CN110807150A

CN110807150A - Information processing method and device, electronic equipment and computer readable storage medium

Info

Publication number: CN110807150A
Application number: CN201910973476.6A
Authority: CN
Inventors: 刘阳; 马文晔
Original assignee: Tencent Technology Shenzhen Co Ltd
Current assignee: Tencent Technology Shenzhen Co Ltd
Priority date: 2019-10-14
Filing date: 2019-10-14
Publication date: 2020-02-18

Abstract

The embodiment of the invention provides an information processing method and device, electronic equipment and a computer readable storage medium, and belongs to the technical field of computers and communication. The method comprises the following steps: determining a first role of a target object; obtaining first object data of the target object according to the first role attribute information of the first role; acquiring article attribute information of an object to be recommended; and processing the first object data and the item attribute information through a neural network model, and determining a first target item recommended to a first role of the target object from the object to be recommended. The technical scheme of the embodiment of the invention provides an information processing method, which can improve the accuracy of personalized recommendation of an object to be recommended.

Description

Information processing method and device, electronic equipment and computer readable storage medium

Technical Field

The present invention relates to the field of computer and communication technologies, and in particular, to an information processing method and apparatus, an electronic device, and a computer-readable storage medium.

Background

In recent years, the development of web games has been rapid, and the web games have become an important entertainment activity. The game can not only be participated by the player, but also enhance the interest of the game, improve the user experience, enhance the stickiness of the user and increase the income of the game platform by selling virtual commodities (such as props).

The virtual goods in the game are various, meanwhile, the game has higher real-time requirement, and the user has the requirement of quickly finding the target prop wanted by the user, so that whether the game platform can provide an efficient and accurate information screening system for the user is extremely important.

It is to be noted that the information disclosed in the above background section is only for enhancement of understanding of the background of the present invention and therefore may include information that does not constitute prior art known to a person of ordinary skill in the art.

Disclosure of Invention

The embodiment of the invention provides an information processing method and device, electronic equipment and a computer readable storage medium, which can improve the accuracy of personalized recommendation of an object to be recommended.

Additional features and advantages of the invention will be set forth in the detailed description which follows, or may be learned by practice of the invention.

The embodiment of the invention provides an information processing method, which comprises the following steps: determining a first role of a target object; obtaining first object data of the target object according to the first role attribute information of the first role; acquiring article attribute information of an object to be recommended; and processing the first object data and the item attribute information through a neural network model, and determining a first target item recommended to a first role of the target object from the object to be recommended.

An embodiment of the present invention provides an information processing apparatus, including: a first object role determination module configured to determine a first role of a target object; a first object data obtaining module configured to obtain first object data of the target object according to first role attribute information of the first role; the article attribute information acquisition module is configured to acquire article attribute information of an object to be recommended; and the first target item determination module is configured to process the first object data and the item attribute information through a neural network model, and determine a first target item recommended to a first role of the target object from the object to be recommended.

An embodiment of the present invention provides a computer-readable storage medium on which a computer program is stored, which, when executed by a processor, implements the information processing method described in the above embodiment.

An embodiment of the present invention provides an electronic device, including: one or more processors; a storage device configured to store one or more programs that, when executed by the one or more processors, cause the one or more processors to implement the information processing method as described in the above embodiments.

In the technical solution provided by some embodiments of the present invention, a first role of a target object is determined, and according to first role attribute information of the first role, first object data of the target object is obtained, and item attribute information of an object to be recommended is also obtained, so that the first object data and the item attribute information may be processed by a neural network model, and a first target item recommended to the first role of the target object is determined from the object to be recommended. On one hand, the target object in the object to be recommended is determined by comprehensively considering the rich role attribute information of the target object and the rich article attribute information of the object to be recommended, so that the accuracy of personalized recommendation of the article can be improved; on the other hand, the complex interaction between the target object and the object to be recommended is described through the nonlinear neural network model, so that the purposes of enhancing model interpretability and improving recommendation effect can be achieved.

It is to be understood that both the foregoing general description and the following detailed description are exemplary and explanatory only and are not restrictive of the invention, as claimed.

Drawings

The accompanying drawings, which are incorporated in and constitute a part of this specification, illustrate embodiments consistent with the invention and together with the description, serve to explain the principles of the invention. It is obvious that the drawings in the following description are only some embodiments of the invention, and that for a person skilled in the art, other drawings can be derived from them without inventive effort. In the drawings:

fig. 1 is a schematic diagram showing an exemplary system architecture to which an information processing method or an information processing apparatus of an embodiment of the present invention can be applied;

FIG. 2 illustrates a schematic structural diagram of a computer system suitable for use with the electronic device to implement an embodiment of the invention;

FIG. 3 schematically shows a flow diagram of an information processing method according to an embodiment of the invention;

FIG. 4 schematically shows a flow chart of an information processing method according to another embodiment of the present invention;

FIG. 5 schematically shows a flow chart of an information processing method according to a further embodiment of the invention;

FIG. 6 is a diagram illustrating a processing procedure of step S320 shown in FIG. 3 in one embodiment;

FIG. 7 is a diagram illustrating a processing procedure of step S340 illustrated in FIG. 3 in one embodiment;

FIG. 8 is a diagram illustrating a processing procedure of step S341 shown in FIG. 7 in one embodiment;

FIG. 9 schematically shows a schematic diagram of a first embedding submodel according to an embodiment of the invention;

FIG. 10 is a diagram illustrating a processing procedure of step S342 shown in FIG. 7 in one embodiment;

FIG. 11 schematically illustrates a schematic diagram of a second embedding submodel according to an embodiment of the invention;

FIG. 12 schematically shows a structural diagram of a neural network model according to an embodiment of the present invention;

FIG. 13 is a diagram illustrating a processing procedure of step S343 shown in FIG. 7 in an embodiment;

FIG. 14 schematically shows a structural schematic of a first neural network submodel according to an embodiment of the present invention;

FIG. 15 is a diagram illustrating a processing procedure of step S344 shown in FIG. 7 in one embodiment;

FIG. 16 schematically shows a structural schematic of a second neural network submodel according to an embodiment of the present invention;

FIG. 17 is a diagram illustrating the processing of step S345 shown in FIG. 7 in one embodiment;

FIG. 18 schematically illustrates an interface diagram of a portal introduction in accordance with an embodiment of the invention;

FIGS. 19 and 20 schematically illustrate interface diagrams for customized character detail in accordance with an embodiment of the present invention;

FIG. 21 is a schematic diagram illustrating an interface for character creation completion in accordance with an embodiment of the present invention;

FIG. 22 schematically illustrates a game interface diagram according to an embodiment of the invention;

FIG. 23 is a schematic diagram illustrating an interface of role attributes in accordance with an embodiment of the present invention;

FIG. 24 schematically illustrates an interface diagram of a gaming mall, according to an embodiment of the present invention;

fig. 25 schematically shows a block diagram of an information processing apparatus according to an embodiment of the present invention.

Detailed Description

Example embodiments will now be described more fully with reference to the accompanying drawings. Example embodiments may, however, be embodied in many different forms and should not be construed as limited to the examples set forth herein; rather, these embodiments are provided so that this disclosure will be thorough and complete, and will fully convey the concept of example embodiments to those skilled in the art.

Furthermore, the described features, structures, or characteristics may be combined in any suitable manner in one or more embodiments. In the following description, numerous specific details are provided to provide a thorough understanding of embodiments of the invention. One skilled in the relevant art will recognize, however, that the invention may be practiced without one or more of the specific details, or with other methods, components, devices, steps, and so forth. In other instances, well-known methods, devices, implementations or operations have not been shown or described in detail to avoid obscuring aspects of the invention.

The block diagrams shown in the figures are functional entities only and do not necessarily correspond to physically separate entities. I.e. these functional entities may be implemented in the form of software, or in one or more hardware modules or integrated circuits, or in different networks and/or processor means and/or microcontroller means.

The flow charts shown in the drawings are merely illustrative and do not necessarily include all of the contents and operations/steps, nor do they necessarily have to be performed in the order described. For example, some operations/steps may be decomposed, and some operations/steps may be combined or partially combined, so that the actual execution sequence may be changed according to the actual situation.

Fig. 1 shows a schematic diagram of an exemplary system architecture 100 to which an information processing method or an information processing apparatus of an embodiment of the present invention can be applied.

As shown in fig. 1, the system architecture 100 may include one or more of

terminal devices

101, 102, 103, a network 104, and a server 105. The network 104 serves as a medium for providing communication links between the

terminal devices

101, 102, 103 and the server 105. Network 104 may include various connection types, such as wired, wireless communication links, or fiber optic cables, to name a few.

It should be understood that the number of terminal devices, networks, and servers in fig. 1 is merely illustrative. There may be any number of terminal devices, networks, and servers, as desired for implementation. For example, server 105 may be a server cluster comprised of multiple servers, or the like.

The user may use the

terminal devices

101, 102, 103 to interact with the server 105 via the network 104 to receive or send messages or the like. The

terminal devices

101, 102, 103 may be various electronic devices having a display screen and supporting web browsing, including but not limited to smart phones, tablets, laptop portable computers, desktop computers, wearable devices, smart home devices, and the like.

The server 105 may be a server that provides various services. For example, the user opens a game client using the terminal device 103 (or the terminal device 101 or 102), determines his or her game character in the game on the game client, and sends a request to the server 105. The server 105 may obtain the role attribute information of the corresponding game role based on the game role information carried in the request, thereby obtaining the object data of the user; meanwhile, the server 105 may also obtain the item attribute information of the to-be-recommended items in the game mall, process the object data of the user and the item attribute information of each to-be-recommended item through the neural network model, determine the target item from the to-be-recommended items, and return the target item to the terminal device 103, so that the user can view the target item recommended to the currently selected game role on the terminal device 103.

Also for example, the terminal device 103 (which may also be the terminal device 101 or 102) may be a smart tv, a VR (virtual Reality)/AR (Augmented Reality) helmet display, or a mobile terminal such as a smart phone, a tablet computer, etc. on which an instant messaging, a navigation, a video Application (APP) and the like are installed, and the user may send various requests to the server 105 through the smart tv, the VR/AR helmet display or the instant messaging, the video APP. The server 105 may obtain, based on the request, feedback information in response to the request, and return the feedback information to the smart television, the VR/AR head mounted display, or the instant messaging and video APP, and then display the returned feedback information through the smart television, the VR/AR head mounted display, or the instant messaging and video APP.

FIG. 2 illustrates a schematic structural diagram of a computer system suitable for use with the electronic device to implement an embodiment of the invention.

It should be noted that the computer system 200 of the electronic device shown in fig. 2 is only an example, and should not bring any limitation to the functions and the scope of the application of the embodiment of the present invention.

As shown in fig. 2, the computer system 200 includes a Central Processing Unit (CPU)201 that can perform various appropriate actions and processes in accordance with a program stored in a Read-Only Memory (ROM) 202 or a program loaded from a storage section 208 into a Random Access Memory (RAM) 203. In the RAM 203, various programs and data necessary for system operation are also stored. The CPU201, ROM 202, and RAM 203 are connected to each other via a bus 204. An input/output (I/O) interface 205 is also connected to bus 204.

The following components are connected to the I/O interface 205: an input portion 206 including a keyboard, a mouse, and the like; an output section 207 including a Cathode Ray Tube (CRT), a Liquid Crystal Display (LCD), and the like, a speaker, and the like; a storage section 208 including a hard disk and the like; and a communication section 209 including a Network interface card such as a LAN (Local Area Network) card, a modem, or the like. The communication section 209 performs communication processing via a network such as the internet. A drive 210 is also connected to the I/O interface 205 as needed. A removable medium 211, such as a magnetic disk, an optical disk, a magneto-optical disk, a semiconductor memory, or the like, is mounted on the drive 210 as necessary, so that a computer program read out therefrom is installed into the storage section 208 as necessary.

In particular, according to an embodiment of the present invention, the processes described below with reference to the flowcharts may be implemented as computer software programs. For example, embodiments of the invention include a computer program product comprising a computer program embodied on a computer-readable storage medium, the computer program comprising program code for performing the method illustrated by the flow chart. In such an embodiment, the computer program may be downloaded and installed from a network through the communication section 209 and/or installed from the removable medium 211. The computer program, when executed by a Central Processing Unit (CPU)201, performs various functions defined in the methods and/or apparatus of the present application.

It should be noted that the computer readable storage medium shown in the present invention can be a computer readable signal medium or a computer readable storage medium or any combination of the two. A computer readable storage medium may be, for example, but not limited to, an electronic, magnetic, optical, electromagnetic, infrared, or semiconductor system, apparatus, or device, or any combination of the foregoing. More specific examples of the computer readable storage medium may include, but are not limited to: an electrical connection having one or more wires, a portable computer diskette, a hard disk, a Random Access Memory (RAM), a Read-Only Memory (ROM), an Erasable Programmable Read-Only Memory (EPROM) or flash Memory), an optical fiber, a portable compact disc Read-Only Memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination of the foregoing. In the present invention, a computer readable storage medium may be any tangible medium that can contain, or store a program for use by or in connection with an instruction execution system, apparatus, or device. In the present invention, however, a computer readable signal medium may include a propagated data signal with computer readable program code embodied therein, for example, in baseband or as part of a carrier wave. Such a propagated data signal may take many forms, including, but not limited to, electro-magnetic, optical, or any suitable combination thereof. A computer readable signal medium may also be any computer readable storage medium that is not a computer readable storage medium and that can communicate, propagate, or transport a program for use by or in connection with an instruction execution system, apparatus, or device. Program code embodied on a computer readable storage medium may be transmitted using any appropriate medium, including but not limited to: wireless, wire, fiber optic cable, RF (radio frequency), etc., or any suitable combination of the foregoing.

The flowchart and block diagrams in the figures illustrate the architecture, functionality, and operation of possible implementations of methods, apparatus, and computer program products according to various embodiments of the present invention. In this regard, each block in the flowchart or block diagrams may represent a module, segment, or portion of code, which comprises one or more executable instructions for implementing the specified logical function(s). It should also be noted that, in some alternative implementations, the functions noted in the block may occur out of the order noted in the figures. For example, two blocks shown in succession may, in fact, be executed substantially concurrently, or the blocks may sometimes be executed in the reverse order, depending upon the functionality involved. It will also be noted that each block of the block diagrams or flowchart illustration, and combinations of blocks in the block diagrams or flowchart illustration, can be implemented by special purpose hardware-based systems which perform the specified functions or acts, or combinations of special purpose hardware and computer instructions.

The modules and/or units and/or sub-units described in the embodiments of the present invention may be implemented by software, or may be implemented by hardware, and the described modules and/or units and/or sub-units may also be disposed in a processor. Wherein the names of such modules and/or units and/or sub-units in some cases do not constitute a limitation on the modules and/or units and/or sub-units themselves.

As another aspect, the present application also provides a computer-readable storage medium, which may be included in the electronic device described in the above embodiments; or may exist separately without being assembled into the electronic device. The computer-readable storage medium carries one or more programs which, when executed by an electronic device, cause the electronic device to implement the method as described in the embodiments below. For example, the electronic device may implement the steps shown in fig. 3, 4, 5, 6, 7, 8, 10, 13, 15, or 17.

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 and the like.

Machine Learning (ML) is a multi-domain cross discipline, and relates to a plurality of disciplines such as probability theory, statistics, approximation theory, convex analysis, algorithm complexity theory and the like. The special research on how a computer simulates or realizes the learning behavior of human beings so as to acquire new knowledge or skills and reorganize the existing knowledge structure to continuously improve the performance of the computer. Machine learning is the core of artificial intelligence, is the fundamental approach for computers to have intelligence, and is applied to all fields of artificial intelligence. Machine learning and deep learning generally include techniques such as artificial neural networks, belief networks, reinforcement learning, transfer learning, inductive learning, and formal education learning.

With the research and progress of artificial intelligence technology, the artificial intelligence technology is developed and applied in a plurality of fields, such as common smart homes, smart wearable devices, virtual assistants, smart speakers, smart marketing, unmanned driving, automatic driving, unmanned aerial vehicles, robots, smart medical care, smart customer service, and the like.

The scheme provided by the embodiment of the invention relates to the technologies such as machine learning of artificial intelligence and the like, and is specifically explained by the following embodiments:

some terms referred to in the following embodiments are explained first.

Deep learning: the deep learning is a learning process based on a deep neural network, and a one-to-one mapping function between input data and target data is finally obtained through an optimization mode such as gradient decrement. For example, given information on age, sex, etc. of a person, it is desirable to predict whether the person likes a pet by using the information, the input data is information on age, sex, etc. of the person, and the target data is "whether the person likes a pet". It is desirable to construct a proper mapping function that maps the input data to the target data. Thus, given the information of another person, it is possible to know whether the person likes a pet. Deep learning is the process of solving this mapping function based on a large amount of data and modeling this function with a deep neural network.

Collaborative Filtering (CF): collaborative filtering is a type of model algorithm for recommendation systems, and the content core of the collaborative filtering is to deduce which users (or commodities) are more similar through past data, so the type of model can be simply divided into "user-based" and "item-based", the former finds which users are more similar, and the latter finds which commodities are more similar. In the former case, after the user who is similar to the user is known, the product purchased by the user before can be recommended to other users who are similar to the user. The latter means that after knowing the similar goods, a user can purchase the goods and deduce that the user likes other goods similar to the goods, so as to recommend other similar goods to the user.

And (3) measurement: a custom, 4-way mapping function that measures the distance between any 2 points in the domain space needs to meet the 4 requirements that the metric satisfies. Simply, defining a new metric is defining a new distance.

In the embodiment of the present invention, a game scene is taken as an example for illustration, but it is understood that the technical solution provided in the embodiment of the present invention is not limited to be applied to a game scene, and may also be applied to other recommended scenes.

The data in the game recommendation scene has the following characteristics:

first, the data is large. The game platform provides services for millions to millions of users, attributes and behaviors of the users can be dynamically changed in the process of playing games, and the users can selectively purchase different game commodities along with the updating of time, so that the game platform accumulates large-scale user behavior data every day.

Second, the data is high dimensional. The data in the game has the characteristic of high dimensionality, attributes associated with the user include personal images of the user (such as real attribute information of the user, such as real age, sex and geographical position of the user), attributes of the game role selected by the user (such as attribute information of name, sex, occupation and the like of the game role), and the like, and meanwhile, the game commodities have various categories and also have multidimensional attributes.

Third, complex structural associations. In contrast to e-commerce, social, etc., platforms, there is not a simple one-to-one correspondence between users and their purchased items in a game platform, e.g., a user may use different game characters at different times, e.g., a user may frequently purchase game item a when using game character a, and may not purchase game item a when using game character B. I.e. in the case of different game characters, the selections made by the same user may be different.

In view of the above, when designing a recommendation method in a game scene, it is necessary to consider the data size and data characteristics in the scene.

Collaborative filtering is a type of recommendation algorithm in the related art, which makes recommendations based on the following two starting points by using collaborative information in a large amount of data: (1) users with similar interests may be interested in the same thing; (2) the user may prefer something similar to what he has purchased.

However, in the conventional recommendation method based on collaborative filtering, when the vector representations of the user and the commodity are learned, only the ID (identification) and the score matrix of the user and the commodity are modeled, and rich background information is ignored or underutilized, that is, in the collaborative filtering, when a certain user in the score matrix purchases a certain commodity, the corresponding score value is 1, otherwise, the corresponding score value is 0, but the attribute information of each dimension of the user and the commodity is not considered.

CTR (Click-through rate prediction) is used not only for Click rate estimation but also in scenes such as commodity ranking of a recommendation system. Compared with the CF model, the CTR directly calculates the probability that the target product is clicked by the target user by inputting the data features, and the general formula can be expressed as follows:

y＝f(X) (1)

wherein X is the input data feature matrix, and y is the probability value between [0,1], which represents the probability of the input data being clicked. For example, Logistic Regression (LR) and Factorization (FM) can be used in the related art. The calculation mode of the LR model is as follows:

where θ is a learnable parameter. Features of each dimension in the LR model are independent of each other, and the combination between features tends to contribute to the improvement of the prediction effect. Based on this, FM considers the intersection between features, and in the case of considering two feature intersections, the FM model formula is as follows:

wherein n represents the dimension of the input data X, n is a positive integer greater than or equal to 1, X_iAnd x_jRespectively represent the value of the ith bit and the jth bit in X, and the value thereof can be 0 or 1, for example. Theta and w_ijThe parameters are learnable parameters obtained in the model training process.

Although the feature intersection can improve the prediction performance, when the purchase probability of a user to a commodity is calculated, simple vector inner product operation is adopted, the traditional machine learning method can only model the linear relation among the features, the model capability has certain limitation, the prediction score value is directly calculated according to the features, and the cooperative information of the user or the commodity is not utilized.

Fig. 3 schematically shows a flow chart of an information processing method according to an embodiment of the present invention. The method provided by the embodiment of the present invention can be executed by any electronic device with computing processing capability, for example, one or more of the

terminal devices

101, 102, 103 and/or the server 105 in fig. 1. In the following description, a terminal device is exemplified as an execution subject.

As shown in fig. 3, an information processing method provided by an embodiment of the present invention may include the following steps.

In step S310, a first character of the target object is determined.

In the embodiment of the present invention, the target object may be a certain game player of a certain game platform. When the game player logs on to the game platform with his game account, he can select one game character a from the game characters as the first character at his current time t 1.

In step S320, first object data of the target object is obtained according to the first character attribute information of the first character.

In an exemplary embodiment, the first role attribute information of the first role may include any one or more of identity information, current level information, and/or current operation information of the first role.

Here, a certain RPG (Role-playing game) game is taken as an example, and it is assumed that seven kinds of identities are preset: darts, catches, hunters, killers, music, swordmen and soldiers, each identity corresponds to a unique river and lake role, and target objects can be selectively added into the dart, catcher, hunter, killer, music, swordsman and swordsman.

In the RPG game, each identity can be further divided into three levels as advanced identities, and each advanced level can obtain stronger manufacturing skills and identity skills. The player can obtain the almanac value by completing the daily tasks, and further consume the almanac value to learn more related skills and identity steps. For example, the identity of the wenshi may include three successively increasing levels of the wenshi, the yashi, and the national shi, and the lowest level of the wenshi and the highest level of the national shi may be set. The line identity can include three grades that increase progressively in proper order of line, ling, name. The killer identity may include three sequentially increasing levels of killer, licker, and deinsectization. The identity of the catching device can comprise three levels of catching device, catching head and catching god which are sequentially increased. The darts identity may include three sequentially increasing grades of darts, darts heads, boomerangs. The swordsman identity can include three grades of swordsman, hero and hero which are sequentially increased. The hunter identity may include three sequentially increasing levels of hunter, hunter.

In the RPG game, game characters with different identities can perform different operations (such as tasks and/or play), and game characters with the same identity can perform different operations at different levels.

It should be understood that the above-mentioned identity information, current level information, current operation information, etc. are only used for illustration, and the first character attribute information may be adaptively adjusted in different application scenarios, or in different types of other games, or in the same type of other games.

In step S330, item attribute information of the object to be recommended is obtained.

Or taking a game scene as an example, the object to be recommended may be a game mall, such as a prop, in the game mall of the target game selected by the target object. The property information of the object at this time may include any one or more of the category to which each item belongs, the function that can be completed, the identity of the corresponding game character, the level of the corresponding game character, the operation of the corresponding game character, and the like, for example. Such as a prop for a game character to exercise light work, a prop for a game character to enhance internal force, and the like. Specifically, the article attribute information may be set according to actual requirements.

In step S340, the first object data and the item attribute information are processed through a neural network model, and a first target item recommended to a first character of the target object is determined from the object to be recommended.

According to the information processing method provided by the embodiment of the invention, the first role of the target object is determined, the first object data of the target object is obtained according to the first role attribute information of the first role, and the article attribute information of the object to be recommended is also obtained, so that the first object data and the article attribute information can be processed through a neural network model, and the first target article recommended to the first role of the target object is determined from the object to be recommended. On one hand, the target object in the object to be recommended is determined by comprehensively considering the rich role attribute information of the target object and the rich article attribute information of the object to be recommended, so that the accuracy of personalized recommendation of the article can be improved; on the other hand, the complex interaction between the target object and the object to be recommended is described through the nonlinear neural network model, so that the purposes of enhancing model interpretability and improving recommendation effect can be achieved.

Fig. 4 schematically shows a flow chart of an information processing method according to another embodiment of the present invention. As shown in fig. 4, compared with the above embodiment, the method provided by the embodiment of the present invention is different in that the following steps may be further included.

In step S410, a second role for the target object is determined.

For example, at another time t2, the game player can change to another game character B as the second character.

In step S420, second object data of the target object is obtained according to the second role attribute information of the second role.

In an exemplary embodiment, the second role attribute information of the second role may include any one or more of identity information, current level information, and/or current operation information of the second role.

In step S430, the second object data and the item attribute information are processed through the neural network model, and a second target item recommended to a second role of the target object is determined from the object to be recommended.

In the embodiment of the invention, the same user (for example, a game player) can select different game characters in the same game at different times, for example, the game character a is selected at the time t1, and is a low-level dart, and the system recommends the props required by the low-level dart to the user; at time t2, if game character B is selected as a high-level killer, the system will recommend to the user the property required by the high-level killer, that is, if the user selects a different game character, the game character may have attributes belonging to different genres and different identities, and the user will perform corresponding operations, for example, a type of property is required for practicing the law, and another type of property is required for practicing the law. The method comprises the steps that a prop pool can be preset, various props are arranged in the prop pool, each prop has corresponding article attribute information, and when a game role selected by a user changes and/or an attribute value of the selected game role changes, the type of the prop recommended to the prop can be changed by a system, or the arrangement sequence of the props recommended to the prop can be changed. In the embodiment of the invention, the selections of the same game commodity made by the same user under different game roles may be different, because the game commodity is a prop of the game role in the game scene, and the purchasing behavior is associated with the role attribute. In addition, if the character attribute of the same user is changed under the same game character, the selection of the same game commodity may be different, for example, the character may not need to buy a luxurious prop at a low level, and as the level is increased, a purchasing behavior may be generated for the prop.

Fig. 5 schematically shows a flowchart of an information processing method according to still another embodiment of the present invention. As shown in fig. 5, compared with the above embodiment, the method provided by the embodiment of the present invention is different in that the following steps may be further included.

In step S510, a historical role of the target object is obtained.

In step S520, historical object data of the target object is obtained according to the historical role attribute information of the historical role.

In step S530, a history operation record of the target object on the object to be recommended is obtained.

In step S540, the historical object data and the label of the item attribute information are determined according to the historical operation record, so as to generate a training data set.

In the embodiment of the invention, the used neural network model can be trained in advance. When training the model, a training data set is first constructed. The historical operation record of the object to be recommended is used. Or taking the above game scenario as an example, the object to be recommended may be each item in an item pool, and the historical operation record may be a purchase record of each item historically by the game player, but the present invention is not limited thereto, and may also be a usage record, a preview record, and the like of each item historically by the game player.

In the embodiment of the present invention, the training data set may be constructed in the following form:

the data can be divided into two parts: target object and first object data composed of first character attribute information thereof

Formed by object to be recommended and attribute information of object

The user related data contains fields as follows: [ user _ id, user _ feature _1, user _ feature _2, …]The commodity-related data contains fields of: [ item _ id, item _ feature _1, item _ feature _2, …]Wherein id and each feature field are represented by numerical values, [ feature _1, feature _2, …]Referred to as attribute information (context). Wherein the feature of the user in the game is not fixed but varies with time, so the same id can correspond to a plurality of attribute information.

(1)

First role attribute information being a target object id and a first role of the target objectFor information splicing, one target object can correspond to a plurality of target objects because the same target object can be in different game roles at different times

For example, the target object id is 1, and in context [0.1,0.3, …,0.2 ═ c]Time of day corresponding

When id is 1, context is 0.5,0.1, …,0.1]Time of day corresponding

The meaning of this representation is: a target object under a particular game character.

(2)

The method can be used for splicing the game commodity id and the article attribute information thereof.

(3) Label (label): each pair of which

And

corresponding to a binary label, 0 indicates that the target object does not purchase the prop under the role attribute information, 1 indicates that the target object purchases the prop under the role attribute information, and a specific label value can be set autonomously.

In step S550, the neural network model is trained using the training data set.

In the embodiment of the invention, the training data can be concentrated in the model training process

And

input to the neural network model, the neural network modelAnd outputting a predicted value, calculating a loss function according to the predicted value and the real label, calculating a parameter gradient by taking the minimized loss function as a target, and performing gradient back propagation to update the model parameter.

The process is iterated, model parameters are updated continuously, model verification is carried out through verification set data, iteration can be stopped when loss functions on the verification sets basically do not decrease any more, and a model with good fitting capacity and generalization capacity is obtained at the moment.

In the embodiment of the invention, the loss function can adopt a binary cross entropy loss function loss, and the calculation mode is as follows:

in the above formula, k is a positive integer greater than or equal to 1 and less than or equal to m, and m is a positive integer greater than or equal to 1, which represents a total of m pairs in the training data set

And

y_kwhen the model is input into the kth pair

And

the value of the corresponding real label is 0 or 1;

when the model is input into the kth pair

And

the predicted value output by the time model is taken as [0,1]]The numerical value in between. The optimization aims atAnd minimizing the loss, wherein the loss is smaller when the predicted value is close to the true value, and the loss is larger when the predicted value deviates from the true value.

The embodiment of the invention also comprises a model test process. Model testing refers to testing the predictive ability of a model on a trained model with a test data set that does not coincide with the training data set. To evaluate the predictive power of the model more intuitively, the evaluation criteria used on the test data set are the accuracy of prediction, recall, AUC (Area Under the receiver operating characteristic Curve (ROC)) Curve and the Area enclosed by the coordinate axes). The accuracy rate can represent the accuracy capability of the model, the recall rate can represent the recall capability of the model, and the AUC is the combination of the accuracy rate and the recall capability.

By comparing the accuracy, recall and AUC values of different models on the test data set, the predictive power of each model can be observed.

In the embodiment of the invention, in the process of training the model, not only the historical operation record of the target object on the object to be recommended is considered, but also the historical role attribute information of the target object and the article attribute information of the object to be recommended are comprehensively considered, so that the prediction accuracy of the model can be improved.

Fig. 6 is a schematic diagram illustrating a processing procedure of step S320 illustrated in fig. 3 in an embodiment. As shown in fig. 6, in the embodiment of the present invention, the step S320 may further include the following steps.

In step S321, object attribute information of the target object is acquired.

For example, in the case where a target object is a certain game player, the object attribute information may be personal figure information such as the real name, age, sex, geographical location, and historical online time of the game player.

In step S322, the first object data is obtained according to the first character attribute information and the object attribute information.

Specifically, the first character attribute information and the object attribute information may be vector-spliced to form the first object data, that is, when recommending a property to the target object, the first object data may not only include virtual attribute information of a currently selected game character, but also comprehensively consider real attribute information of the target object, thereby further improving recommendation accuracy.

Fig. 7 is a schematic diagram illustrating a processing procedure of step S340 illustrated in fig. 3 in an embodiment. In the embodiment of the present invention, the neural network model may include a first neural network submodel, a second neural network submodel, and a third neural network submodel.

As shown in fig. 7, in the embodiment of the present invention, the step S340 may further include the following steps.

In step S341, an object vector is generated from the first object data.

In step S342, an item vector of the object to be recommended is generated according to the item attribute information.

In step S343, the object vector and the item vector are processed by the first neural network sub-model, so as to obtain an object preference vector of the first character of the target object and an item characteristic vector of the object to be recommended.

In step S344, the object vector and the item vector are processed through the second neural network sub-model, so as to obtain an interaction relationship vector between the first role of the target object and the object to be recommended.

In step S345, the object preference vector, the item characteristic vector, and the interaction relation vector are processed through the third neural network sub-model, so as to obtain a prediction probability value of the first character operation of the target object on the object to be recommended.

In step S346, a first target item recommended to a first persona of the target object is determined according to the predicted probability value.

In the embodiment of the present invention, according to the predicted probability values of the properties in the property pool, sorting from large to small, selecting the properties of a predetermined number (for example, 10, the specific value is not limited) or a predetermined proportion (for example, 10%, the specific value is not limited) as the first target object, returning the first target object to the game client for display, displaying the maximum predicted probability value in the first position, and displaying the minimum predicted probability value in the first target object in the last position, but the present invention is not limited thereto. In other embodiments, if the number of all the props in the prop pool is not large, all the props can be displayed on the game client, and the props are sorted according to the size of the predicted probability value.

In the embodiment of the present invention, the data input into the model may be preprocessed first. The features of each dimension of the data comprise discrete features (discrete features) and continuous features (dense features), for the discrete features, sparse vectors can be converted into dense representation by embedding (embedding), for the continuous features, as the data distribution of partial continuous features is often long-tailed distribution, the numerical value with larger difference under the distribution can cause interference to the model, therefore, the continuous features are firstly subjected to binning, and then subjected to data normalization processing.

Fig. 8 is a schematic diagram illustrating a processing procedure of step S341 shown in fig. 7 in an embodiment. In an embodiment of the present invention, the neural network model may further include a first embedded sub-model, and the first object data may include object discrete features and object continuous features.

As shown in fig. 8, in the embodiment of the present invention, the step S341 may further include the following steps.

In step S3411, the object discrete features are processed by the first embedding sub-model to obtain an object embedding vector.

In step S3412, the object continuous features are subjected to binning processing to obtain an object discrete representation.

In the embodiment of the present invention, any one of an equal frequency binning mode, an equidistant binning mode, a chi-square binning mode, and the like may be adopted, which is not limited to this.

In step S3413, the discrete representation of the object is normalized.

In step S3414, the object embedding vector and the normalized object discrete representation are concatenated to generate the object vector.

Fig. 9 schematically shows a schematic diagram of a first embedding submodel according to an embodiment of the invention.

As shown in fig. 9, the first object data

Assume 4, 3, …,0.2, 0.5, …. Where 4 and 3 are object discrete features whose one-hot codes are 00001 and 0001, respectively, which are input to the embedding layer of the first embedding submodel, respectively, outputting a dense object embedding vector. 0.2 and 0.5 are the discrete representation of the normalized object continuous characteristic, which is kept unchanged, and cascaded with the embedded object vector to output the object vector

Fig. 10 is a schematic diagram illustrating a processing procedure of step S342 illustrated in fig. 7 in an embodiment. In the embodiment of the present invention, the neural network model may further include a second embedded sub-model, and the article attribute information of the object to be recommended may include an article discrete feature and an article continuous feature.

As shown in fig. 10, in the embodiment of the present invention, the step S342 may further include the following steps.

In step S3421, the discrete features of the article are processed by the second embedding sub-model to obtain an article embedding vector.

In step S3422, the item continuous features are subjected to binning processing to obtain an item discrete representation.

In step S3423, the discrete representation of the item is normalized.

In step S3424, the item embedding vector and the normalized item discrete representation are spliced to generate the item vector.

Fig. 11 schematically shows a schematic diagram of a second embedding submodel according to an embodiment of the invention.

As shown in fig. 11, the article attribute information

Assume 2, 1, …, 0.4, 0.1, …. Where 2 and 1 are discrete features of the article, their one-hot codes are 00100 and 0100 respectively, which are input to the embedding layer of the second embedding sub-model, respectively, outputting a dense article embedding vector. 0.4 and 0.1 are normalized article discrete representation of article continuous characteristics, which is kept unchanged, and cascaded with the article embedding vector to output the article vector

Fig. 12 schematically shows a structural diagram of a neural network model according to an embodiment of the present invention.

As shown in fig. 12, first object data is input to the first embedding submodel, and an object vector is output; and inputting the article attribute information into a second embedding sub-model and outputting an article vector. Inputting the object vector into a first neural network sub-model, and outputting an object preference vector and an article characteristic vector; and inputting the object vector into a second neural network submodel, and outputting an interactive relation vector. And inputting the object preference vector, the article characteristic vector and the interactive relation vector into a third neural network submodel, and outputting a prediction probability value.

Fig. 13 is a schematic diagram illustrating a processing procedure of step S343 shown in fig. 7 in an embodiment. In an embodiment of the present invention, the first neural network submodel may include a first neural network unit and a second neural network unit. The first neural network sub-model is a representation learning module (representation learning), which can be further divided into two parts: one part is used for learning the vector representation of the target object, and the other part is used for learning the vector representation of the object to be recommended.

As shown in fig. 13, in the embodiment of the present invention, the step S343 may further include the following steps.

In step S3431, the object vector is processed by the first neural network unit to obtain the object preference vector.

In step S3432, the item vector is processed by the second neural network unit to obtain the item characteristic vector.

Fig. 14 schematically shows a structural schematic diagram of a first neural network submodel according to an embodiment of the present invention.

As shown in fig. 14, the object vector is generated

Input to the first neural network unit MLP_user，MLP_userThe multilayer ceramic comprises a layer 1, a layer 2 and a layer n1 which are connected in sequence, wherein n1 is a positive integer greater than or equal to 1, and n1 can be 3-5. First neural network element MLP_userOutputting object preference vectors

To vector an article

Input to a second neural network element MLP_item，MLP_itemThe multilayer ceramic comprises a layer 1, a layer 2 and a layer n2 which are sequentially connected, wherein n2 is a positive integer greater than or equal to 1, n2 can be 3-5, and n1 can be the same as or different from n2, and the invention is not limited to this. Second neural network element MLP_itemOutputting the article characteristic vector

In the embodiment of fig. 14, the first neural network Unit and the second neural network Unit are exemplified by using a Multilayer Perceptron (MLP), but in other embodiments, the first neural network Unit and/or the second neural network Unit may also use other deep learning networks, such as LSTM (Long Short-term memory), GRU (Gated Recurrent Unit), etc., and the first neural network Unit may use the same deep learning network as the second neural network Unit, or may use different deep learning networks, which is not limited in this disclosure. Wherein the content of the first and second substances,a multi-layered perceptron is a neural network of forward structure that maps a set of input vectors to a set of output vectors. In the embodiment of FIG. 14, MLP learning vector characterization is used, i.e., input separately

And

respectively output corresponding

And

fig. 15 is a schematic diagram illustrating a processing procedure of step S344 shown in fig. 7 in an embodiment. As shown in fig. 15, in the embodiment of the present invention, the step S344 may further include the following steps.

In step S3441, the object vector and the article vector are spliced to obtain a spliced vector.

In step S3442, the stitching vector is processed through the second neural network submodel to obtain the interaction relationship vector.

FIG. 16 schematically shows a structural schematic diagram of a second neural network submodel according to an embodiment of the present invention.

As shown in fig. 16, the object vector is generated

And item vector

Cascading (splicing) is carried out to obtain a splicing vector; inputting the stitching vector into a second neural network submodel MLP_{user_item}，MLP_{user_item}The multilayer ceramic comprises a layer 1, a layer 2 and a layer n3 which are sequentially connected, wherein n3 is a positive integer greater than or equal to 1, n3 can be 3-5, and n3 can be equal to or different from n1 or n2, and the invention is not limited to this.Second neural network submodel MLP_{user_item}Outputting an interaction relation vector

In the embodiment of the invention, the second neural network submodel MLP_{user_item}It may also be called a relationship learning (relationship learning) module. The representation of the target object and the representation of the object to be recommended are obtained by separate learning in the characterization learning module, and the relationship learning module is designed to learn the complex interaction relationship between the target object and the object to be recommended. In this section, first, theAnd

are connected in series and then the splicing vector is sent into MLP_{user_item}. Compared with the method that the relation between the target object and the object to be recommended is described by using a linear function of vector inner product in the related technology, the nonlinear structure of the MLP is more beneficial to describing a complex process from characterization to result.

In other embodiments, the second neural network submodel may employ other neural networks or deep learning networks in addition to MLP, such as LSTM, GRU, etc. The first neural network submodel and the second neural network submodel can both adopt MLP, can also both adopt other same neural networks, and can also respectively adopt different neural networks.

Deep learning is a technology which is suitable for processing large-scale complex data and has strong expressive force, and is applied to a recommendation system, the embodiment of the invention designs a collaborative filtering network structure based on deep learning in a game scene, and a neural network replaces the traditional vector inner product to be used for calculating the interactive relation between a user and a commodity.

Fig. 17 is a schematic diagram illustrating a processing procedure of step S345 illustrated in fig. 7 in an embodiment. As shown in fig. 17, in the embodiment of the present invention, the step S345 may further include the following steps.

In step S3451, the object preference vector is multiplied by the item characteristic vector to obtain a point product vector.

In step S3452, the point-multiplied vector and the interaction relation vector are concatenated to obtain a concatenated vector.

In step S3453, the concatenated vector is processed by the third neural network sub-model to obtain the prediction probability value.

In this embodiment of the present invention, the third neural network sub-model may also be referred to as a joint prediction layer (prediction layer). After passing through the characterization learning module and the relation learning module, the purchase probability of the target object to the game commodity needs to be predicted, and the final prediction result is obtained through calculation by combining the output of the characterization learning module and the relation learning module of the prediction layer. For example, the calculation of the joint prediction layer may be as follows:

in the above formula, ⊙ denotes vector element multiplication,

representing vector concatenation, the selected function f can be a single-layer full-connection layer plus softmax function, output

Is a two-dimensional vector, shaped asThe first bit represents the probability value that the user does not purchase the commodity, the second bit represents the probability value that the user purchases the commodity, namely the predicted probability value, and the sum of the two probabilities is 1.

The following description will take an example in which the scheme provided by the above embodiment is applied to a property recommendation scene of an RPG game. In the game, a user can select game roles, and the game roles are associated with rich attributes such as role modification, power and the like (the value becomes larger and larger along with the side length of online time duration); the user can enter the game mall to purchase the prop required by the role, and the prop is also associated with various attributes. The purpose of recommendation in this scenario is to recommend one or more items that are most likely to be of interest to a user according to attribute information of the user (which may comprehensively consider attribute information of the user such as the real gender, not limited to the user id for example), the role, and the items.

The RPG Game is a MMORPG (Massive Multiplayer Online Role Playing Game) Game of martial arts themes, which comprises eight barren men groups, including Taibai, Shenwei, Tangmen, beggar's upper, Zhenwu, Tianxiang, Wudu and Shaolin groups; the device also comprises various industries such as rivers, lakes and major, wherein the scales of the groups such as quick catches, darts, hunters, swordmen, killers, music and souls, and the like are the largest, and the groups such as eye-hanging, merchant and city wells are also provided. The user may download and install the game client. After the game client is installed, the game login device of the game can be operated by clicking the game icon on the desktop. The game account login interface can be accessed by clicking 'enter game' on the right side of the login device, the game account and the password are input on the interface, then the 'enter game' is clicked, the server list can be opened by checking 'more servers', the game area and the server to be logged in are selected (or the server which is logged in recently can be directly selected on the right side), and the 'start game' is clicked, so that the river and lake can be started to travel.

FIG. 18 schematically illustrates an interface diagram of a portal introduction in accordance with an embodiment of the invention.

As shown in fig. 18, the user first performs creation of a character. The first step is to select the gate. If the user logs in the game for the first time, the game account does not have any role under the server, a gate selection interface of the game can be seen, and any opened gate can be selected under the interface to create the role of the user. And clicking any one of the opened door dispatches and entering a door dispatcher introduction interface. The currently selected gate and the appearance of the character can be seen in the center of the interface, the fighting characteristics of the gate are listed on the right side of the interface, the gender of the character can be switched below, and the character can be viewed by clicking the right mouse button. The left side can switch role gate rapidly.

Fig. 19 and 20 schematically illustrate interface diagrams for role detail customization, according to an embodiment of the present invention.

As shown in FIG. 19, clicking on "customize details" can further customize the character details, and more details of the character can be seen under this interface and can be fine-tuned for each item. There are more than 200 adjustable parameters in total on 48 bones, based on the natural distribution of bones and muscles in the human face. Also as shown in fig. 20, previews of multiple (here 4 for example) facial expressions and multiple (here 5 for example) external fits are provided. After the role name is input below the interface, the user role can be established by clicking 'creation completed'. After the character is established, the established character is selected, and clicking "start game" below the character as shown in fig. 21 enters the game scene as shown in fig. 22.

As shown in fig. 22, side shortcut bar, chat interface, character status, bottom shortcut bar, function buttons (character and mall), minimap, task play directions.

The function buttons on the game interface are the entrances of the most important function modules of the game. Wherein, the role includes: attributes, meridians, heart law, equipment, etc. The identity includes: darts, killers, swordmen, hunters, music, soldiers, quick catches, eye-hanging, merchant and city wells are not opened temporarily. The shopping mall includes: common use, appearance, pet-riding pendant, treasure, member exclusive share, daily reward point, binding point coupon, purchasing member, point coupon consignment and point coupon recharge.

The character state is information for viewing qi and blood, inspiration, strength, disinhibition and various gate-specific recruits of the current character.

The bottom shortcut bar corresponds to 15 keys of QERTG and 0-9, and the right page-up and page-down buttons can be switched. The shortcut bar can be used for placing props and skills and can be used quickly after corresponding keys are pressed down. The key position on the shortcut bar can be changed in the system setting.

The side shortcut bars correspond to 25 key positions of F1-F10, Ctrl0-Ctrl9 and HVCXZ, can be used for placing props and skills, and can be used quickly after corresponding keys are pressed down. The key position on the shortcut bar can be changed in the system setting.

The chat interface may be used to display chat messages and system messages currently with other players in the game. The channel of the utterance can be selected and the content of the utterance can be entered in the upper part of the chat window.

The player information can be used for checking the head portrait, the gate, the name and the grade information of the current character.

The minimap here shows the surrounding map information of the environment in which the player is currently located. The key at the upper left corner is a time-rain calendar which can display the current area, date and time and weather; the key at the upper right corner is a mailbox; the keys on the lower right corner are: zoom in, zoom out, view world maps, view large maps.

The task play guide is used to track current storyline task/play event information. Clicking the link above can guide to the target point, and clicking the plot/play page mark above can switch to view the information of the plot task and the play activity.

In this game, the reading is closely related to the elevation of the character level, that is, the experience of the character. The reading value is displayed at the top of the game client, and a yellow progress bar is used for displaying the currently owned reading and the reading required by next level-up. Besides the main line task, the reading process also comprises ways of gate style sitting, daily meeting, swinging, other supplementary playing methods and the like, and the 'must do every day' on the left side of the small map on the right upper side of the game interface can be opened to specifically view the completion situation every day.

FIG. 23 is an interface diagram that schematically illustrates a role attribute, in accordance with an embodiment of the present invention.

As shown in fig. 23, the role attributes may include: a house number (e.g., dong men), title, achievement value, kill, modification, vitality, etc., wherein modification is a meridian and heart method that can be used to refine a character. A click may select to view all attributes, the meridian attribute, the psychology attribute, and the equipment attribute. All attributes may include base attributes and combat attributes. The basic properties may include qi and blood, endogenous qi, strength, root and bone, strength of qi, insights, and physical techniques. The combat attributes may in turn include exo-attack, endo-attack, brute force attack, hit rate, heart-beat injury, exo-work defense, endo-work defense, brute force defense, hold-off, and toughness.

FIG. 24 schematically illustrates an interface diagram for a gaming mall, according to an embodiment of the present invention.

As shown in fig. 24, all the props, equipped props, stone needle props, heart methods props, and identity props in the game mall can be selected for viewing. All the properties can include purple heart annotation, rhizoma Matteucciae chinensis, Tibetan antelope skin material, phoenix dance picture, Margarita, TIANZHIDAN, SHUSHENGLI, jade colored glaze, golden silk ear, and LONGDAN Sand. The display setting period of the commodity attribute is 7 days, and the arrangement sequence of the props is updated every 7 days according to the change of the role attribute of the user.

As shown in fig. 25, an information processing apparatus 2500 according to an embodiment of the present invention may include: a first object role determination module 2510, a first object data obtaining module 2520, an item attribute information obtaining module 2530, and a first target item determination module 2540.

Among other things, the first object persona determination module 2510 may be configured to determine a first persona of a target object. The first object data obtaining module 2520 may be configured to obtain first object data of the target object according to first character attribute information of the first character. The item attribute information obtaining module 2530 may be configured to obtain item attribute information of an item to be recommended. The first target item determination module 2540 may be configured to determine, from the object to be recommended, a first target item recommended to a first persona of the target object by processing the first object data and the item attribute information through a neural network model.

In an exemplary embodiment, the information processing apparatus 2500 may further include: a second object role determination module configurable to determine a second role for the target object; a second object data obtaining module, configured to obtain second object data of the target object according to second role attribute information of the second role; the second target item determination module may be configured to process the second object data and the item attribute information through the neural network model, and determine a second target item recommended to a second role of the target object from the object to be recommended.

In an exemplary embodiment, the information processing apparatus 2500 may further include: a historical role obtaining module configured to obtain a historical role of the target object; a historical object data obtaining module, configured to obtain historical object data of the target object according to historical role attribute information of the historical role; the historical operation record obtaining module can be configured to obtain the historical operation record of the target object on the object to be recommended; a training set generation module configured to determine labels of the historical object data and the item attribute information according to the historical operation records to generate a training data set; a model training module may be configured to train the neural network model using the training data set.

In an exemplary embodiment, the neural network model may include a first neural network submodel, a second neural network submodel, and a third neural network submodel. Among other things, the first target item determination module 2540 may include: an object vector generation unit that may be configured to generate an object vector from the first object data; an item vector generating unit configured to generate an item vector of the object to be recommended according to the item attribute information; a target object characteristic vector obtaining unit, which may be configured to process the object vector and the object vector through the first neural network sub-model, to obtain an object preference vector of a first character of the target object and an object characteristic vector of the object to be recommended; the interactive relationship vector obtaining unit may be configured to process the object vector and the item vector through the second neural network sub-model to obtain an interactive relationship vector between the first role of the target object and the object to be recommended; the prediction probability obtaining unit may be configured to process the object preference vector, the item characteristic vector, and the interaction relation vector through the third neural network sub-model to obtain a prediction probability value of the first character operation of the target object on the object to be recommended; a first target item determination unit may be configured to determine a first target item recommended to a first persona of the target object according to the predicted probability value.

In an exemplary embodiment, the first neural network submodel may include a first neural network element and a second neural network element. Wherein the object item characteristic vector obtaining unit may include: an object preference vector obtaining subunit, which may be configured to process the object vector by the first neural network unit to obtain the object preference vector; an item characteristic vector obtaining subunit, which may be configured to process the item vector by the second neural network unit, to obtain the item characteristic vector.

In an exemplary embodiment, the interaction relation vector obtaining unit may include: a splicing vector obtaining subunit, configured to splice the object vector and the article vector to obtain a splicing vector; and the interactive relation vector obtaining subunit may be configured to process the stitching vector through the second neural network submodel to obtain the interactive relation vector.

In an exemplary embodiment, the prediction probability obtaining unit may include: a point-multiplied vector obtaining subunit, which may be configured to multiply the object preference vector and the article characteristic vector to obtain a point-multiplied vector; a concatenated vector obtaining subunit which may be configured to concatenate the point-multiplied vector and the interaction relationship vector to obtain a concatenated vector; and the prediction probability obtaining subunit may be configured to process the concatenated vector through the third neural network submodel to obtain the prediction probability value.

In an exemplary embodiment, the neural network model may further include a first embedded sub-model, and the first object data may include object discrete features and object continuous features. Wherein the object vector generation unit may include: an object embedding vector obtaining subunit, configured to process the object discrete feature through the first embedding submodel to obtain an object embedding vector; the object discrete representation obtaining subunit may be configured to perform binning processing on the object continuous features to obtain an object discrete representation; an object normalization processing subunit, which can be configured to perform normalization processing on the object discrete representation; and the object vector generating subunit may be configured to splice the object embedding vector and the normalized object discrete representation to generate the object vector.

In an exemplary embodiment, the neural network model may further include a second embedded sub-model, and the item attribute information of the item to be recommended may include an item discrete feature and an item continuous feature. Wherein the item vector generation unit may include: an article embedding vector obtaining subunit, which may be configured to process the article discrete features through the second embedding submodel, to obtain an article embedding vector; an article discrete representation obtaining subunit, configured to perform binning processing on the article continuous features to obtain an article discrete representation; an item normalization processing subunit, which can be configured to perform normalization processing on the item discrete representation; and the article vector generating subunit may be configured to splice the article embedding vector and the normalized article discrete representation to generate the article vector.

In an exemplary embodiment, the first object data obtaining module 2520 may include: an object attribute information acquisition unit that may be configured to acquire object attribute information of the target object; a first object data obtaining unit may be configured to obtain the first object data based on the first character attribute information and the object attribute information.

In an exemplary embodiment, the first character attribute information of the first character may include identity information, current level information, and/or current operation information of the first character.

The specific implementation of each module, unit, and sub-unit in the information processing apparatus provided in the embodiment of the present invention may refer to the content in the information processing method, and will not be described herein again.

It should be noted that although several modules, units and sub-units of the apparatus for action execution are mentioned in the above detailed description, such division is not mandatory. Indeed, the features and functionality of two or more modules, units and sub-units described above may be embodied in one module, unit or sub-unit in accordance with an embodiment of the invention. Conversely, the features and functions of one module, unit and sub-unit described above may be further divided into embodiments by a plurality of modules, units and sub-units.

Through the above description of the embodiments, those skilled in the art will readily understand that the exemplary embodiments described herein may be implemented by software, or by software in combination with necessary hardware. Therefore, the technical solution according to the embodiment of the present invention can be embodied in the form of a software product, which can be stored in a non-volatile storage medium (which can be a CD-ROM, a usb disk, a removable hard disk, etc.) or on a network, and includes several instructions to enable a computing device (which can be a personal computer, a server, a touch terminal, or a network device, etc.) to execute the method according to the embodiment of the present invention.

Other embodiments of the invention will be apparent to those skilled in the art from consideration of the specification and practice of the invention disclosed herein. This application is intended to cover any variations, uses, or adaptations of the invention following, in general, the principles of the invention and including such departures from the present disclosure as come within known or customary practice within the art to which the invention pertains. It is intended that the specification and examples be considered as exemplary only, with a true scope and spirit of the invention being indicated by the following claims.

It will be understood that the invention is not limited to the precise arrangements described above and shown in the drawings and that various modifications and changes may be made without departing from the scope thereof. The scope of the invention is limited only by the appended claims.

Claims

1. An information processing method characterized by comprising:

determining a first role of a target object;

obtaining first object data of the target object according to the first role attribute information of the first role;

acquiring article attribute information of an object to be recommended;

and processing the first object data and the item attribute information through a neural network model, and determining a first target item recommended to a first role of the target object from the object to be recommended.

2. The method of claim 1, further comprising:

determining a second role for the target object;

obtaining second object data of the target object according to second role attribute information of the second role;

and processing the second object data and the item attribute information through the neural network model, and determining a second target item recommended to a second role of the target object from the object to be recommended.

3. The method of claim 1, further comprising:

obtaining a historical role of the target object;

obtaining historical object data of the target object according to historical role attribute information of the historical roles;

obtaining a historical operating record of the target object on the object to be recommended;

determining labels of the historical object data and the article attribute information according to the historical operation records to generate a training data set;

training the neural network model using the training data set.

4. The method of claim 1, wherein the neural network model comprises a first neural network submodel, a second neural network submodel, and a third neural network submodel; the method for determining the first target object recommended to the first role of the target object from the object to be recommended by processing the first object data and the object attribute information through a neural network model comprises the following steps:

generating an object vector from the first object data;

generating an article vector of the object to be recommended according to the article attribute information;

processing the object vector and the article vector through the first neural network submodel to obtain an object preference vector of a first character of the target object and an article characteristic vector of the object to be recommended;

processing the object vector and the item vector through the second neural network sub-model to obtain an interactive relation vector between a first role of the target object and the object to be recommended;

processing the object preference vector, the article characteristic vector and the interaction relation vector through the third neural network submodel to obtain a prediction probability value of the first character operation of the target object on the object to be recommended;

and determining a first target item recommended to a first role of the target object according to the predicted probability value.

5. The method of claim 4, wherein the first neural network submodel comprises a first neural network element and a second neural network element; wherein, processing the object vector and the item vector through the first neural network submodel to obtain an object preference vector of a first character of the target object and an item characteristic vector of the object to be recommended, comprises:

processing the object vector by the first neural network unit to obtain the object preference vector;

and processing the item vector through the second neural network unit to obtain the item characteristic vector.

6. The method of claim 4, wherein processing the object vector and the item vector through the second neural network submodel to obtain an interaction relationship vector between the first character of the target object and the object to be recommended comprises:

splicing the object vector and the object vector to obtain a spliced vector;

and processing the splicing vector through the second neural network submodel to obtain the interactive relation vector.

7. The method of claim 4, wherein processing the object preference vector, the item characteristic vector and the interaction relation vector through the third neural network submodel to obtain a predicted probability value of the first character operation of the target object on the object to be recommended comprises:

multiplying the object preference vector by the article characteristic vector to obtain a point product vector;

connecting the point multiplication vector and the interaction relation vector in series to obtain a connected vector;

and processing the concatenated vector through the third neural network submodel to obtain the prediction probability value.

8. The method of claim 4, wherein the neural network model further comprises a first embedded submodel, the first object data comprising object discrete features and object continuous features; wherein generating an object vector from the first object data comprises:

processing the object discrete features through the first embedding sub-model to obtain an object embedding vector;

performing box separation processing on the object continuous characteristics to obtain an object discrete representation;

performing normalization processing on the object discrete representation;

and splicing the object embedded vector and the normalized object discrete representation to generate the object vector.

9. The method according to claim 4, wherein the neural network model further comprises a second embedded submodel, and the item attribute information of the object to be recommended comprises an item discrete feature and an item continuous feature; generating an article vector of the object to be recommended according to the article attribute information, wherein the method comprises the following steps:

processing the discrete features of the article through the second embedding sub-model to obtain an article embedding vector;

performing box separation processing on the continuous features of the articles to obtain discrete representations of the articles;

performing normalization processing on the discrete representation of the article;

and splicing the article embedding vector and the normalized article discrete representation to generate the article vector.

10. The method of claim 1, wherein obtaining first object data of the target object according to the first character attribute information of the first character comprises:

acquiring object attribute information of the target object;

and obtaining the first object data according to the first character attribute information and the object attribute information.

11. The method of claim 1, wherein the first persona attribute information for the first persona includes identity information, current level information, and/or current operation information for the first persona.

12. An information processing apparatus characterized by comprising:

a first object role determination module configured to determine a first role of a target object;

a first object data obtaining module configured to obtain first object data of the target object according to first role attribute information of the first role;

the article attribute information acquisition module is configured to acquire article attribute information of an object to be recommended;

and the first target item determination module is configured to process the first object data and the item attribute information through a neural network model, and determine a first target item recommended to a first role of the target object from the object to be recommended.

13. An electronic device, comprising:

one or more processors;

a storage device configured to store one or more programs that, when executed by the one or more processors, cause the one or more processors to implement the information processing method according to any one of claims 1 to 11.

14. A computer-readable storage medium storing a computer program, wherein the computer program is executed by a processor to implement the information processing method according to any one of claims 1 to 11.