target object, and determining a first object set which is higher in hierarchy than the target object and similar to the target object by combining the hierarchy of the object, the object characteristics and the like; subsequently, further determining a recommendation set of game items used by the first object set in the objects so as to recommend the game items in the recommendation set to the target object; so that the low-level object can learn the prop usage habits of objects that are similar to him and have more game skills than him. And the game logic and complex characteristic engineering do not need to be deeply analyzed, so that the practicability of game property recommendation is improved.

Description

Game item recommendation method and device, computer equipment, storage medium and product

Technical Field

The application relates to the technical field of computer networks, games and the like, in particular to a game prop recommendation method, device, computer equipment, storage medium and product.

Background

Along with the rapid development of the game industry, nowadays, various types of games come out endlessly, and the competition of the game industry is more and more intense. Many of these games provide virtual objects representing the user's image with game objects, which are virtual items used in the game. In order to increase the purchase rate of users for game items, many game platforms will actively recommend game items.

In the related art, game item recommendations may include: one is to recommend the same item list to each object by manually configuring a fixed item list. One is based on the interest characteristics and purchasing behavior data of the user, usually a model is used for constructing complex characteristic engineering, and a large amount of data are mined and calculated to match game props interested by the user so as to recommend the game props interested by the user.

The method is adopted to recommend the props, or the prop lists of all users are the same, so that the purchasing desire of the users is low, namely the actual recommending efficiency is low; or complex characteristic engineering and large calculation amount are needed, an algorithm engineer needs to deeply know the advantages and disadvantages of all properties of the game, a large amount of labor cost is needed for analysis, and the recommendation cost is high. Therefore, the game property recommendation method in the related art is poor in practicability.

Disclosure of Invention

The application provides a game property recommendation method, a game property recommendation device, computer equipment, a storage medium and a product, which can solve the problem of poor practicability of the game property recommendation method in the related technology. The technical scheme is as follows:

in one aspect, a game item recommendation method is provided, the method comprising:

determining a target level of a target object based on game skill information of the target object;

determining a first object set based on the target hierarchy and object features of target objects, wherein the hierarchy of the objects in the first object set is higher than the target hierarchy, and feature similarity between the objects in the first object set and the target objects meets a target condition;

and determining a recommendation set of the target object based on the item use information of the first object set, and recommending game items in the recommendation set to the target object.

In another aspect, a game item recommendation device is provided, the device comprising:

the level determining module is used for determining a target level of a target object based on game skill information of the target object;

the object set determining module is used for determining a first object set based on the target hierarchy and object features of target objects, the hierarchy of the objects in the first object set is higher than the target hierarchy, and feature similarity between the objects in the first object set and the target objects meets a target condition;

and the recommending module is used for determining a recommending set of the target object based on the item using information of the first object set and recommending the game items in the recommending set to the target object.

In another aspect, a computer device is provided, which includes a memory, a processor, and a computer program stored on the memory, wherein the processor executes the computer program to implement the game item recommendation method described above.

In another aspect, a computer-readable storage medium is provided, on which a computer program is stored, which, when executed by a processor, implements the game item recommendation method described above.

In another aspect, a computer program product is provided, comprising a computer program, which when executed by a processor implements the game item recommendation method described above.

The technical scheme provided by the embodiment of the application has the following beneficial effects:

according to the game item recommendation method provided by the embodiment of the application, the target level of the target object is determined based on the game skill information of the target object, so that the object is divided into a plurality of levels with different game skills; determining a first object set based on the target hierarchy and the object characteristics of the target object, and determining a first object set which is higher in hierarchy than the target object and similar to the target object by combining the hierarchy of the object, the object characteristics and the like; subsequently, a recommendation set comprising game items used by the first object set is further determined, so that the game items in the recommendation set are recommended to the target object; the method and the device enable the low-level object to learn the use habit of the props to the high-level object, and the game skill of the high-level object is more skilled than that of the low-level object, so that the possibility that the recommended props are converted into effective props is improved, and the actual recommendation efficiency is improved. In addition, the game logic and the advantages and disadvantages of all the props do not need to be deeply analyzed, the frequently-selected props of the feature similar objects of higher levels can be recommended in a self-adaptive mode based on the levels and the feature similarity, complex feature engineering and deep understanding of the game logic are not needed, and labor cost is greatly saved. By isolating the game logic and game property recommendation process of different games and without an individualized machine learning model and the like, the method can be reused for any game to perform property recommendation, has strong reusability, universality and mobility, and can also learn the recommendation mode of using the property to a high level through a low level even if the game property is the latest one, so that the possibility that a new property is recommended and purchased is improved, the problem of cold start of the property is well solved, and the practicability of game property recommendation is further improved.

Drawings

In order to more clearly illustrate the technical solutions in the embodiments of the present application, the drawings used in the description of the embodiments of the present application will be briefly described below.

FIG. 1 is a schematic diagram of an implementation environment of a method for recommending game items according to an embodiment of the present application;

FIG. 2 is a schematic flow chart illustrating a method for recommending game items according to an embodiment of the present disclosure;

FIG. 3 is a schematic diagram of an object hierarchy provided in an embodiment of the present application;

FIG. 4 is a schematic view of a game item recommendation process provided in an embodiment of the present application;

FIG. 5 is a schematic structural diagram of a game item recommendation device according to an embodiment of the present application;

fig. 6 is a schematic structural diagram of a computer device according to an embodiment of the present application.

Detailed Description

Embodiments of the present application are described below in conjunction with the drawings in the present application. It should be understood that the embodiments set forth below in connection with the drawings are exemplary descriptions for explaining technical solutions of the embodiments of the present application, and do not limit the technical solutions of the embodiments of the present application.

As used herein, the singular forms "a", "an", "the" and "the" are intended to include the plural forms as well, unless the context clearly indicates otherwise. It should be further understood that the terms "comprises" and/or "comprising," when used in this specification in connection with embodiments of the present application, specify the presence of stated features, information, data, steps, operations, elements, and/or components, but do not preclude the presence or addition of other features, information, data, steps, operations, elements, components, and/or groups thereof, as embodied in the art. It will be understood that when an element is referred to as being "connected" or "coupled" to another element, it can be directly connected or coupled to the other element or intervening elements may be present. Further, "connected" or "coupled" as used herein may include wirelessly connected or wirelessly coupled. The term "and/or" as used herein indicates at least one of the items defined by the term, e.g., "a and/or B" may be implemented as "a", or as "B", or as "a and B".

It is understood that in the embodiments of the present application, any data related to the user, such as game level, resource value of transaction prop, transaction record, object online time, object registration time, number of times of activity field parameters, etc., is referred to, when the following embodiments of the present application are applied to specific products or technologies, user permission or consent needs to be obtained, and collection, use and processing of the related data need to comply with relevant laws and regulations and standards of relevant countries and regions.

To make the objects, technical solutions and advantages of the present application more clear, embodiments of the present application will be described in further detail below with reference to the accompanying drawings.

The game item recommendation method provided by the embodiment of the application relates to the following artificial intelligence technology, computer vision technology, big data, cloud games and the like. For example, cloud computing techniques among artificial intelligence techniques may be utilized to rank individual objects in a game based on their game skill information; of course, it is also possible to store a plurality of items of data such as the hierarchy, characteristics, game skill information, and property use information of each object in the game by using techniques such as a distributed database and a distributed file system in the big data technique. Illustratively, the recommended game items may also be presented in a game page using computer vision techniques. For example, each object may be played online in a web page by using cloud game technology, or may be played by using a game platform provided with multiple game entries or a standalone game application, which is not limited in this embodiment of the present application.

Artificial Intelligence (AI) is a theory, method, technique and application system that uses a digital computer or a machine controlled by a digital computer to simulate, extend and expand human Intelligence, perceive the environment, acquire knowledge and use the knowledge to obtain the best results. In other words, artificial intelligence is a comprehensive technique of computer science that attempts to understand the essence of intelligence and produce a new intelligent machine that can react in a manner similar to human intelligence. Artificial intelligence is the research of the design principle and the realization method of various intelligent machines, so that the machines have the functions of perception, reasoning and decision making.

The artificial intelligence technology is a comprehensive subject and relates to the field of extensive technology, namely the technology of a hardware level and the technology of a software level. The artificial intelligence infrastructure generally includes technologies such as sensors, dedicated artificial intelligence chips, cloud computing, distributed storage, big data processing technologies, operation/interaction systems, mechatronics, and the like. The artificial intelligence software technology mainly comprises a computer vision technology, a voice processing technology, a natural language processing technology, machine learning/deep learning, automatic driving, intelligent traffic and the like.

Computer Vision technology (CV) Computer Vision is a science for researching how to make a machine "see", and further refers to that a camera and a Computer are used for identifying and measuring a target instead of human eyes, and further image processing is performed, so that the Computer processing becomes an image more suitable for human eyes to observe or is transmitted to an instrument to detect. As a scientific discipline, computer vision research-related theories and techniques attempt to build artificial intelligence systems that can capture information from images or multidimensional data. Computer vision techniques typically include image processing, image recognition, image semantic understanding, image retrieval, OCR, video processing, video semantic understanding, video content/behavior recognition, three-dimensional object reconstruction, 3D techniques, virtual reality, augmented reality, and the like.

Big data (Big data) refers to a data set which cannot be captured, managed and processed by a conventional software tool within a certain time range, and is a massive, high-growth-rate and diversified information asset which can have stronger decision-making power, insight discovery power and flow optimization capability only by a new processing mode. With the advent of the cloud era, big data has attracted more and more attention, and the big data needs special technology to effectively process a large amount of data within a tolerance elapsed time. The method is suitable for the technology of big data, and comprises a large-scale parallel processing database, data mining, a distributed file system, a distributed database, a cloud computing platform, the Internet and an extensible storage system.

Cloud gaming (Cloud gaming), also known as game on demand (gaming), is an online gaming technology based on Cloud computing technology. Cloud gaming technology enables light-end devices (thin clients) with relatively limited graphics processing and data computing capabilities to run high-quality games. In a cloud game scene, a game is not operated in a player game terminal but in a cloud server, and the cloud server renders the game scene into a video and audio stream which is transmitted to the player game terminal through a network. The player game terminal does not need to have strong graphic operation and data processing capacity, and only needs to have basic streaming media playing capacity and capacity of acquiring player input instructions and sending the instructions to the cloud server.

Fig. 1 is a schematic diagram of an implementation environment of a game item recommendation method provided in the present application. As shown in fig. 1, the implementation environment includes: a computer device 101 and a terminal 102, where the computer device 101 may be a background server of a game application or a cloud server supporting a game scene, and the like. The terminal 102 may be a terminal of a target object. The computer device 101 may interact with the terminal 102 to send the recommended play items for the target object to the terminal 102.

In one possible scenario example, the terminal 102 may be installed with a game application, the computer device 101 sends the recommended game item to the terminal 102 based on the game application, and the terminal 102 may display item information such as an icon, a name and the like of the recommended game item in a page of the game application.

In another example scenario, the computer device 101 and the terminal stick 102 may interact with data based on cloud gaming technology. For example, the terminal 102 may display an online game page of a cloud game, and the computer device 101 may transmit game screen rendering data including items to be recommended to the terminal 102. Of course, the present application is only illustrated by the above two examples, and the present application is not limited to the scenario. For example, the computer device 101 may further send the item to be recommended to the terminal 102 based on the live game application, and the terminal 102 pops up a recommendation page including the item to be recommended in a page of the live game application, or displays a website link corresponding to the recommendation page, and the like.

The target object is used to represent an avatar of the user in the game, for example, the target object may be a game character in the game. The user may set at least one avatar in a game. The game item may be a virtual item that interacts with an in-game object, having an effect on an attribute of the in-game object. A user may purchase a game item through a transaction resource, which may be a tool that measures the virtual value of the game item, a medium from which the game item is purchased, e.g., a transaction resource may be a virtual medium available in a game for trading items, e.g., a game coin, a virtual gold coin, a virtual diamond, etc.

The gaming applications include, but are not limited to: shooting games, multiplayer online tactical sports games, multiplayer gunfight type survival games, role playing games, instant strategy games, racing games, music games, chess and card type games, and the like. Play items include, but are not limited to: attack props, defense props, dressing props, virtual appliances, virtual medicines, virtual ammunition, virtual keys, virtual cards, fragments, and the like.

The server may be an independent physical server, a server cluster or a distributed system formed by a plurality of physical servers, or a cloud server or a server cluster providing basic cloud computing services such as a cloud service, a cloud database, cloud computing, a cloud function, cloud storage, a Network service, cloud communication, middleware service, a domain name service, a security service, a CDN (Content Delivery Network), a big data and artificial intelligence platform. Such networks may include, but are not limited to: a wired network, a wireless network, wherein the wired network comprises: a local area network, a metropolitan area network, and a wide area network, the wireless network comprising: bluetooth, Wi-Fi, and other networks that enable wireless communication. The terminal may be a smart phone (e.g., an Android phone, an iOS phone, etc.), a smart voice interaction device, a smart home appliance, a vehicle-mounted terminal, a tablet computer, a notebook computer, a digital broadcast receiver, an MID (Mobile Internet Devices), a PDA (personal digital assistant), a desktop computer, a vehicle-mounted terminal (e.g., a vehicle-mounted navigation terminal, a vehicle-mounted computer, etc.), a smart speaker, a smart watch, etc., and the terminal and the server may be directly or indirectly connected through a wired or wireless communication manner, but are not limited thereto. The determination may also be based on the requirements of the actual application scenario, and is not limited herein.

Fig. 2 is a schematic flow chart of a game item recommendation method provided in the embodiment of the present application. The execution subject of the method may be a computer device. As shown in fig. 2, the method includes the following steps.

Step 201, the computer device determines a target level of the target object based on the game skill information of the target object.

The game skill information is used to indicate proficiency in game skill of the target object. The computer device may obtain game skill information of the target object, and determine a target level corresponding to the game skill information according to the proficiency level of the game skill of the target object indicated by the game skill information. The higher the proficiency of the game skill of the target object, the higher the target level. For example, the computer device may hierarchically divide the objects according to a hierarchical division rule based on game skill information of the objects. The hierarchical division rule includes: the more proficient the game skill of the object in the game, the higher the level the object corresponds to. The hierarchy of the object can be represented by numerical values, and the height of the hierarchy can be changed corresponding to the change of the size of the numerical values. For example, a first level is higher than a second level, which is higher than a third level, and so on; for example, the first level is the highest level, the game skill of the object at the first level being most proficient; the second level is a next highest level, the game skills of the objects in the second level are next to each other, and so on.

In one possible embodiment, the game skill information is information representing proficiency in mastering and applying game logic; the higher the proficiency level of the target object in mastering and applying game logic is, the higher the target level of the target object is; the game skill information comprises at least one of game grade, resource value of trade prop, object online time length, object registration time and activity participation frequency; accordingly, in this step, the process of determining, by the computer device, the target level of the target object based on the game skill information of the target object may be implemented in at least one of the following manners.

The first way, the game skill information includes game level. And the computer equipment determines the object hierarchy of the target object as the target hierarchy corresponding to the target hierarchy section when responding to that the game hierarchy of the target object is positioned in the target hierarchy section.

The game level may be a level of the target object in the game, and a higher level indicates a more proficient game skill of the target object, and a better grasp of game logic in the game. The computer device may extract a game level of the target object from object data of the target object.

The computer device may store in advance an association relationship between a level section and a level of an object, may determine a target level section in which a game level of the target object is located, and may acquire, according to the target level section, a level corresponding to the target level section from the association relationship between the level section and the level of the object, and may take the acquired level as a target level of the target object.

In one possible example, the computer device may divide the hierarchy of objects in the game based on the game level. For example, the computer device may divide objects belonging to the same level section into the same hierarchy based on the level section. The higher the game level is, the more proficient the game skill of the target object is, the higher the corresponding target level is, and the more core the object is. For example, the first level is the highest level, the second level is the next highest level, each game level in the level section corresponding to the first level is greater than the game level in the level section corresponding to the second level, and so on.

In one possible example, the computer device may determine a level interval corresponding to each level using a location of an inflection point in the level distribution curve. Illustratively, the process of the computer device determining the association between the level interval and the level of the object includes: the computer equipment can determine a grade distribution curve corresponding to the game based on the game grade of each object in the game, wherein the grade distribution curve comprises a variation relation between the game grade and the number of the corresponding objects; the computer equipment determines a plurality of grade intervals based on the inflection point positions of the grade distribution curve, and respectively associates the grade intervals with a plurality of grades according to the grades included in the grade intervals; wherein, the higher the grade in the grade interval, the higher the corresponding grade level of the grade interval. For example, the inflection point may be an extreme point, a maximum point, a minimum point, or the like of the curve. For example, the grade distribution curve may be a curve in a two-dimensional coordinate system having the game grade as the horizontal axis and the number of objects as the vertical axis. The computer device may use an abscissa corresponding to a position of an inflection point in the equal-grade distribution curve as an endpoint of a grade interval to be divided, and sequentially obtain a plurality of grade intervals.

In a second mode, the game skill information includes resource values for trading items. And when the computer equipment responds that the resource value of the transaction prop of the target object is positioned in the target resource interval, determining the object level of the target object as the target level corresponding to the target resource interval.

The resource value of the trading item may be the amount of trading resource consumed by the target object when trading the item in the game. The larger the resource value of the trading prop is, the more the trading prop is, the higher the trading degree of the prop of the target object is, the larger the trading potential is correspondingly, and the more proficient the game skill is. The computer equipment can extract the transaction record of the target object and count the resource value of the transaction prop of the target object. It should be noted that the data such as the resource value of the transaction record and the transaction prop are obtained after the user agrees.

The computer device may pre-store an association relationship between a resource interval and a hierarchy of an object, determine a target resource interval in which a resource value of a transaction prop of the target object is located, obtain, according to the target resource interval, a hierarchy corresponding to the target resource interval from the association relationship between the resource interval and the hierarchy of the object, and use the obtained hierarchy as a target hierarchy of the target object.

In one possible example, the computer device may divide the hierarchy of objects in the game based on resource values of the transaction props. It should be noted that, the larger the resource value of the transaction item is, the more or higher the game item used in the game of the target object is, the more skilled the corresponding game skill is, the higher the corresponding target level is, and the more core the object is. For example, the first hierarchy is the highest hierarchy, the second hierarchy is the next highest hierarchy, each resource value in the resource interval corresponding to the first hierarchy is greater than the resource value in the resource interval corresponding to the second hierarchy, and so on.

In one possible example, the computer device may determine the resource intervals corresponding to the respective levels by using inflection positions in the resource distribution curve of the trading prop. Illustratively, the process of the computer device determining an associative relationship between the resource interval and the hierarchy of objects includes: the computer equipment can determine a resource distribution curve corresponding to the game based on the resource numerical values of the trading objects of all the objects in the game, wherein the resource distribution curve comprises the resource numerical values of the trading objects and the change relation between the number of the corresponding objects; the computer equipment determines a plurality of resource intervals based on the inflection point position of the resource distribution curve, and correspondingly associates the resource intervals with a plurality of levels respectively based on the magnitude of resource numerical values included in the resource intervals; in the resource intervals, the larger the resource value in the resource interval is, the higher the hierarchy corresponding to the resource interval is. The process of obtaining the association relationship between the resource interval and the hierarchy by the computer device using the resource distribution curve is the same as the above first way of obtaining the association relationship between the level interval and the hierarchy by using the level distribution curve, and is not described herein again.

And in a third mode, the game skill information comprises the online time length of the object. And the computer equipment determines the object hierarchy of the target object as the target hierarchy corresponding to the target duration interval when responding to the situation that the online duration of the target object is located in the target duration interval.

The object online duration may be a cumulative duration that the target object is online at least once within a specified period. The specified time period may be approximately one month, approximately one week, etc. The numerical value of the online time length of the object indicates that the longer the time of the target object game is, the higher the activity degree of the target object in the game is, and the more skillful the corresponding game skill is; for example, object A, which was online for a period of 2 hours in the last month, and object B, which was online for a period of 50 hours in the last month, obviously object B played more time and object B played more proficient. The computer device can extract the online record of the target object in a specified period and count the accumulated time length of at least one online in the specified period. It should be noted that the online record, the online duration of the object, and other data are obtained after the user agrees.

The computer device may pre-store an association relationship between the duration interval and the hierarchy of the object, determine a target duration interval in which the object online duration of the target object is located, obtain, according to the target duration interval, a hierarchy corresponding to the target duration interval from the association relationship between the duration interval and the hierarchy of the object, and use the obtained hierarchy as the target hierarchy of the target object.

In one possible example, the computer device may divide the hierarchy of objects in the game based on the length of time the objects are online. It should be noted that the larger the object online time length is, the more the object game time is described, the more proficient the game skill is grasped in the game time, the higher the object level is, and the more the object is core. For example, the value of each online duration in the duration interval corresponding to the first level is greater than the value of the online duration in the duration interval corresponding to the second level, and so on. In one possible example, the computer device may determine the duration intervals corresponding to the respective levels using inflection positions in the online duration distribution curve. The process is the same as the process of determining the corresponding interval based on the inflection point position in the curve in the first and second modes, and is not repeated here.

A fourth mode, the game skill information includes an object registration time. And the computer equipment determines the object hierarchy of the target object as the target hierarchy corresponding to the target time interval when responding to the object registration time of the target object in the target time interval.

The object registration time may be a time when the user registers a game account in the game, a time when the game is started for the first time, a time when the user configures the target object in the game, or the like. The earlier the object registration time is, the longer the target object or the user represented by the target object is in contact with the game is, the more the game experience accumulated by the target object is, and the more proficient the corresponding game skill is. It should be noted that the object registration time is obtained after the user agrees.

The computer device may store an association relationship between a time interval and a hierarchy of an object in advance, determine a target time interval in which an object registration time of the target object is located, acquire a hierarchy corresponding to the target time interval from the association relationship between the time interval and the hierarchy of the object according to the target time interval, and use the acquired hierarchy as a target hierarchy of the target object.

The earlier the object registration time is, the longer the target object game time is, the more skilled the corresponding game skill is, the higher the corresponding target level is, and the more core the object is. For example, the first level is the highest level, the second level is the next highest level, each time in the time interval corresponding to the first level is earlier than the time in the time interval corresponding to the second level, and so on. In one possible example, the computer device may determine time intervals corresponding to respective tiers using inflection locations in a registration time profile. The process is the same as the process of determining the corresponding interval based on the inflection point position in the curve in the first and second modes, and is not repeated here.

Fifth, the game skill information includes activity participation. And when the computer equipment responds that the activity participation times of the target object are positioned in the target time interval, determining the object hierarchy of the target object as the target hierarchy corresponding to the target time interval.

The number of activity engagements may be game-related activities that the user has engaged in over a specified period of time. For example, the number of activity parameters in the last month and in the last week. Illustratively, the game may be a specified game. The game-related activities are not limited to online and offline activities; for example, online mission-earned rewards campaigns, free point-card-earned campaigns, discounted prop offers, and the like; or live events provided by off-line live game stands. For example, the more times of activity participation, the higher the activity level of the target object or the user represented by the target object in the game, the higher the interest level of the game, and the more proficient the corresponding game skill. It should be noted that the number of activity participation is obtained after the user agrees.

The computer device may store an association relationship between a frequency interval and a hierarchy of an object in advance, determine a target frequency interval in which the activity participation frequency of the target object is located, acquire a hierarchy corresponding to the target frequency interval from the association relationship between the frequency interval and the hierarchy of the object according to the target frequency interval, and use the acquired hierarchy as a target hierarchy of the target object.

The more times of activity participation, the higher the interest degree of the target object in the game, the more skilled the corresponding game skill, the higher the corresponding target level and the more core the object. For example, the first hierarchy is the highest hierarchy, the second hierarchy is the next highest hierarchy, the number of the activity participation times in the time interval corresponding to the first hierarchy is greater than the number of the activity participation times in the time interval corresponding to the second hierarchy, and so on. In one possible example, the computer device may determine the time intervals corresponding to the respective levels using inflection positions in the activity participation time distribution curve. The process is the same as the process of determining the corresponding interval based on the inflection point position in the curve in the first and second modes, and is not repeated here.

It should be noted that the above five ways are respectively ways of determining the target hierarchy based on one kind of information. The computer device may also combine any two or more of the five approaches described above to determine the target level. In one example, the first mode and the second mode are combined for illustration. The computer device may be configured with level intervals and resource intervals corresponding to respective levels, and may determine, based on the game level of the target object and the resource value of the transaction item, a target level interval in which the game level is located and a target resource interval in which the resource value of the transaction item is located, and determine, according to the target level interval and the target resource interval, a target level corresponding to the target level interval and the target resource interval.

In still another example, the first mode and the second mode are combined for illustration. The computer device can also calculate scores corresponding to the information, and determine the target level based on the scores corresponding to the at least two kinds of information. The computer device is configured with an association between a plurality of hierarchies and a plurality of scoring components. The computer equipment can also calculate a first score of the target object based on the weight of the game level and the level score corresponding to the game level of the target object; calculating to obtain a second score of the target object based on the weight of the resource of the transaction prop and the score corresponding to the resource value of the transaction prop of the target object; taking the sum of the first score and the second score as the total score of the target object; and taking the corresponding hierarchy between the target scoring areas where the total score is located as the target hierarchy of the target object. It should be noted that, as for the implementation process when the first mode and the second mode are combined, only the two examples are illustrated, but the actual mode of combining the first mode and the second mode is not limited. The above description is given by way of example only in combination of the first embodiment and the second embodiment. Of course, for any other combination manner of the five manners, the corresponding implementation process is the same as the implementation process of the combination of the first manner and the second manner, which is described above, and is not described herein again.

Fig. 3 is a schematic diagram of an object hierarchy provided in an embodiment of the present application. As shown in fig. 3, the computer device may layer the in-game objects to separate high-level, high-activity users. For example, the objects may be layered according to game levels, resource values of trade items, online time lengths of the objects, object registration time, and activity participation times, or may be layered according to frequency of trade items, and the objects may be divided into core player levels of various degrees. The core degrees of users in different layers are different, and the smaller the layer number is, the more core users are. For example, the highest level is the first level which is the most core player, the second level which is the core player, the third level which is the sub-core player, the fourth level which is the sub-core player, the fifth level which is the active player, and the like, and the payment, the game level, the inactivity of the online time period lower than the fifth level (for example, the login number is lower than 3, the payment is 0, and the like) can be divided into other players. Wherein, the low-level group B can learn the prop using habit of the high-level group A.

The target level of the target object is determined based on game skill information, the level division of each object in the game based on the game skill is realized, the item recommendation aiming at different levels is realized based on the subsequent steps, the learning of game item use is carried out on a high-level group by a low-level group, the complicated characteristic engineering is not needed, the deep understanding of game logic is not needed, and the item recommendation process has strong reusability. Further, the game skill information comprises at least one of game grade, resource value of trade prop, on-line time length of the object, object registration time and activity participation frequency; the level of each object is accurately positioned by one or more items of information included in the game skill information, so that the skill proficiency of the object can be accurately measured by the level of the object, and the accuracy of item recommendation is further improved.

Step 202, the computer device determines a first set of objects based on the target hierarchy and object characteristics of the target objects.

The level of the objects in the first object set is higher than the target level, and the feature similarity between the objects in the first object set and the target object meets a target condition. The computer device determines a second set of objects based on the target hierarchy, the hierarchy of objects in the second set of objects being higher than the target hierarchy; the computer device screens out a first object set which meets a target condition with the feature similarity of the target object from the second object set based on the feature of the target object. Wherein the target condition may include that the feature similarity is not lower than a target similarity threshold. Wherein the computer device may determine feature similarities between the target object and each object in the second set of objects, respectively, based on the features of the target object and the objects in the second set of objects; the computer device screens out the first set of objects from the second set of objects based on a feature similarity between the target object and each object in the second set of objects, respectively. For example, the computer device may obtain a feature vector of the target object based on object data of the target object. For example, the computer device may use a trained feature extraction network to input the object data of the target object into the feature extraction network, so as to obtain the feature vector output by the feature extraction network. The feature vector may characterize features of the target object in multiple dimensions. The object data includes, but is not limited to: game skill information of the target object, win or loss of each game, team formation information at the time of the game, and the like, it is noted that the object data is obtained after authorization and approval by the user.

In one possible implementation, prior to determining the first set of objects based on the target hierarchy and the object characteristics of the target objects, the computer device may also determine characteristics of the respective objects in the game application such that a similarly characterized set of objects is determined based on the characteristics of the objects in step 202. For example, the process of determining the characteristics of each object may include: the computer device determines a characteristic of each object based on team information for at least two objects in the gaming application. For example, the team information of each object may include a teammate object of the object, and the teammate object may be an object of a teammate in a game team in which the object is located as the object. The computer device may learn a team characteristic of each object based on team information of at least two objects to obtain a characteristic of each object.

In one possible example, the computer device may perform machine learning in a graph-like manner to obtain the features of each object. Illustratively, the process of determining the characteristic of each object based on the team information of at least two objects in the game application may be implemented through the following steps S1 to S3.

Step S1, the computer device constructs an undirected graph based on the team information of the at least two objects.

The undirected graph comprises at least two nodes corresponding to at least two objects, and an edge exists between the two nodes corresponding to two objects which are mutually grouped in the at least two objects. Illustratively, according to the team formation information of at least two objects in the game application, an undirected graph is constructed with at least two objects as nodes, wherein there is a variable basis for whether teams between the objects are formed as corresponding nodes, for example, when two objects have been formed with each other in the last month, an edge exists between two nodes corresponding to the two objects. This results in a large undirected graph that may not include weights.

For example, an undirected graph G ═ (V, E) can be constructed using team information for only one month for all objects in a certain game application; wherein, V represents a node in the undirected graph, and the node represents a corresponding object; e denotes an edge in the undirected graph,

step S2, the computer device acquires at least two sequences from the undirected graph in a random walk manner.

Each sequence includes objects arranged in a walking order during the walking process. After the graph G is constructed, any one node is taken as an initial node, and one node is randomly selected from a plurality of nodes with edges between the node and the previous node during each wandering until the maximum step length t is reached, namely the wandering times reach t times, so that a sequence with the length of t is obtained. Furthermore, this truncated random walk approach provides two advantages: on one hand, local exploration wandering is easy to parallelize, and several different walkers can wander a plurality of different lines at the same time. On the other hand, information is acquired from the truncated random walks, and repeated learning is not needed when the graph structure is changed slightly.

The random walk may be a fixed length random walk. The random walk is for sampling work to collect training data. The principle of random walk mainly includes: and randomly finding the next node as a root node for walking by taking each node in the undirected graph, wherein the walking length is fixed t. Each node may have multiple walkers to walk in parallel, which are hyper-parameters that may be configured based on need. After random walk, a plurality of sequence samples are obtained. These sequence samples can be used as training samples.

Step S3, the computer device inputs the at least two sequences into a target model, and performs unsupervised training on the target model to obtain the features of each object.

The computer equipment inputs the at least two sequences into a target model, and trains the target model in an unsupervised training mode to obtain the characteristics of each object. Each sequence includes a plurality of objects arranged in a walking order. In one possible example, the step of obtaining at least two sequences from the undirected graph in a random walk manner may include: for each node in the undirected graph, the computer device performs random walk with each node as a starting point based on the target walk times and the walk step size to obtain the at least two sequences. Accordingly, the step of inputting the at least two sequences into the target model and performing unsupervised training on the target model to obtain the features of each object by the computer device may include: the computer equipment acquires the initialized characterization matrixes of the at least two objects; for each sequence obtained by each wandering, the computer device traverses at least two subsequences included in each sequence by taking each node in each sequence as a central point based on the size of a target window, wherein each subsequence includes the central point and adjacent nodes located in the size range of the target window of the central point; for each of the sequences, the following steps are performed: for each subsequence of each sequence, the computer device calculates a value of a target loss function based on each subsequence and the central point of each subsequence, and optimizes the initialized characterization matrix based on the value of the target loss function until the value of the loss function reaches a target optimization condition, and then the optimization is stopped to obtain characterization matrices of at least two objects. The target optimization condition may include that the value of the loss function does not exceed the target optimization threshold, is within a target optimization threshold interval, and the like. The characterization matrix may be a node characterization matrix Φ, where Φ is a matrix of | V | × d, and each node has a d-dimensional vector, that is, an Embedding vector having a d-dimension corresponding to each object. Wherein the objective loss function can be a function characterizing the conditional probability; the target loss function is used to characterize the conditional probability of the occurrence of each subsequence if the center point of each subsequence occurs. The process of the computer device calculating the value of the objective loss function based on the each subsequence and the center point of the each subsequence may include: the computer device calculates a conditional probability of the occurrence of the subsequence under the condition that the central point occurs based on each subsequence and the central point of each subsequence, and obtains a value of the objective loss function based on the conditional probability.

Each object may be characterized by an Embedding vector of the object. Wherein the process of the computer device determining the feature similarity between the target object and each object in the second set of objects may include: the computer device calculates cosine similarity between the target object and each object in the second object set respectively based on the Embedding vector of the target object and the Embedding vector of each object in the second object set, and takes the cosine similarity as the feature similarity. Therefore, cosine similarity is calculated for the Embedding vector of each object in each layer and the Embedding vector of each object higher than the object in the upper layer, and similar objects of each object in an object set higher than the object in the layer are obtained. Alternatively, the euclidean distance between two Embedding vectors may also be calculated, and the euclidean distance may be used as the feature similarity. Accordingly, the target conditions may include, but are not limited to: the Euclidean distance is not higher than a target distance threshold, the cosine similarity is not lower than a target similarity threshold, and the like.

The random walk is mainly to randomly find the next node as root for the walk for each node in the undirected graph, and the walk length is fixed t. Each node may have several walkers that walk in parallel, which are hyper-parameters that may be configured based on need. After random walk, a plurality of sequence samples are obtained. These sequence samples can be used as training samples. The specific process of obtaining the sequence based on random walk and generating the Embedding vector by using the model is as follows:

Algorithm 1 DeepWalk(G，w，d，γ，t)

Input：graph G(V，E)

windowsizew

Embeddingsized

walkspervertexγ

walklengtht

Output：matrixofvertexrepresentationsΦ∈R^∣V∣×d

1：Initialization：SampleФfrom U^∣V∣×d

2:Build a binary Tree Tfrom V

3：for i＝0toγdo

4：O＝Shuffle(V)

5：for eachv_i∈O do

6：Wv_i＝Random Walk(G，v_i，t)

7：SkipGram(Ф，Wv_i，w)

8：endfor

9：endfor

the expression "Algorithm 1deep walk (G, w, d, γ, t)" may be implemented by adopting a deep walk Algorithm, where G is a directed graph, V denotes a node set in the directed graph, and E denotes an edge in the graph. "Input: graph G (V, E); window size w; embeddingsize d; walkspervertex γ; walklength t; "The input of the algorithm is shown as a graph G (V, E), the window size is w, the dimensionality of the Embedding vector is d, the number of random walk times of each node is gamma, and the step length of the random walk is t. "Output: matrix xoffvertexrepresentations phi epsilon R^∣V∣×d"the node characterization matrix phi representing the output is a matrix of | V | × d, each node has a vector of d dimensions, and the vector corresponds to the Embedding vector of each object. In step 1, a vector space of each node is initialized, and a U can be obtained through initialization^∣V∣×dThe initial characterization matrix of (a). And 2, establishing a Huffman (Huffman) tree, wherein the Huffman tree can be established according to the occurrence times of random walk nodes. Step 3, representing the circulation wandering from 0 to gamma times. And 4, disordering the node set V to obtain O. And step 5, traversing each node in the O, and d represents entering a loop. Step 6, obtaining_iThe node starts a random walk sequence of steps t. Step 7, training is performed using the SkipGram algorithm to optimize parameters, which may include optimization of the node characterization matrix Φ. And 8, exiting the inner-layer cycle. And 9, exiting the outer layer cycle.

The SkipGram algorithm related in the algorithm process is a training mode of word2vec in the field of natural language processing, in the deep walk algorithm, a large number of training samples are generated by random walk, and then the SkipGram is used for carrying out unsupervised training to obtain an Embegding vector of each node. The SkipGram algorithm has the following specific flow:

Algorithm 2 SkipGram(Ф，Wv_i，w)

1：for eachv_j∈Wv_ido

2：for eachu_k∈Wv_i[j-w:j+w]do

3：J(Ф)＝-logPr(u_k|Ф(v_j))

4：

5：endfor

6：endfor

wherein "Algorithm 2SkipGram (phi, Wv)_iW) "representsAlgorithm 2SkipGram algorithm, phi represents the node characterization matrix of the output, Wv_iRepresents a sequence generated by random walk and w represents a window size. Wherein, step 1, outer layer circulation, for sequence Wv_iEach node in the set performs a loop: traverse Wv_iEach node v in the sequence_iAnd entering a circulation. And 2, performing inner-layer circulation, namely operating the sequence with the window size of each node as w: u. of_k∈Wvi[j-w:j+w]Go through v_iEach node of the node front-back spacing w obtains a subsequence u_kAnd enters the cycle. Step 3, for each subsequence u_kCalculating the value of the corresponding objective loss function J (phi) at each cycle, wherein Pr (u)_k|Ф(v_j) Is expressed at phi (v)_j) Occurrence of u under the conditions of occurrence_kIs, i.e., u_kIn v_jThe probability of a context occurrence; the goal of the optimization here is to move towards u_kAt v_jThe greater the probability of occurrence under the conditions of occurrence, i.e. Pr (u)_k|Ф(v_j) Larger, i.e., J (Φ) is closer to 0. And 4, performing iterative training by adopting a gradient descent method, wherein alpha represents a learning rate. And 5, exiting the inner-layer cycle. And 6, exiting the outer layer cycle.

It should be noted that, in the above example, the deep walk algorithm is used to calculate the Embedding vector of the obtained object, and other graph representing algorithms, such as LINE (Large-scale Information Embedding) algorithm, Node2Vec (Scalable-Feature Learning for Networks) algorithm, structure 2Vec (Learning Node from structure identifier) algorithm, etc., may also be tried to perform comparative experiment analysis, and select the algorithm with the best effect.

The first object set higher than the target object level is screened out based on the target level, the subsequent property recommendation based on the property use habit of the high-level object is realized, and the more superior players are, the more proficient the game logic is mastered, and the more reasonable the selected property collocation is. For each layer of players, the method can learn the habits, such as property selection habits, property collocation and the like, of the higher-level players similar to the player, so as to learn the game skills of the higher-level players, and further improve the practicability of property recommendation.

Step 203, the computer device determines a recommendation set of the target object based on the item usage information of the first object set.

The computer device can determine a recommendation set meeting the item screening condition based on the item use information. For example, the item usage information may include a record of item usage of the object. The item screening condition at least comprises: the props used by the objects belonging to the first set of objects. Props used by subjects may include, but are not limited to: the current props used, the historical props used, etc.

In a possible implementation manner, the computer device may further filter the props of the objects in the first object set based on the existing props of the target objects, the prop transaction levels of the target objects, and the like, to obtain a recommendation set. For example, the item screening condition may further include: not exceeding the item transaction level of the target object, not including the existing items of the target object, etc. Accordingly, step 203 may be implemented by steps 2031-2033 below.

Step 2031, the computer device determines a first recommendation set based on the item usage information of each object in the first object set.

The computer device can obtain props of all objects in a first object set based on the prop use information, and determine the first recommendation set, wherein the first recommendation set comprises the props of all objects in the first object set. For example, the items commonly used by the objects in the first object set may be further screened out as the first recommendation set based on the number of times of use of the items. The item screening condition may further include: the number of uses used by each object in the first set of objects exceeds a target number threshold. The computer device may determine a prop for each object in the first set of objects based on the prop usage information; and counting the use times of props of all objects in the first object set, and screening out a first recommendation set with the use times exceeding a target time threshold value from the props of all objects in the first object set. For example, the computer device may determine, based on the item usage information, items used by the respective objects in the first set of objects within the target time period; for example, props that have been or are being used in the last month or week. Wherein the computer device can count the number of times each prop is used by at least one object in the first object set in a target period, for example, for a certain item a used by the object A in the first object set, the accumulated value of the number of times the prop a is used by the object A and other objects in the first object set in the last month can be counted; further obtaining the common props of the first object set.

In one possible example, the item screening condition may further include: the objects in the first set of objects use game items in each game for which the outcome of the game is a win. The computer device may also determine a first set of recommendations in conjunction with the game outcome for the object at each game play. For example, the computer device may further determine the first recommendation set based on the item usage information of each object in the first object set and the game result of each object in each game, wherein the first recommendation set includes items used by the objects in the first object set in the game for which the game result is a win. Of course, the first recommendation set that the number of usage times exceeds the target number threshold value in each game that wins the game result may be further screened; or screening out a first recommendation set with the use times exceeding a target time threshold in each game won in the target period, and the like, which are not described in detail herein.

Step 2032, the computer device determines the item transaction level of the target object based on the resource value of the transaction item of the target object.

The item transaction level refers to the transaction capacity of the target object to the game item; for example, the amount of resources paid out by the target object to purchase a game item. In one possible example, the computer device may evaluate a consumption level of a target object based on a resource value of the target object at least two times when the object is traded. For example, the average value, the maximum value, the median value, or the like of the game virtual resources spent by the target object to purchase the game item a plurality of times in the last month is taken as the item transaction level of the target object.

In one possible example, the computer device may further evaluate the consumption level by predicting a trading potential of the target object based on historical trading records of the target object at past times. The computer equipment can count the resource value of the target object trading the prop within the designated time, and determine the trading potential value of the target object based on the resource value of the trading prop within the designated time; and the computer equipment determines the property trading level of the target object according to the trading potential value and the resource value of the traded property in the designated time. The trading potential value is used for representing the potential ability of the target object to trade props, for example, the potential for purchasing props. For example, it may be counted that player A has purchased the game gold 1800, 2000, 2200, maximum 2200, in the last 3 consecutive months, each month, and the transaction potential value, i.e., the potential increased purchasing power 200, and the maximum consumption level of the target object in month 4 may be 2400.

Step 2033, the computer device screens out a second recommendation set from the first recommendation set based on the existing item of the target object and the item transaction level.

The second recommendation set does not include the existing prop and the resource value of the prop in the second recommendation set does not exceed the prop trading level. The computer equipment screens out a candidate set which does not comprise the existing prop from the first recommendation set based on the existing prop, and determines the prop which does not exceed the trade level of the prop in the candidate set as a second recommendation set based on the resource value of each prop in the candidate set. In one possible example, the computer device may further store an association between the second recommendation set and the target object in association, so as to subsequently recommend game items in the second recommendation set to the target object. Wherein, the existing prop means a prop which is owned and can be directly used by the target object at present; such as items that are currently purchased, gift to friends, or already exist in the item repository of the target object, etc.

And determining a recommendation set comprising the props used by the first object set based on the prop use information of the objects in the first object set, thereby finding out the props used by the players with the hierarchy higher than that of the target object for subsequent recommendation. Furthermore, the method can further screen the props which better meet the requirements of the target objects by combining with certain prop screening conditions, such as screening the props which do not exceed the consumption level of the target objects, filtering the existing props of the target objects, and the like, so that the finally obtained second recommendation set can better meet the props requirements of the target objects, and further improve the individuation and the accuracy of the prop recommendation.

Step 204, the computer device recommends the game item in the recommendation set to the target object.

The computer device can sort the props in the recommendation set and recommend the props according to a certain order. Alternatively, the computer device may recommend a number of game items in the recommendation set to the target object in accordance with a preconfigured recommendation number. In one possible implementation, the computer device may make the recommendation using the magnitude of the heat value of the prop. Step 204 may include: the computer equipment counts the heat value of at least two props in the recommendation set in the first object set; the computer equipment performs descending arrangement on the at least two props in the recommendation set based on the heat values of the at least two props; the computer device recommends the game items in the recommendation set to the target object in descending order. For example, the computer device recommends a number of game items from the recommendation set to the target object in descending order. Illustratively, the heat value of the prop is used to indicate the popularity of the prop in the first set of objects. For example, the heat value may include a number of usage objects that the prop corresponds to in the first set of objects. The recommendation set may be a second recommendation set, and the computer device may count the number of objects used for each prop in the second recommendation set, and rank each prop in the second recommendation set in a descending order based on the number of objects used. The number of the used objects can be the number of the objects using the prop in the first object set; for example, prop a is used by 10 players in the last month, and the number of used objects is 10.

In another possible implementation manner, the computer device may perform descending order arrangement on the items in the recommendation set according to the attribute values of the game items in the recommendation set, and sequentially perform recommendation according to the descending order arrangement. Illustratively, the attribute value is used to indicate a capability value that a game item plays when used in a game; for example, the range of a virtual gun in the game, the attack strength of a virtual sword, the defense strength of a virtual shield, and the like.

In one possible implementation, the computer device may make the recommendation in conjunction with a game level at which the player is playing when recommending items; alternatively, the recommendation may be made in conjunction with configuration information for the game item. In an example one, the computer device may determine a current game stage of the target object, screen out a target item corresponding to the game stage from the recommendation set, and recommend the target item to the target object. For example, when detecting that the target object is playing the game stage, the computer device may send the target item to the terminal where the target object is located, so that the terminal displays the target item in a game screen corresponding to the game stage. For example, an icon showing the target item, the number of coins spent for the game, the value of the attribute, the power of the attack, the rules of use, and the like. For example, the target prop corresponding to the game stage may include: a game item specific to the game stage, an object use number of the object use item of the game stage exceeding a target numerical value, and the like. For example, if more than 100 objects in the first set of objects use prop b when passing through game stage E, prop b may be a target prop corresponding to the game stage. In example two, the computer device may determine configuration information of an object in the first object set on a game item in the recommendation set, and recommend the game item in the recommendation set and its corresponding configuration information to the target object. For example, the computer device may send the game item and the configuration information in the recommendation set to the terminal where the target object is located, and the terminal may display the game item and the configuration information thereof. Configuration information for a play object may include, but is not limited to: the configuration of the attribute parameters of the game item and the configuration of accessories required by the game item. For example, an attack value of the attack-like virtual appliance, a traveling speed of the virtual vehicle, and the like. For example, if the target in the first set of objects is configured with a virtual shotgun having a quadruple mirror, then the virtual shotgun may be recommended to also include a quadruple mirror for the target in its corresponding configuration information.

It should be noted that the computer device may correspondingly store the association relationship between the target object and the recommendation set (or the second recommendation set), and may recommend one or more game items in the recommendation set based on the manner shown in the above implementation manner or example each time recommendation is performed. The recommendation process is illustrated by only a few implementation manners and examples, and the number of items recommended each time and the recommendation process are not specifically limited in the embodiment of the present application.

In one possible implementation, for the object with the highest hierarchy, the computer device may determine the first set of objects and the recommendation set no longer based on the

above step

202 and 203, but rather make targeted recommendations. In one example, if the target level is the highest level in the game application, the computer device may recommend the item with the highest item attribute value in the game application to the target object. That is, if the target level is the highest level, the step 202-203 can be replaced by: the computer device obtains the item with the highest attribute value in the game application to obtain the recommendation set, and recommends the game item in the recommendation set to the target object based on step 204. Of course, filtering may also be performed based on the existing props of the target object, for example, the computer device obtains at least two game props with the highest prop attribute values, and filters the existing props of the target object in the at least two game props, to obtain the recommendation set. For example, the most aggressive attack-class virtual appliance, the most recently-derived most highly configured virtual shotgun, etc. are recommended. In another example, if the target level is not the highest level in the game application, the computer device performs the process of

step

202 and 203, determines the first object set, obtains the recommendation set based on the item usage information of the objects in the first object set, and then performs item recommendation through step 204.

In a possible implementation manner, when the recommendation set is determined based on the above step 2031-2033, if the step 2033 is based, the second recommendation set obtained after the first recommendation set is screened according to the item screening condition that does not exceed the item transaction level of the target object and does not include the existing items of the target object is empty, that is, the item in the first recommendation set can be filtered based on the item screening condition, and then there is no game item that meets the condition. Step 204 may be replaced with: the computer equipment recommends a prop exchange gift bag to the target object, wherein the prop exchange gift bag can be universal and can exchange other prop gift bags. Of course, when the recommendation set is empty (or the second recommendation set is empty), the computer device may further obtain the item exchange gift package corresponding to the target level according to the target level of the target object, and recommend the item exchange gift package to the target object. For example, a sub-core player corresponds to a premium gift package, a normal player corresponds to an advanced gift package, and so on.

Currently, in the related art, for game item recommendation, manual configuration of game planning is adopted, and personalized recommendation through a machine learning algorithm is adopted.

Among them, the recommendation of manual configuration by game planning has the following technical defects: firstly, the recommended content of the game props is single, namely the content seen by all users is the same, even props purchased by some users are still recommended repeatedly, and the waste of popularization resources is formed for a long time. Secondly, because the prop list is fixed, in most cases, a user cannot quickly find a favorite game prop from the first page of the prop list, and usually turns a page many times, so that the user experience is very poor. Thirdly, because the required props are difficult to find, the purchasing desire of the user is greatly reduced, and the game revenues are reduced to a great extent. Fourth, the game list is not updated in time, the game is usually released along with the version, and the latest and best game props cannot be seen by the user quickly.

The personalized recommendation method based on the machine learning algorithm is initially applied to the field of electronic commerce, and based on the interest characteristics and purchasing behavior data of users, a complex characteristic project is built by using a model, and information and commodities which are interested by the users are recommended to the users through data mining calculation. The system is systematized to form a high-level intelligent recommendation system, called as an individual recommendation system, which provides completely individual decision support and information service for the user in the shopping process, thereby bringing better commodity purchasing experience and invisibly improving the commodity sales quantity and revenues. Similarly, the game prop is a virtual commodity and also belongs to the category of personalized recommendation. Currently, algorithms commonly used in the industry include algorithms based on a tree model, Deep Network (Deep Factorization Machine) Network, DIN (Deep Interest Network), etc.), collaborative filtering, etc. Although the existing common personalized recommendation algorithm is mature, the following disadvantages may exist in the field of games: first, more complex feature engineering is required. Secondly, the algorithm engineer needs to understand the game logic and the advantages and disadvantages of various properties deeply, a large amount of labor cost is needed for analysis, the recommended scheme is deeply coupled with specific services, and the designed scheme is difficult to multiplex and migrate on different services. Thirdly, a large amount of new props can be frequently generated in the game for selling, and the current commonly-used recommendation scheme can not well solve the problem of cold starting of the props.

Aiming at the defects of the traditional game plan manual configuration recommendation, such as single recommendation content, fixed property list, incapability of updating the recommendation result in time and the like, the game property recommendation method provided by the application can self-adaptively layer objects in the game at regular time and then recommend, and the recommendation result can be automatically and routinely updated, and aiming at the problems existing in the current common recommendation algorithm, such as the need of complex characteristic engineering, the need of an algorithm engineer to deeply understand the game logic and the difficulty of well solving the problems of cold start of the property and the like, the game property recommendation method provided by the application can self-adaptively recommend the property which is similar to the game logic and is selected most recently than other objects to the target object without the need of complex characteristic engineering and the need of deep understanding of the game logic under the condition of not deeply analyzing the game logic and the advantages and disadvantages of various properties, the cost is greatly saved. The method has strong reusability, universality and mobility, and particularly aims at the problem of prop cold start, the recommendation set is determined by directly using information based on the first object set, and the object at the highest level directly recommends the latest prop and the strongest prop, so that the use habit of the props can be learned from the object at the high level through the low level, even the latest prop can be effectively recommended with a high probability, and the problem of prop cold start can be well solved.

To better explain the game item recommendation process of the present application, the game item recommendation process shown in fig. 4 is used to further exemplify the game item recommendation method of the present application. Aiming at the defects of the prior art scheme, the game prop recommendation method provided by the application makes full use of the log data of the object in the game, applies the technologies of machine learning/deep learning and the like, enables the recommendation system to learn the object Embedding vector, and adaptively recommends the prop which is similar to and more severe than other objects and is selected most recently to the object under the condition that the game logic and the advantages and disadvantages of various props do not need to be deeply analyzed. The game property recommendation method does not need complex characteristic engineering, does not need deep understanding of game logic, has strong reusability, universality and migratability, and can solve the problem of cold start of properties. As shown in fig. 4, the game item recommendation process of the present application may include the following steps (1) to (7).

Step (1), object layering: according to the characteristics of the game level, the payment characteristic, the online time and the like of the objects, all the objects are divided into a plurality of layers, and the more core the object with higher level (the smaller the layer number) is, the higher the game skill is. The purpose of the layering is to wish to recommend props to the subject that are similar to and more frequent than those of the subject.

Step (2), calculating an object Embedding vector: according to the characteristics of the objects in the in-game team, the objects are used as node composition, random walk is conducted by means of a Deepwalk algorithm, and an Embedding vector of each node is obtained through learning.

And (3) calculating the similarity of the objects: cosine similarity is calculated for each object Embedding vector in each layer and Embedding vector of each object higher than the game level of the object in the upper layer, except for the object in the first layer with the highest level, each object in other layers can obtain topN (wherein N can be valued according to actual conditions) most similar objects.

Step (4), determining a recommended prop candidate set: in step (3), except for the first layer of objects, each object in other layers can be matched with the most similar topN (the first N, N is a positive integer) objects, and the game props most frequently used by each similar object in the topN are counted. Therefore, each object except the first layer can obtain topN recommended prop candidate sets.

Step (5), rule filtering: according to the maximum sum of the items purchased by each object in the last half year as the payment capability of the object, in the step (4), a recommended item candidate set of each object except the first layer is obtained, items which do not accord with the object payment capability are filtered from the candidate set according to item prices and the payment capability of the objects, and existing items of the objects are filtered, so that each object except the first layer can obtain the recommended item candidate set S.

Illustratively, object u is at the ith level (i)>1, obtained by layering part of users at 3.2.1), the Embedding vector is recorded as eⁱThe game level is pⁱ. Finding game level greater than p in the objects of layer 1 to layer i-1ⁱ(filtering may be performed according to other information in actual service), assuming that the set has n objects, each object in the set has a corresponding Embedding vector, and the set of the Embedding vectors is denoted as Es ═ e¹，e²，e³，...，e^j，...eⁿCalculating the Embedding vector e of the user u according to the cosine similarityⁱSimilarity to each Embedding vector in Es. Obtaining the most similar topk (k) according to the inverse sorting of the similarity<n) upper level objects. Counting the k pairsIn the elephant, each object has the most recent (the last week, the last month, or the like according to the actual business situation) most frequently used props, and k props can be obtained as a recommended candidate set of the object u.

By the above logic, each user of other layers can get the recommended candidate set S of k items except the first layer object.

Step (6), sequencing the utilization degree: in the step (5), except for the first layer of objects, each object of other layers can obtain a recommended prop candidate set S, the number of different objects of each prop in the S used in the upper layer is counted again, and the props in the S are sorted in a reverse order according to the number of users of each prop, so that a final recommended prop candidate set S' is obtained.

Through the step (5), except for the first-layer object, each of the rest objects can obtain k item recommendation candidate sets S, but if items in the recommendation candidate sets are higher in price than the payment capability of the object or the items existing in the object, if the items are recommended to the object, the recommendation is invalid because the object cannot purchase the items, the operation resources are wasted, and the game experience of the object is possibly poor. Therefore, the updated prop candidate set S' can be obtained by filtering the existing props through the payment capability.

Aiming at an object to be recommended, the maximum sum value of single payment in the last half year of the object can be counted as the payment capability of the object, props with the price higher than the payment capability in the recommended candidate set are filtered, existing props of the object are checked, existing props of the recommended candidate set are filtered, and an updated prop candidate set S' is obtained.

Except for the first-layer object, each of the remaining objects can obtain a property candidate set S ', but when online recommending is performed, each object can be recommended with m (determined according to business requirements, for example, m may be 1, 2, 5, and the like) properties, all the properties of S ' may not be recommended at one time, so it is necessary to sort the properties in S ', for example, the number of users of each property in S "from layer 1 to layer i-1 is counted, and it can be considered that the larger the number of users is, the more reasonable the recommendation result is, and the higher the recommended reliability is. And (5) inversely sequencing the props in the S according to the number of users. Through the above processing, except for the first-layer object, each of the remaining objects can obtain the sorted recommended candidate set S ″.

Step (7), recommending props for the first-layer objects: because the first-layer object is the most core, most valuable and highest-grade object in the game, the part of the object can be defaulted to be without personalized recommendation, and the strongest and latest item in the game is directly recommended.

In summary, the purpose of the layering is to distinguish game skill mastery and activity according to the object, and it can be considered that the more the object on the upper layer is, the more proficient the game logic mastery is, and the more reasonable the property matching is selected. For each layer of objects, the users hope to learn the property selection habits of the similar upper layer objects, for the most core player group at the top layer, the group has the highest mastering degree on game logic, the users can accurately select the properties with the maximum profit, and the recommendation system only needs to recommend the latest strongest properties to the group.

According to the game item recommendation method provided by the embodiment of the application, the target level of the target object is determined based on the game skill information of the target object, so that the object is divided into a plurality of levels with different game skills; determining a first object set based on the target hierarchy and the object characteristics of the target object, and determining a first object set which is higher in hierarchy than the target object and similar to the target object by combining the hierarchy of the object, the object characteristics and the like; subsequently, further determining a recommendation set comprising game items used by the first object set so as to recommend the game items in the recommendation set to the target object; the method and the device enable the low-level object to learn the use habit of the props to the high-level object, and the game skill of the high-level object is more skilled than that of the low-level object, so that the possibility that the recommended props are converted into effective props is improved, and the actual recommendation efficiency is improved. In addition, the game logic and the advantages and disadvantages of all the props do not need to be deeply analyzed, the frequently-selected props of the feature similar objects of higher levels can be recommended in a self-adaptive mode based on the levels and the feature similarity, complex feature engineering and deep understanding of the game logic are not needed, and labor cost is greatly saved. By isolating the game logic and game property recommendation process of different games and without an individualized machine learning model and the like, the method can be reused for any game to perform property recommendation, has strong reusability, universality and mobility, and can also learn the recommendation mode of using the property to a high level through a low level even if the game property is the latest one, so that the possibility that a new property is recommended and purchased is improved, the problem of cold start of the property is well solved, and the practicability of game property recommendation is further improved.

Fig. 5 is a schematic structural diagram of a game item recommendation device provided in an embodiment of the present application, and as shown in fig. 5, the game item recommendation device may include:

a level determining module 501, configured to determine a target level of a target object based on game skill information of the target object;

an object set determining module 502, configured to determine a first object set based on the target hierarchy and object features of a target object, where a hierarchy of the objects in the first object set is higher than the target hierarchy, and feature similarity between the objects in the first object set and the target object meets a target condition;

and the recommending module 503 is configured to determine a recommending set of the target object based on the item use information of the first object set, and recommend the game item in the recommending set to the target object.

In one possible implementation, the apparatus further includes:

a characteristic determination module for determining a characteristic of each object based on team information of at least two objects in the game application;

the object set determining module 502 is configured to determine a second object set based on the target hierarchy, where a hierarchy of objects in the second object set is higher than the target hierarchy; determining feature similarity between the target object and each object in the second object set respectively based on features of the target object and the objects in the second object set; and screening the first object set from the second object set based on the feature similarity between the target object and each object in the second object set respectively.

In one possible implementation, the feature determination module includes:

the construction unit is used for constructing an undirected graph based on the formation information of the at least two objects, wherein the undirected graph comprises at least two nodes corresponding to the at least two objects, and an edge exists between the two nodes corresponding to the two objects which are formed into a formation with each other in the at least two objects;

a sequence obtaining unit, configured to obtain at least two sequences from the undirected graph in a random walk manner, where each sequence includes nodes arranged in a walk order in a walk process;

and the characteristic acquisition unit is used for inputting the at least two sequences into the target model and carrying out unsupervised training on the target model to obtain the characteristic of each object.

In a possible implementation manner, the sequence obtaining unit is configured to perform random walk on each node in the undirected graph based on a target walk number and a walk step length, with the each node as a starting point, to obtain the at least two sequences;

correspondingly, the feature obtaining unit is configured to obtain the initialized characterization matrices of the at least two objects; for each sequence obtained by each wandering, based on the size of a target window, traversing at least two subsequences included in each sequence by taking each node in each sequence as a central point, wherein each subsequence includes the central point and adjacent nodes located in the size range of the target window of the central point; for each of the sequences, the following steps are performed: and for each subsequence of each sequence, calculating a numerical value of a target loss function based on each subsequence and the central point of each subsequence, and optimizing the initialized characterization matrix based on the numerical value of the target loss function until the numerical value of the target loss function reaches a target optimization condition, and stopping optimization to obtain the characterization matrices of at least two objects.

In one possible implementation, the game skill information is information representing a proficiency level in mastering and applying game logic; the higher the proficiency level of the target object in mastering and applying game logic, the higher the target level of the target object;

the game skill information comprises at least one of game grade, resource value of trade prop, object online time length, object registration time and activity participation frequency;

accordingly, the hierarchy determination module 501 is configured to at least one of:

when the game level of the target object is in the target level interval, determining that the object level of the target object is a target level corresponding to the target level interval, wherein the game skill information comprises the game level;

when the resource value of the transaction prop responding to the target object is positioned in the target resource interval, determining the object level of the target object as a target level corresponding to the target resource interval;

when the online object time length of the target object is within the target time length interval, determining the object hierarchy of the target object as a target hierarchy corresponding to the target time length interval;

when the object registration time of the target object is in the target time interval, determining the object hierarchy of the target object as a target hierarchy corresponding to the target time interval;

and when the activity participation frequency of the target object is positioned in the target frequency interval, determining the object level of the target object as the target level corresponding to the target frequency interval.

In a possible implementation manner, the recommending module 503 is configured to determine a first recommending set based on the item usage information of each object in the first set of objects; determining the trading level of the target object based on the resource value of the trading prop of the target object; and screening a second recommendation set from the first recommendation set based on the existing prop of the target object and the prop trading level, wherein the second recommendation set does not comprise the existing prop and the resource value of the prop in the second recommendation set does not exceed the prop trading level.

In a possible implementation manner, the recommending module 503 is configured to count heat values of at least two items in the recommended set in the first set of objects; based on the heat values of the at least two props, performing descending arrangement on the at least two props in the recommendation set; and recommending the game items in the recommendation set to the target object according to the descending order.

In one possible implementation, the recommending module 503 includes at least one of:

determining a current game stage of the target object, screening a target prop corresponding to the game stage from the recommendation set, and recommending the target prop to the target object;

determining the configuration information of the objects in the first object set to the game items in the recommendation set, and recommending the game items in the recommendation set and the corresponding configuration information to the target object.

In a possible implementation manner, the recommending module 503 is further configured to recommend the item with the highest item attribute value in the game application to the target object if the target hierarchy is the highest hierarchy in the game application.

According to the game item recommendation device provided by the embodiment of the application, the target level of the target object is determined based on the game skill information of the target object, so that the object is divided into a plurality of levels with different game skills; determining a first object set based on the target hierarchy and the object characteristics of the target object, and determining a first object set which is higher in hierarchy than the target object and similar to the target object by combining the hierarchy of the object, the object characteristics and the like; subsequently, further determining a recommendation set comprising game items used by the first object set so as to recommend the game items in the recommendation set to the target object; the method and the device enable the low-level object to learn the use habit of the props to the high-level object, and the game skill of the high-level object is more skilled than that of the low-level object, so that the possibility that the recommended props are converted into effective props is improved, and the actual recommendation efficiency is improved. In addition, the game logic and the advantages and disadvantages of all the props do not need to be deeply analyzed, the frequently-selected props of the feature similar objects of higher levels can be recommended in a self-adaptive mode based on the levels and the feature similarity, complex feature engineering and deep understanding of the game logic are not needed, and labor cost is greatly saved. Through separating the game logic of different games, game props and game props recommendation process, and need not individualized machine learning model etc. for the method of this application can multiplex to any game and carry out the props recommendation, has stronger reusability, commonality, mobility, even be the newest game props, also can be through the low level to the high level study recommendation mode of using the props, improved the new way utensil recommended, the possibility of being purchased, fine solution props cold start problem, and then improved the practicality of game props recommendation.

The game item recommendation device of this embodiment can execute the game item recommendation method of the above embodiments of this application, and the implementation principles are similar, and are not described herein again.

Fig. 6 is a schematic structural diagram of a computer device provided in an embodiment of the present application. As shown in fig. 6, the computer apparatus includes: the processor executes the computer program to realize the steps of the game item recommendation method, and compared with the related art, the method can realize the following steps:

dividing an object into a plurality of levels having different game skills by determining a target level of the target object based on game skill information of the target object; determining a first object set based on the target hierarchy and the object characteristics of the target object, and determining a first object set which is higher in hierarchy than the target object and similar to the target object by combining the hierarchy of the object, the object characteristics and the like; subsequently, further determining a recommendation set comprising game items used by the first object set so as to recommend the game items in the recommendation set to the target object; the method and the device enable the low-level object to learn the use habit of the props to the high-level object, and the game skill of the high-level object is more skilled than that of the low-level object, so that the possibility that the recommended props are converted into effective props is improved, and the actual recommendation efficiency is improved. In addition, the game logic and the advantages and disadvantages of all the props do not need to be deeply analyzed, the frequently-selected props of the feature similar objects of higher levels can be recommended in a self-adaptive mode based on the levels and the feature similarity, complex feature engineering and deep understanding of the game logic are not needed, and labor cost is greatly saved. By isolating the game logic and game property recommendation process of different games and without an individualized machine learning model and the like, the method can be reused for any game to perform property recommendation, has strong reusability, universality and mobility, and can also learn the recommendation mode of using the property to a high level through a low level even if the game property is the latest one, so that the possibility that a new property is recommended and purchased is improved, the problem of cold start of the property is well solved, and the practicability of game property recommendation is further improved. In an alternative embodiment, a computer device is provided, as shown in FIG. 6, the computer device 600 shown in FIG. 6 comprising: a processor 601 and a memory 603. The processor 601 is coupled to the memory 603, such as via a bus 602. Optionally, the computer device 600 may further include a transceiver 604, and the transceiver 604 may be used for data interaction between the computer device and other computer devices, such as data transmission and/or data reception. It should be noted that the transceiver 604 is not limited to one in practical applications, and the structure of the computer device 600 is not limited to the embodiment of the present application.

The Processor 601 may be a CPU (Central Processing Unit), a general-purpose Processor, a DSP (Digital Signal Processor), an ASIC (Application Specific Integrated Circuit), an FPGA (Field Programmable Gate Array) or other Programmable logic device, a transistor logic device, a hardware component, or any combination thereof. Which may implement or perform the various illustrative logical blocks, modules, and circuits described in connection with the disclosure. The processor 601 may also be a combination of computing functions, e.g., comprising one or more microprocessors, DSPs and microprocessors, and the like.

Bus 602 may include a path that transfers information between the above components. The bus 602 may be a PCI (Peripheral Component Interconnect) bus, an EISA (Extended Industry Standard Architecture) bus, or the like. The bus 602 may be divided into an address bus, a data bus, a control bus, and the like. For ease of illustration, only one thick line is shown in FIG. 6, but this is not intended to represent only one bus or type of bus.

The Memory 603 may be a ROM (Read Only Memory) or other types of static storage devices that can store static information and instructions, a RAM (Random Access Memory) or other types of dynamic storage devices that can store information and instructions, an EEPROM (Electrically Erasable Programmable Read Only Memory), a CD-ROM (Compact Disc Read Only Memory) or other optical Disc storage, optical Disc storage (including Compact Disc, laser Disc, optical Disc, digital versatile Disc, blu-ray Disc, etc.), a magnetic Disc storage medium \ other magnetic storage device, or any other medium that can be used to carry or store a computer program and that can be Read by a computer, without limitation.

The memory 603 is used for storing computer programs for executing the embodiments of the present application, and is controlled by the processor 601 for execution. The processor 601 is adapted to execute a computer program stored in the memory 603 for implementing the steps shown in the foregoing method embodiments.

Among these, computer devices include, but are not limited to: servers, service clusters, cloud computing centers, and the like.

Embodiments of the present application provide a computer-readable storage medium, on which a computer program is stored, and when being executed by a processor, the computer program may implement the steps and corresponding contents of the foregoing method embodiments.

Embodiments of the present application further provide a computer program product, which includes a computer program, and when the computer program is executed by a processor, the steps and corresponding contents of the foregoing method embodiments may be implemented.

The terms "first," "second," "third," "fourth," "1," "2," and the like in the description and in the claims of the present application and in the above-described drawings (if any) are used for distinguishing between similar elements and not necessarily for describing a particular sequential or chronological order. It should be understood that the data so used are interchangeable under appropriate circumstances such that the embodiments of the application described herein are capable of operation in other sequences than illustrated or otherwise described herein.

It should be understood that, although each operation step is indicated by an arrow in the flowchart of the embodiment of the present application, the implementation order of the steps is not limited to the order indicated by the arrow. In some implementation scenarios of the embodiments of the present application, the implementation steps in the flowcharts may be performed in other sequences as desired, unless explicitly stated otherwise herein. In addition, some or all of the steps in each flowchart may include multiple sub-steps or multiple stages based on an actual implementation scenario. Some or all of these sub-steps or stages may be performed at the same time, or each of these sub-steps or stages may be performed at different times, respectively. In a scenario where execution times are different, an execution sequence of the sub-steps or the phases may be flexibly configured according to requirements, which is not limited in the embodiment of the present application.

The foregoing is only an optional implementation manner of a part of implementation scenarios in this application, and it should be noted that, for those skilled in the art, other similar implementation means based on the technical idea of this application are also within the protection scope of the embodiments of this application without departing from the technical idea of this application.

Claims

1. A game item recommendation method, the method comprising:

2. The method of claim 1, wherein prior to determining the first set of objects based on the target hierarchy and object characteristics of target objects, the method further comprises:

determining a characteristic of each object based on team information of at least two objects in the game application;

the determining a first set of objects based on the target hierarchy and object features of target objects comprises:

determining a second set of objects based on the target hierarchy, the hierarchy of objects in the second set of objects being higher than the target hierarchy;

determining feature similarity between the target object and each object in the second object set respectively based on features of the target object and the objects in the second object set;

and screening the first object set from the second object set based on the feature similarity between the target object and each object in the second object set.

3. The method of claim 2, wherein determining the characteristic of each object based on team information for at least two objects in the game application comprises:

constructing an undirected graph based on the formation information of the at least two objects, wherein the undirected graph comprises at least two nodes corresponding to the at least two objects, and an edge exists between the two nodes corresponding to the two objects which are formed into a formation with each other in the at least two objects;

acquiring at least two sequences from the undirected graph in a random walk mode, wherein each sequence comprises nodes arranged according to a walk sequence in a walk process;

and inputting the at least two sequences into a target model, and carrying out unsupervised training on the target model to obtain the characteristics of each object.

4. The method according to claim 3, wherein the obtaining at least two sequences from the undirected graph by using a random walk manner comprises:

for each node in the undirected graph, performing random walk by taking each node as a starting point based on target walk times and walk step length to obtain at least two sequences;

correspondingly, the inputting the at least two sequences into the target model and performing unsupervised training on the target model to obtain the features of each object includes:

acquiring an initialized characterization matrix of the at least two objects;

for each sequence obtained by each wandering, based on the size of a target window, traversing at least two subsequences included in each sequence by taking each node in each sequence as a central point, wherein each subsequence includes the central point and adjacent nodes located in the size range of the target window of the central point;

for each of said sequences, performing the steps of:

and for each subsequence of each sequence, calculating a numerical value of a target loss function based on each subsequence and the central point of each subsequence, and optimizing the initialized characterization matrix based on the numerical value of the target loss function until the numerical value of the target loss function reaches a target optimization condition, and stopping optimization to obtain the characterization matrices of at least two objects.

5. The method according to claim 1, wherein the game skill information is information representing proficiency in mastering and applying game logic; the higher the proficiency level of the target object in mastering and applying game logic is, the higher the target level of the target object is;

the game skill information comprises at least one of game grade, resource value of trade prop, object online time, object registration time and activity participation frequency;

correspondingly, the determining the target level of the target object based on the game skill information of the target object comprises at least one of the following steps:

when the game level of the target object is located in a target level interval, determining that the object level of the target object is a target level corresponding to the target level interval, wherein the game skill information comprises the game level;

when the resource value of the transaction prop responding to the target object is located in a target resource interval, determining the object level of the target object as a target level corresponding to the target resource interval;

when the online object time length of the target object is within a target time length interval, determining the object hierarchy of the target object as a target hierarchy corresponding to the target time length interval;

when the object registration time of the target object is located in a target time interval, determining the object hierarchy of the target object as a target hierarchy corresponding to the target time interval;

and when responding to the activity participation times of the target object in a target time interval, determining the object level of the target object as a target level corresponding to the target time interval.

6. The method of claim 1, wherein determining the recommended set of target objects based on the item usage information for the first set of objects comprises:

determining a first recommendation set based on the item use information of each object in the first object set;

determining the trading level of the object based on the resource value of the trading prop of the object;

and screening out a second recommendation set from the first recommendation set based on the existing prop of the target object and the prop trading level, wherein the second recommendation set does not comprise the existing prop and the resource value of the prop in the second recommendation set does not exceed the prop trading level.

7. The method of claim 1, wherein said recommending game items in the recommendation set to the target object comprises:

counting the heat value of at least two props in the recommendation set in the first object set;

based on the heat values of the at least two props, performing descending arrangement on the at least two props in the recommendation set;

and recommending the game items in the recommendation set to the target object according to the descending order.

8. The method of claim 1, wherein the recommending game items in the recommended set to the target object comprises at least one of:

9. The method of claim 1, further comprising:

and if the target level is the highest level in the game application, recommending the item with the highest item attribute value in the game application to the target object.

10. A game item recommendation device, the device comprising:

11. A computer device comprising a memory, a processor and a computer program stored on the memory, wherein the processor executes the computer program to implement the game item recommendation method of any one of claims 1 to 9.

12. A computer-readable storage medium, on which a computer program is stored, which, when being executed by a processor, carries out a game item recommendation method according to any one of claims 1 to 9.

13. A computer program product comprising a computer program, wherein the computer program, when executed by a processor, implements the game item recommendation method of any one of claims 1 to 9.