Object recognition method and device, electronic equipment and storage medium

Abstract

The invention discloses an object recognition method and device, electronic equipment and a storage medium, and is applied to the field of computers. The method comprises the following steps: acquiring basic feature data corresponding to an account object; a first probability is determined according to the basic feature data, the first probability is the probability that the account object becomes a dyeing account object of the platform application program and a preset operation is executed on an application program object in the platform application program, and the dyeing account object refers to an account executing a registration operation on the application program object in the platform application program; a second probability is determined according to the basic feature data, the second probability refers to the probability that the account object becomes a key account object, and the key account object is an account of which the condition of executing a preset operation meets a preset requirement; a target probability is determined according to the first probability and the second probability, the target probability refers to the probability that the account object becomes a target object, and the target object refers to a dyed account object with the condition that preset operation is executed on the application program object meets the preset requirement. According to the method, target object recognition is realized.

Description

Technical Field

[0001] The present application relates to the field of computer technology, and in particular to object recognition methods, devices, electronic devices, and storage media. Background Art

[0002] With the continuous development and popularization of internet technology, many applications are increasing their interaction with account objects through platform applications, or application communities, facilitating interaction between account objects and these applications. For a platform application, the first registration of an account object in an application (application object) associated with the platform application is considered the completion of a coloring operation on the account object, making the account object a colored account object with respect to the platform application.

[0003] Different colored account objects perform different preset operations within application objects. Colored account objects that can perform preset operations that meet preset requirements are of great significance to application objects and even platform applications, and are the account objects that platform applications and application objects focus on serving. Therefore, predictive identification of these account objects is crucial for platform applications. However, the relevant art lacks a technical solution that can accurately identify these account objects. Summary of the Invention

[0004] An embodiment of the present application provides an object recognition method that significantly improves the recognition accuracy of target objects that are important for platform applications. The target object is a colored account object that performs preset operations that meet preset requirements, thereby solving at least one of the aforementioned technical problems.

[0005] According to one aspect of an embodiment of the present application, a method for object recognition is provided, the method comprising:

[0006] Get the basic feature data corresponding to the account object;

[0007] determining, based on the basic feature data, a first probability, the first probability being a probability that the account object becomes a dyed account object and performs a preset operation on an application object in the platform application, the dyed account object being an account object that performs a registration operation on the application object in the platform application;

[0008] determining a second probability based on the basic feature data, where the second probability indicates a probability that the account object becomes a key account object, where the key account object is an account that satisfies the preset requirement when performing the preset operation;

[0009] A target probability is determined based on the first probability and the second probability, where the target probability indicates a probability that the account object becomes a target object, where the target object refers to a colored account object whose situation of performing the preset operation on the application object meets the preset requirements.

[0010] According to one aspect of an embodiment of the present application, there is provided an object recognition device, the device comprising:

[0011] Basic data acquisition module, used to obtain basic feature data corresponding to account objects;

[0012] The object recognition module is used to perform the following operations:

[0013] determining, based on the basic feature data, a first probability, the first probability being a probability that the account object becomes a dyed account object and performs a preset operation on an application object in the platform application, the dyed account object being an account object that performs a registration operation on the application object in the platform application;

[0014] determining a second probability based on the basic feature data, where the second probability indicates a probability that the account object becomes a key account object, where the key account object is an account that satisfies the preset requirement when performing the preset operation;

[0015] A target probability is determined based on the first probability and the second probability, where the target probability indicates a probability that the account object becomes a target object, where the target object refers to a colored account object whose situation of performing the preset operation on the application object meets the preset requirements.

[0016] According to one aspect of an embodiment of the present application, a computer device is provided, comprising a processor and a memory, wherein the memory stores at least one instruction, at least one program, a code set, or an instruction set, and the at least one instruction, the at least one program, the code set, or the instruction set is loaded and executed by the processor to implement the above-mentioned object recognition method.

[0017] According to one aspect of an embodiment of the present application, a computer-readable storage medium is provided, wherein the storage medium stores at least one instruction, at least one program, a code set, or an instruction set, and the at least one instruction, the at least one program, the code set, or the instruction set is loaded and executed by a processor to implement the above-mentioned object recognition method.

[0018] According to one aspect of an embodiment of the present application, a computer program product is provided, the computer program product including computer instructions stored in a computer-readable storage medium. A processor of a computer device reads the computer instructions from the computer-readable storage medium and executes the computer instructions, causing the computer device to implement the object recognition method described above.

[0019] The technical solutions provided in the embodiments of the present application can bring the following beneficial effects:

[0020] In order to promote the interaction between account objects and platform applications and application objects related to the platform applications, enhance the viscosity of account objects, and promote the transformation of account objects into target objects, an embodiment of the present application provides an object recognition method. The object recognition method can be based on the Bayesian principle and convert the recognition problem of colored account objects that meet preset requirements when performing preset operations, that is, the recognition problem of target objects, into two sub-recognition problems, which are Problem 1 and Problem 2 respectively.

[0021] Problem 1 involves identifying associated account objects. Associated account objects are objects that are stained with platform applications and perform preset operations on application objects. The identification result of Problem 1 is the probability that an account object is an associated account object. Problem 2 involves identifying whether an account object can become a key account object. A key account object is an account object that performs preset operations that meet preset requirements, or can be simply understood as a head account object. The identification result of Problem 2 is the probability that the account object is a head account object. Based on the identification results of Problems 1 and 2, the probability that an account object stained with a platform application and performing preset operations on an application object meets the preset requirements can be determined, that is, the probability that the account object is a target object, thereby indirectly achieving target object identification. This method solves the target object identification problem by decomposing the technical problem and achieves high recognition accuracy. BRIEF DESCRIPTION OF THE DRAWINGS

[0022] In order to more clearly illustrate the technical solutions in the embodiments of the present application, the following briefly introduces the drawings required for use in the description of the embodiments. Obviously, the drawings described below are only some embodiments of the present application. For ordinary technicians in this field, other drawings can be obtained based on these drawings without any creative work.

[0023] Figure 1 This is a schematic diagram of the relationship between account objects in a game scenario provided by an embodiment of the present application;

[0024] Figure 2 is a schematic diagram of an application program operating environment provided by an embodiment of the present application;

[0025] Figure 3 is a flowchart of an object recognition method provided by an embodiment of the present application;

[0026] Figure 4 This is a flowchart of a method for identifying associated account objects provided by an embodiment of the present application;

[0027] Figure 5 This is a flow chart of a first probability determination method provided by an embodiment of the present application;

[0028] Figure 6 This is a neural network architecture diagram provided by one embodiment of the present application;

[0029] Figure 7 This is a schematic diagram of an interface provided by an embodiment of the present application;

[0030] Figure 8 is a block diagram of an object recognition device provided by one embodiment of the present application;

[0031] Figure 9 This is a structural block diagram of a computer device provided in one embodiment of the present application. DETAILED DESCRIPTION

[0032] Before introducing the method embodiments provided in the present application, a brief introduction is first given to the relevant terms or nouns that may be involved in the method embodiments of the present application to facilitate understanding by those skilled in the art in the field of the present application.

[0033] Cloud technology refers to a hosting technology that unifies hardware, software, and network resources within a wide area network (WAN) or local area network (LAN) to enable data computing, storage, processing, and sharing. Cloud technology is a general term for network, information technology, integration technology, management platform technology, and application technology, all based on the cloud computing business model. It can form a resource pool that can be used on demand with flexibility and convenience. Cloud computing technology will become a crucial support. Backend services for technical network systems, such as video websites, image websites, and more portals, require extensive computing and storage resources. With the rapid development and application of the internet industry, every item will likely have its own unique identification mark and will need to be transmitted to backend systems for logical processing. Data of varying levels will be processed separately, and data from all industries will require a strong system backend, which can only be achieved through cloud computing.

[0034] Account object: An abstract representation of a user's identity in the computer world when using an application.

[0035] Virtual resource recommendation: The main purpose of virtual resource recommendation is to recommend virtual resources to account objects within an application, thereby increasing the probability that the account objects can perform certain operations within the application.

[0036] Before describing the embodiments of the present application in detail, the relevant technical background related to the embodiments of the present application is introduced to facilitate understanding by those skilled in the art in the art of the present application.

[0037] With the continuous development and popularization of Internet technology, many applications increase their contact with account objects through relevant application recommendation channels, promoting the probability of interaction between account objects and applications. In the embodiments of this application, platform applications are used to represent the applications corresponding to the application recommendation channels, and application objects are used to represent applications displayed or recommended in the platform application. The role of the platform application is to build an interaction channel between application objects and account objects. The platform application can be understood as an application community product that can display various application objects.

[0038] For a platform application, the first registration of an account object with an application object is considered the completion of a coloring operation for that account object, and the account object becomes a colored account object with respect to both the platform application and the application object. For example, if the platform application is a gaming community product, the platform application can provide gaming community features for account objects, such as displaying various gaming applications. If an account object registers a gaming application for the first time within the platform application, the account object becomes a colored account object.

[0039] Different dyeing account objects may perform different operations in the application object, and the depth of interaction with the application object and the viscosity relative to the application object are also different. Taking the application object as a game application as an example, each dyeing account object can perform various preset operations in the application object, such as payment operations or certain game-related operations. The degree to which different dyeing account objects perform the preset operations is different. For example, some dyeing account objects can frequently perform payment operations until the accumulated payment amount is greater than a certain payment threshold, or some dyeing account objects can frequently perform certain game-related operations until a certain operation frequency threshold is reached. These dyeing account objects that have a high degree of execution for certain specific preset operations and are high enough to meet a certain requirement of the application object for the preset operation are high-value objects for the game application, and are also the objects of key services and attention of the application object and even the platform application. In the embodiment of the present application, such dyeing account objects are referred to as target objects. Please refer to Figure 1, which shows a schematic diagram of the relationship between account objects in the game scenario in this embodiment of the present application. When an account object logs into a platform application and completes registration for a game application within the platform application, the account object becomes a dyed account object. After the dyed account object performs a payment operation such as topping up the game, the dyed account object becomes a paid account object. After the paid account object's cumulative payment exceeds a certain payment threshold, the paid account object becomes a target object.

[0040] Undoubtedly, predictive identification of target objects is extremely important for platform applications. When an account object accesses the platform application, quickly and accurately identifying whether the account object can become the target object is very important for promoting information interaction between the account object and the platform application, as well as improving the community interaction effect of the platform application.

[0041] For example, consider a scenario where the platform application is a game community product, the application object is the game application software, and the target object is a dyed account object whose cumulative game payment amount exceeds a certain payment threshold. The platform application in which the account object registers the game—that is, the game community product in which the dyeing operation is completed—is crucial to the revenue of the game community product. This is because the amount of money the dyed account object subsequently pays in-game can be allocated proportionally to the game community product that completed the dyeing operation. In other words, if a platform application completes the dyeing operation on an account object, making it a dyed account object, and if the dyed account object has a high payment amount for the corresponding game, the platform application that completed the dyeing operation can also generate significant revenue. Therefore, attracting dyed account objects with high payment amounts, namely, target objects, is crucial for both the game community product (platform application) and its affiliated games (application objects). The prerequisite for attracting dyed account objects with high payment amounts is to accurately identify account objects with the potential to become dyed account objects with high payment amounts among the account objects accessing the platform application, namely, identifying the target objects.

[0042] However, when an account object accesses a platform application, the platform application's prediction accuracy for whether the account object has the potential to become a designated account object, or even a target object, is insufficient. Consequently, it is difficult to recommend virtual resources to such potential target account objects, promote the conversion of such account objects to target objects through targeted, high-quality virtual resource recommendations, and improve the conversion rate of target objects. This results in related art failing to effectively promote interaction between the account object and the platform application, as well as between the account object and related application objects, when the account object accesses the platform application.

[0043] In order to promote the interaction between account objects and platform applications and application objects related to the platform applications, enhance the viscosity of account objects, and promote the transformation of account objects into target objects, an embodiment of the present application provides an object recognition method. The object recognition method can be based on the Bayesian principle and convert the recognition problem of colored account objects that meet preset requirements when performing preset operations, that is, the recognition problem of target objects, into two sub-recognition problems, which are Problem 1 and Problem 2 respectively.

[0044] Problem 1 involves identifying associated account objects. Associated account objects are objects that are stained with platform applications and perform preset operations on application objects. The identification result of Problem 1 is the probability that an account object is an associated account object. Problem 2 involves identifying whether an account object can become a key account object. A key account object is an account object that performs preset operations that meet preset requirements, or can be simply understood as a head account object. The identification result of Problem 2 is the probability that the account object is a head account object. Based on the identification results of Problems 1 and 2, the probability that an account object stained with a platform application and performing preset operations on an application object meets the preset requirements can be determined, that is, the probability that the account object is a target object, thereby indirectly achieving target object identification. This method solves the target object identification problem by decomposing the technical problem and achieves high recognition accuracy.

[0045] In order to make the objectives, technical solutions and advantages of this application clearer, the implementation methods of this application will be further described in detail below with reference to the accompanying drawings.

[0046] Before describing, it should be pointed out that all the data used in the examples of this application have been desensitized. The data after that is used legally and has been fully authorized by the relevant parties.

[0047] Please refer to Figure 2 , which shows a schematic diagram of an application running environment provided by an embodiment of the present application. The application running environment may include: a terminal 10 and a server 20.

[0048] The terminal 10 includes, but is not limited to, electronic devices such as mobile phones, computers, intelligent voice interaction devices, smart home appliances, car terminals, game consoles, e-book readers, multimedia playback devices, wearable devices, etc. The terminal 10 may be installed with a client of an application.

[0049] In an embodiment of the present application, the above-mentioned application can be any application that provides or uses object recognition services. Typically, the application can be a software community application or a software platform application. Various application objects can be associated with the software community application or the software platform application, such as game applications, news applications, social applications, interactive entertainment applications, browser applications, shopping applications, content sharing applications, virtual reality (VR) applications, augmented reality (AR) applications, application store applications, etc., which are not limited in the embodiment of the present application. The embodiment of the present application is not limited in this. Optionally, a client of the above-mentioned application is running in the terminal 10.

[0050] The server 20 is used to provide background services for the client of the application in the terminal 10. For example, the server 20 can be the background server of the above-mentioned application. The server 20 can be a single server or a server cluster composed of multiple physical servers, that is, any server in a distributed cluster. The distributed cluster can respond to requests from multiple clients 10 concurrently. It can also be a cloud server that provides basic cloud computing services such as cloud services, cloud databases, cloud computing, cloud functions, cloud storage, network services, cloud communications, middleware services, domain name services, security services, CDN (Content Delivery Network), and big data and artificial intelligence platforms.

[0051] Optionally, the terminal 10 and the server 20 may communicate with each other via a network 30. The terminal 10 and the server 20 may be directly or indirectly connected via wired or wireless communication, which is not limited in this application.

[0052] Please refer to Figure 3 , which shows a flow chart of an object recognition method provided by an embodiment of the present application. The method can be applied to a computer device, which refers to an electronic device with data calculation and processing capabilities. For example, the execution subject of each step can be Figure 2 Any terminal 10 or server 20 in the application running environment shown. The method may include the following steps:

[0053] S301. Obtain basic feature data corresponding to the account object.

[0054] The account object in the embodiments of this application refers to an account object that has not been colored by a platform application, or an account object that has not been registered by an application object in the platform application. This application does not limit the content of the basic feature data, for example, it can include object attribute data and / or object target records.

[0055] Object attribute data can be understood as characteristics that describe the static profile of the account object itself. This is related to the specific platform application, which refers to software community applications or software platform applications. Object attribute data refers to the data used to describe the account object within the platform application. Object attribute data is the existing data within the platform application. Taking the platform application as a gaming community platform as an example, object attribute data may include the account object's own attributes (for example, the account object's network address, contact information, etc.), the account object's login terminal, the account object's identity information, the account object's activity level, the account object's reservation information (such as the number of historical game reservations), the operating system, and so on.

[0056] Object target records refer to the records generated by the interaction between account objects and platform applications. Taking the gaming community platform as an example, object target records may include at least one of the following:

[0057] The first category of features: number of weekly game login days in the past month, the growth rate of monthly login days compared to the previous month, weekly game online time in the past month, the growth rate of monthly online time compared to the previous month, the number of weekly game logins in the past month, the growth rate of monthly game logins compared to the previous month, the increase rate of the average daily login days on weekends compared to the average daily login days on weekdays in the past month, and the proportion of login time in each time period of each week in the past month (working hours, noon, after get off work, early morning).

[0058] The second type of features: the number of participations in the non-deleted game activities in the past month, the length of stay in the game activities, the number of gift packs received in each level of game activities, the number of coupons received in each level, the number of tasks completed, etc.

[0059] Coloring features: refers to features related to registration operation prediction, such as the number of historically colored games, the number of historically active games, the number of visits to the game area, the length of time spent in the game area, whether there is a reservation, etc.

[0060] In some embodiments, the original basic feature data may be subjected to at least one of the following preprocessing steps to obtain preprocessed basic feature data, which can be used for prediction of various probabilities below.

[0061] The embodiments of the present application do not limit the preprocessing method. For example, at least one of the following processing methods can be adopted: missing value processing, feature engineering processing.

[0062] Missing value processing: The general method for processing missing values ​​is to delete missing value records or fill them with specified values. If data integrity is considered, missing values ​​can be filled with the average value of the data.

[0063] Feature engineering processing includes at least one of the following operations: discretization and normalization.

[0064] Discretization: The purpose of discretization is to increase generalization ability. For example, rank data can be discretized using the equidistant interval method. The specific operation is to group the rank data according to the principle of the same grouping interval, and assign the same rank characteristics to the ranks that fall into the same group.

[0065] Normalization: The normalization calculation formula is as follows. The purpose of normalization is to speed up the output of calculation results.

[0066] x'=(x-X_min) / (X_max-X_min), where x represents the original data, X_min represents the minimum value of the original data, X_max represents the maximum value of the original data, and x' represents the normalized value.

[0067] S302. Determine a first probability based on the basic characteristic data, where the first probability is the probability that the account object becomes a dyed account object and performs a preset operation on the application object in the platform application. The dyed account object refers to an account object that performs a registration operation on the application object in the platform application.

[0068] The account object obtained in step S301 has not been registered with the application object in the platform application and has not yet been colored for the application object in the platform application. The first probability refers to the probability that the account object belongs to an associated account object. The associated account object is the colored account object that performs a preset operation on the application object in the platform application. Therefore, step S302 actually solves the first sub-problem involved in the technical problem of target object identification: the problem of identifying associated account objects. The embodiments of this application do not limit the method for identifying associated account objects. For example, neural network technology can be used to directly identify associated account objects based on the basic feature data of the account object.

[0069] S303. Determine a second probability based on the basic characteristic data, where the second probability indicates a probability that the account object becomes a key account object, where the key account object is an account that satisfies the preset requirements when performing the preset operation.

[0070] The second probability refers to the probability of whether the account object is a key account object. The key account object is simply the head account object. The head account object is an account that meets the preset requirements when performing the above-mentioned preset operation. Taking the preset operation as a payment operation as an example, the head account object refers to an account with a higher payment limit. Of course, the embodiment of the present application does not limit the specific content of the preset requirements and the measurement criteria for the payment limit, and does not constitute an implementation obstacle. The head account object can be a colored account object or an uncolored account object. It can pay for the application object of the platform application or not. The head account object has the right to choose. Step S303 actually solves the second sub-problem involved in the technical problem of target object identification - the problem of head account object identification. The embodiment of the present application does not limit the head account object identification method. For example, the head account object can be directly identified using neural network technology based on the basic feature data of the account object.

[0071] In one embodiment, the second probability is determined based on the basic feature data, including: extracting first feature data from the basic feature data, where the first feature data is data related to the preset operation; inputting the first feature data into a first recognition model to obtain the second probability, where the first recognition model is trained based on the historical first feature data of the historical account object and the first label of the historical account object, where the first label represents whether the preset operation of the corresponding historical account object meets the preset requirements.

[0072] In the embodiment of the present application, the first feature data can be understood as the data related to the preset operation in the above-mentioned basic feature data. Taking the payment operation as an example, the first feature data is the data related to the payment operation in the above-mentioned basic feature data. In fact, it is data useful for predicting the payment ability. The embodiment of the present application does not limit the acquisition method and specific content of the first feature data. It can be screened from the basic feature data and does not constitute an implementation obstacle.

[0073] The meanings of the historical account object and historical first feature data are similar to those of the account object and first feature data, respectively, except for their timing. The historical account object and historical first feature data are already known data before executing S303 and serve as training data for the first recognition model. The first label is a label indicating whether the corresponding historical account object is a head account object. The model structure and training method of the first recognition model are not limited in this embodiment and do not constitute an implementation obstacle.

[0074] S304. Determine a target probability based on the first probability and the second probability. The target probability indicates the probability that the account object becomes a target object. The target object refers to a colored account object that meets the preset requirements when the preset operation is performed on the application object.

[0075] Taking the preset operation as a paid operation as an example, the target object refers to the object that is dyed through the platform application and pays a higher fee to the application object. Based on the product of the above-mentioned first probability and the above-mentioned second probability, the above-mentioned target probability can be determined, thereby obtaining the probability that the account object becomes the target object. In the embodiment of the present application, the technical problem of identifying the target object is converted into two sub-identification problems, which are Problem 1 and Problem 2. Problem 1 is the problem of identifying associated account objects, which is solved in step S302, and Problem 2 is the problem of identifying head account objects, which is solved in step S303. According to the respective identification results of Problem 1 and Problem 2, the probability of the account object becoming the target object can be determined, thereby indirectly realizing the identification of the target object. This target object identification method solves the technical problem in the related art that it is difficult to accurately identify the target object.

[0076] In some embodiments, when the target probability is greater than a preset threshold, virtual resources are sent to the account object. The present application embodiment does not limit the preset threshold, and it can be set according to actual conditions. Taking the preset operation as a payment operation as an example, the target probability is large, which means that the account object is likely to be stained by the platform application and pay a lot for the application object therein. In this case, the account object is a high-value customer for both the platform application and the application object. By sending virtual resources to high-value customers, the transformation of the account object to the target object can be promoted, and the interaction between the account object and the platform application and application object can be promoted, thereby improving the efficiency of interaction, expanding the depth of interaction, and thus improving the quality of interaction.

[0077] In one embodiment, please refer to Figure 4 , which shows a flow chart of the method for identifying associated account objects provided by an embodiment of the present application. The above-mentioned determination of the first probability based on the above-mentioned basic feature data includes:

[0078] S401. Determine a first probability component based on the basic feature data, where the first probability component is the probability that the account object is the key account object and performs a preset operation on the application object.

[0079] Regardless of whether an account object is a head account object, or in other words, a key account object, it can choose to perform or not perform a preset operation on an application object in a platform application. The first probability component quantifies the head account object's ability to perform the preset operation on an application object in the platform application. For example, if the preset operation is a payment operation, the first probability component quantifies the probability that an account object with strong payment capabilities will pay for the application object in the platform application.

[0080] If r=1,n u =1 respectively represent the two conditions that the account object is the head account object and the account object performs the payment operation, then step S401 predicts γ u =p(r=1,n u =1), that is, the probability that both conditions are met at the same time. γ u This is the first probability component.

[0081] The embodiment of the present application does not limit the method for determining the first probability component. For example, the first probability component can be directly output using neural network technology based on the basic feature data of the account object.

[0082] In one embodiment, the above-mentioned determination of the first probability component based on the above-mentioned basic feature data includes: inputting the above-mentioned basic feature data into a second recognition model to obtain the above-mentioned first probability component, the above-mentioned second recognition model is trained based on the historical basic feature data of the historical account object and the second label of the above-mentioned historical account object, the above-mentioned second label represents whether the corresponding historical account object meets the first requirement, and the above-mentioned first requirement is that the corresponding historical account object is a key account object and performs a preset operation on the above-mentioned application object.

[0083] The meanings of the historical account object and historical basic feature data are similar to those of the account object and basic feature data, respectively, except for their timing. The historical account object and historical basic feature data are already known data before executing S401 and serve as training data for the second recognition model. The second label indicates whether the corresponding historical account object meets the first requirement. The present embodiment does not limit the model structure and training method of the second recognition model and does not constitute an implementation obstacle.

[0084] S402. Determine a second probability component based on the basic feature data, the second probability component including a probability that the account object is not the key account object and a probability that the account object is the key account object but does not perform the preset operation on the application object;

[0085] Step S402 predicts the payment capability other than the situation corresponding to step S401.u =p(r=1,n u =0)+p(r=0,n u =1)+p(r=0,n u =0) = p(n u =0)+p(n u =1)*p(r=0|n u =1), where n u =0 means not to perform the preset operation, r=1 means it is a key account object (head account object), r=0 means it is not a key account object (head account object), n u =1 means executing the preset operation.

[0086] The embodiment of the present application does not limit the method for determining the second probability component. For example, the second probability component can be directly output using neural network technology based on the basic feature data of the account object.

[0087] In one embodiment, the second probability component is determined based on the basic feature data, including: inputting the basic feature data into a third recognition model to obtain the second probability component, the third recognition model is trained based on the historical basic feature data of the historical account object and the third label of the historical account object, the third label represents whether the corresponding historical account object meets the second requirement, and the second requirement is that the corresponding historical account object is not the key account object or is the key account object but does not perform the preset operation.

[0088] The meanings of the historical account object and historical basic feature data are similar to those of the account object and basic feature data, respectively, except for their timing. The historical account object and historical basic feature data are already known data before executing S402 and serve as training data for the second recognition model. The third label indicates whether the corresponding historical account object meets the second requirement. The model structure and training method of the third recognition model are not limited in this embodiment and do not constitute an obstacle to implementation.

[0089] S403. Determine the first probability based on the product of the first probability component and the second probability component.

[0090] Furthermore, in the process of solving problem 1, the embodiments of the present application can further determine the probability that the account object can be colored and perform a preset operation on the application object based on the Bayesian principle, according to a first probability component that quantifies the probability that the account object is a head account object and performs a preset operation on the application object in the platform application, and a second probability component that quantifies the preset operation capability of the account object in other situations. In other words, by flexibly based on the Bayesian principle, the problem of obtaining probabilities that are not easy to directly predict (the probability that the account object is colored and performs a preset operation on the application object) is decomposed into probabilities that are easy to predict, such as the problem of obtaining the first probability component and the second probability component, thereby solving the aforementioned problem of identifying associated account objects. By decomposing the problem, a higher recognition result is obtained, and the recognition accuracy of associated account objects is improved. The improvement in the recognition accuracy of associated account objects further promotes the improvement in the recognition accuracy of target objects.

[0091] Please refer to Figure 5 , which shows a flow chart of a method for determining a first probability in an embodiment of the present application. The method of determining the first probability based on the product of the first probability component and the second probability component includes:

[0092] S501. Determine a target base probability based on the basic feature data, where the target base probability indicates the probability of the account object becoming a dyed account object;

[0093] The target base probability quantifies how easily the account object is colored, that is, how easy it is to register the application object in the platform application. This embodiment of the application does not limit the method for determining the target base probability. For example, a neural network technology can be used to directly predict the basic feature data of the account object.

[0094] In one embodiment, second feature data is extracted from the above-mentioned basic feature data, and the above-mentioned second feature data is data related to the registration operation; the above-mentioned second feature data is input into the fourth recognition model to obtain the above-mentioned target basic probability, and the above-mentioned fourth recognition model is trained based on the historical second feature data of the historical account object and the fourth label of the above-mentioned historical account object, and the above-mentioned fourth label represents whether the corresponding historical account object is a dyed account object.

[0095] In the embodiments of this application, the second feature data can be understood as data related to the registration operation within the aforementioned basic feature data. This data is actually data useful for predicting registration ability, and may include, for example, the aforementioned staining characteristics. This embodiment of the application does not limit the method for obtaining or the specific content of the second feature data; it can be obtained from the basic feature data, and this does not constitute an implementation obstacle.

[0096] The meanings of the historical account object and the historical second feature data are similar to those of the account object and the second feature data, respectively, except for their timing. The historical account object and the historical first feature data are already known data before executing S501 and serve as training data for the fourth recognition model. The fourth label is a label indicating whether the corresponding historical account object is a colored account object. The present embodiment does not limit the model structure and training method of the fourth recognition model and does not constitute an obstacle to its implementation.

[0097] S502. Determine the first probability based on the product of the first probability component, the second probability component, and the target basic probability.

[0098] The first probability component and the second probability component respectively quantify the ability to perform preset operations on the application object, and the target basic probability quantifies the coloring tendency. The first probability is obtained by multiplying the above-mentioned first probability component, the above-mentioned second probability component and the above-mentioned target basic probability. The first probability quantifies the probability of being colored and performing the preset operation. Taking the preset operation as a payment operation as an example, the first probability quantifies the probability of being colored and paying, that is, the probability that the account object is an associated account object, thereby realizing the identification of associated account objects.

[0099] In the embodiment of the present application, the first recognition model, the second recognition model, the third recognition model and the fourth recognition model can all be constructed based on a neural network and obtained by training the neural network. The architecture of the neural network used by the first recognition model, the second recognition model, the third recognition model and the fourth recognition model can be the same or different. In an exemplary embodiment, reference can be made to Figure 6 , which shows a neural network architecture diagram of an embodiment of the present application, and the neural network architecture can be used to construct a first recognition model, a second recognition model, a third recognition model and / or a fourth recognition model. The neural network architecture includes an embedding layer, a feature extraction layer and a prediction layer. The embedding layer converts the input data of the neural network into embedded features (Embedding). The Embedding of the neural network is a technology that converts high-order sparse features into low-dimensional dense feature vectors. The essence of Embedding is a mapping from semantic space to vector space, while maintaining the relationship of the original samples in the semantic space in the vector space as much as possible. After the embedded features are input into several feature extraction layers, the specific recognition task is completed through the prediction layer.

[0100] The embodiment of the present application further experiments on the technical effect of the target object identification method provided in the embodiment of the present application by means of AB experiments. AB experiment is a commonly used experimental method for evaluating the effect of a technical solution. The AB experiment is divided into an experimental group and a control group. The experimental group and the control group implement different technical solutions, and the AB experimental results are obtained by comparing the execution results of the technical solutions. In the embodiment of the present application, the control group screened target accounts from a large number of account objects based on prior experience, and a total of 500 account objects were screened. The experimental group used the technical solution provided in the embodiment of the present application to automatically screen target accounts from a large number of account objects and obtained 500 account objects. To ensure that other operational intervention measures are the same for the experimental group and the control group, the accuracy of the target accounts screened by the experimental group is 39% higher than that of the control group.

[0101] The embodiments of this application can be applied to platform applications, please refer to Figure 7 , which shows a schematic diagram of the interface provided by the embodiment of the present application. When the platform application detects that the target probability corresponding to the account object is greater than a preset threshold, the account object can be regarded as a potential high-value account, and the Figure 7 The interface diagram in the figure shows the virtual resources provided for the account object, which promotes the account object to register the application object and allocates virtual resources to it after successful registration, thereby achieving targeted target object conversion.

[0102] Please refer to Figure 8 , which shows a block diagram of an object recognition device provided by one embodiment of the present application. The device has the function of implementing the above-mentioned object recognition method. The above-mentioned function can be implemented by hardware or by hardware executing corresponding software. The device can be a computer device or can be set in a computer device. The above-mentioned device includes:

[0103] Basic data acquisition module 801, used to obtain basic feature data corresponding to the account object;

[0104] The object recognition module 802 is configured to perform the following operations:

[0105] Determining a first probability based on the basic feature data, the first probability being a probability that the account object becomes a dyed account object and performs a preset operation on the application object in the platform application, where the dyed account object refers to an account object that performs a registration operation on the application object in the platform application;

[0106] Determining a second probability based on the basic characteristic data, the second probability indicating a probability that the account object becomes a key account object, where the key account object is an account that satisfies the preset requirements when performing the preset operation;

[0107] Based on the above-mentioned first probability and the above-mentioned second probability, the target probability is determined, and the above-mentioned target probability indicates the probability that the above-mentioned account object becomes the target object, and the above-mentioned target object refers to the colored account object that meets the above-mentioned preset requirements when the above-mentioned preset operation is performed on the above-mentioned application object.

[0108] In one embodiment, the object recognition module 802 is configured to perform the following operations:

[0109] Determining a first probability component based on the basic feature data, where the first probability component is a probability that the account object is the key account object and performs a preset operation on the application object;

[0110] determining a second probability component based on the basic feature data, the second probability component including a probability that the account object is not the key account object and a probability that the account object is the key account object but does not perform the preset operation on the application object;

[0111] The first probability is determined based on the product of the first probability component and the second probability component.

[0112] In one embodiment, the object recognition module 802 is configured to perform the following operations:

[0113] Extracting first feature data from the basic feature data, where the first feature data is data related to the preset operation;

[0114] Inputting the first feature data into a first recognition model to obtain the second probability, wherein the first recognition model is trained based on the historical first feature data of the historical account object and a first label of the historical account object, wherein the first label indicates whether a preset operation of the corresponding historical account object meets the preset requirement;

[0115] The determining of the target probability according to the first probability and the second probability includes: determining the target probability based on the product of the first probability and the second probability.

[0116] In one embodiment, the object recognition module 802 is configured to perform the following operations:

[0117] Determine a target basic probability based on the basic feature data, where the target basic probability indicates a probability that the account object becomes a dyed account object;

[0118] The first probability is determined based on the product of the first probability component, the second probability component, and the target basic probability.

[0119] In one embodiment, the object recognition module 802 is configured to perform the following operations:

[0120] The above-mentioned basic feature data is input into the second recognition model to obtain the above-mentioned first probability component. The above-mentioned second recognition model is trained based on the historical basic feature data of the historical account object and the second label of the above-mentioned historical account object. The above-mentioned second label represents whether the corresponding historical account object meets the first requirement. The above-mentioned first requirement is that the corresponding historical account object is a key account object and performs a preset operation on the above-mentioned application object.

[0121] In one embodiment, the object recognition module 802 is configured to perform the following operations:

[0122] The above-mentioned basic feature data is input into the third recognition model to obtain the above-mentioned second probability component. The above-mentioned third recognition model is trained based on the historical basic feature data of the historical account object and the third label of the above-mentioned historical account object. The above-mentioned third label represents whether the corresponding historical account object meets the second requirement. The above-mentioned second requirement is that the corresponding historical account object is not the above-mentioned key account object or is the above-mentioned key account object but does not perform the above-mentioned preset operation.

[0123] In one embodiment, the object recognition module 802 is configured to perform the following operations:

[0124] Extracting second feature data from the basic feature data, where the second feature data is data related to the registration operation;

[0125] The above-mentioned second feature data is input into the fourth recognition model to obtain the above-mentioned target basic probability. The above-mentioned fourth recognition model is trained based on the historical second feature data of the historical account object and the fourth label of the above-mentioned historical account object. The above-mentioned fourth label represents whether the corresponding historical account object is a colored account object.

[0126] In one embodiment, the object recognition module 802 is configured to perform the following operations:

[0127] When the target probability is greater than a preset threshold, virtual resources are sent to the account object.

[0128] It should be noted that the apparatus provided in the above embodiments, when implementing its functions, is only illustrated by the division of the above functional modules. In actual applications, the above functions can be assigned to different functional modules as needed, that is, the internal structure of the device can be divided into different functional modules to complete all or part of the functions described above. In addition, the apparatus and method embodiments provided in the above embodiments are based on the same concept. The specific implementation process is detailed in the method embodiment and will not be repeated here.

[0129] Please refer to Figure 9 , which shows a block diagram of a computer device provided by an embodiment of the present application. The computer device may be a server for executing the above-mentioned object recognition method. Specifically:

[0130] Computer device 900 includes a central processing unit (CPU) 901, a system memory 904 including a random access memory (RAM) 902 and a read-only memory (ROM) 903, and a system bus 905 connecting system memory 904 and CPU 901. Computer device 900 also includes a basic input / output system (I / O system) 906 that facilitates information transfer between various components within the computer, and a mass storage device 907 for storing an operating system 913, application programs 914, and other program modules 915.

[0131] The basic input / output system 906 includes a display 908 for displaying information and an input device 909, such as a mouse and keyboard, for inputting information. The display 908 and the input device 909 are both connected to the central processing unit 901 via an input / output controller 190 connected to the system bus 905. The basic input / output system 906 may also include an input / output controller 190 for receiving and processing input from a variety of other devices, such as a keyboard, mouse, or electronic stylus. Similarly, the input / output controller 190 also provides output to a display screen, printer, or other types of output devices.

[0132] The mass storage device 907 is connected to the central processing unit 901 via a mass storage controller (not shown) connected to the system bus 905. The mass storage device 907 and its associated computer-readable media provide non-volatile storage for the computer device 900. In other words, the mass storage device 907 may include a computer-readable medium (not shown) such as a hard disk or a CD-ROM (Compact Disc Read-Only Memory) drive.

[0133] Without loss of generality, computer-readable media may include computer storage media and communication media. Computer storage media include volatile and non-volatile, removable and non-removable media implemented by any method or technology for storing information such as computer-readable instructions, data structures, program modules or other data. Computer storage media include RAM, ROM, EPROM (Erasable Programmable Read Only Memory), EEPROM (Electrically Erasable Programmable Read Only Memory), flash memory or other solid-state memory technology, CD-ROM, DVD (Digital Video Disc) or other optical storage, tape cassettes, magnetic tape, disk storage or other magnetic storage devices. Of course, those skilled in the art will appreciate that computer storage media are not limited to the above-mentioned ones. The above-mentioned system memory 904 and mass storage device 907 can be collectively referred to as memory.

[0134] According to various embodiments of the present application, the computer device 900 may also be connected to a remote computer on a network such as the Internet for operation. That is, the computer device 900 may be connected to a network 912 via a network interface unit 911 connected to the system bus 905, or the network interface unit 911 may be used to connect to other types of networks or remote computer systems (not shown).

[0135] The memory further includes a computer program, which is stored in the memory and configured to be executed by one or more processors to implement the object recognition method.

[0136] In an exemplary embodiment, a computer-readable storage medium is also provided, in which at least one instruction, at least one program, a code set or an instruction set is stored. When the at least one instruction, the at least one program, the code set or the instruction set is executed by a processor, the object recognition method is implemented.

[0137] Specifically, the object recognition method includes:

[0138] Get the basic feature data corresponding to the account object;

[0139] Determining a first probability based on the basic feature data, the first probability being a probability that the account object becomes a dyed account object and performs a preset operation on the application object in the platform application, where the dyed account object refers to an account object that performs a registration operation on the application object in the platform application;

[0140] Determining a second probability based on the basic characteristic data, the second probability indicating a probability that the account object becomes a key account object, where the key account object is an account that satisfies the preset requirements when performing the preset operation;

[0141] Based on the above-mentioned first probability and the above-mentioned second probability, the target probability is determined, and the above-mentioned target probability indicates the probability that the above-mentioned account object becomes the target object, and the above-mentioned target object refers to the colored account object that meets the above-mentioned preset requirements when the above-mentioned preset operation is performed on the above-mentioned application object.

[0142] In one embodiment, determining the first probability based on the basic feature data includes:

[0143] Determining a first probability component based on the basic feature data, where the first probability component is a probability that the account object is the key account object and performs a preset operation on the application object;

[0144] determining a second probability component based on the basic feature data, the second probability component including a probability that the account object is not the key account object and a probability that the account object is the key account object but does not perform the preset operation on the application object;

[0145] The first probability is determined based on the product of the first probability component and the second probability component.

[0146] In one embodiment, determining the second probability based on the basic feature data includes:

[0147] Extracting first feature data from the basic feature data, where the first feature data is data related to the preset operation;

[0148] Inputting the first feature data into a first recognition model to obtain the second probability, wherein the first recognition model is trained based on the historical first feature data of the historical account object and a first label of the historical account object, wherein the first label indicates whether a preset operation of the corresponding historical account object meets the preset requirement;

[0149] The determining of the target probability according to the first probability and the second probability includes: determining the target probability based on the product of the first probability and the second probability.

[0150] In one embodiment, determining the first probability based on the product of the first probability component and the second probability component includes:

[0151] Determine a target basic probability based on the basic feature data, where the target basic probability indicates a probability that the account object becomes a dyed account object;

[0152] The first probability is determined based on the product of the first probability component, the second probability component, and the target basic probability.

[0153] In one embodiment, determining the first probability component based on the basic feature data includes:

[0154] The above-mentioned basic feature data is input into the second recognition model to obtain the above-mentioned first probability component. The above-mentioned second recognition model is trained based on the historical basic feature data of the historical account object and the second label of the above-mentioned historical account object. The above-mentioned second label represents whether the corresponding historical account object meets the first requirement. The above-mentioned first requirement is that the corresponding historical account object is a key account object and performs a preset operation on the above-mentioned application object.

[0155] In one embodiment, determining the second probability component based on the basic feature data includes:

[0156] The above-mentioned basic feature data is input into the third recognition model to obtain the above-mentioned second probability component. The above-mentioned third recognition model is trained based on the historical basic feature data of the historical account object and the third label of the above-mentioned historical account object. The above-mentioned third label represents whether the corresponding historical account object meets the second requirement. The above-mentioned second requirement is that the corresponding historical account object is not the above-mentioned key account object or is the above-mentioned key account object but does not perform the above-mentioned preset operation.

[0157] In one embodiment, determining the target basic probability based on the basic feature data includes:

[0158] Extracting second feature data from the basic feature data, where the second feature data is data related to the registration operation;

[0159] The above-mentioned second feature data is input into the fourth recognition model to obtain the above-mentioned target basic probability. The above-mentioned fourth recognition model is trained based on the historical second feature data of the historical account object and the fourth label of the above-mentioned historical account object. The above-mentioned fourth label represents whether the corresponding historical account object is a colored account object.

[0160] In one embodiment, the above method further comprises:

[0161] When the target probability is greater than a preset threshold, virtual resources are sent to the account object.

[0162] Optionally, the computer-readable storage medium may include: ROM (Read Only Memory), RAM (Random Access Memory), SSD (Solid State Drives), or an optical disk, etc. Among them, the random access memory may include ReRAM (Resistance Random Access Memory) and DRAM (Dynamic Random Access Memory).

[0163] In an exemplary embodiment, a computer program product or computer program is also provided, the computer program product or computer program including computer instructions stored in a computer-readable storage medium. A processor of a computer device reads the computer instructions from the computer-readable storage medium and executes the computer instructions, causing the computer device to perform the object recognition method described above.

[0164] It should be understood that the "multiple" mentioned in this article refers to two or more. "And / or" describes the association relationship of associated account objects, indicating that three relationships may exist. For example, A and / or B can represent three situations: A exists alone, A and B exist at the same time, and B exists alone. The character " / " generally indicates that the previous and subsequent associated account objects are in an "or" relationship. In addition, the step numbers described in this article only illustrate a possible execution sequence between the steps. In some other embodiments, the above steps may not be executed in the order of the numbers, such as two steps with different numbers are executed at the same time, or two steps with different numbers are executed in the opposite order of the diagram. The embodiments of this application are not limited to this.

[0165] In the embodiments of the present application, the term "module" or "unit" refers to a computer program or a part of a computer program that has a predetermined function and works together with other related parts to achieve a predetermined goal, and can be implemented in whole or in part by using software, hardware (such as processing circuits or memories) or a combination thereof. Similarly, a processor (or multiple processors or memories) can be used to implement one or more modules or units. In addition, each module or unit can be part of an overall module or unit that includes the function of the module or unit.

[0166] In addition, in the specific implementation of this application, when the above embodiments of this application are applied to specific products or technologies, data related to object information, etc., must be obtained from the object for permission or consent, and the collection, use and processing of relevant data must comply with relevant laws, regulations and standards of relevant countries and regions.

[0167] The above are merely exemplary embodiments of the present application and are not intended to limit the present application. Any modifications, equivalent replacements, improvements, etc. made within the spirit and principles of the present application shall be included in the scope of protection of the present application.

Claims

1. A method for object recognition, characterized in that: The method comprises: Get the basic feature data corresponding to the account object; determining, based on the basic feature data, a first probability, the first probability being a probability that the account object becomes a dyed account object and performs a preset operation on an application object in the platform application, the dyed account object being an account object that performs a registration operation on the application object in the platform application; determining a second probability based on the basic feature data, where the second probability indicates a probability that the account object becomes a key account object, where the key account object is an account that satisfies the preset requirement when performing the preset operation; A target probability is determined based on the first probability and the second probability, where the target probability indicates a probability that the account object becomes a target object, where the target object refers to a colored account object whose situation of performing the preset operation on the application object meets the preset requirements.

2. The method according to claim 1, characterized in that The determining of the first probability according to the basic feature data includes: determining, based on the basic feature data, a first probability component, where the first probability component is a probability that the account object is the key account object and performs a preset operation on the application object; determining a second probability component based on the basic feature data, the second probability component including a probability that the account object is not the key account object and a probability that the account object is the key account object but does not perform the preset operation on the application object; The first probability is determined based on a product of the first probability component and the second probability component.

3. The method according to claim 1 or 2, characterized in that The determining the second probability according to the basic feature data includes: Extracting first feature data from the basic feature data, where the first feature data is data related to the preset operation; Inputting the first feature data into a first recognition model to obtain the second probability, where the first recognition model is trained based on the historical first feature data of the historical account object and a first label of the historical account object, where the first label indicates whether a preset operation of the corresponding historical account object meets the preset requirement; Determining the target probability according to the first probability and the second probability includes: determining the target probability based on the product of the first probability and the second probability.

4. The method according to claim 2, characterized in that The determining the first probability based on the product of the first probability component and the second probability component includes: determining a target basic probability based on the basic feature data, the target basic probability indicating a probability that the account object becomes a dyed account object; The first probability is determined based on a product of the first probability component, the second probability component, and the target base probability.

5. The method according to claim 2, characterized in that The determining of the first probability component according to the basic feature data includes: The basic feature data is input into a second recognition model to obtain the first probability component. The second recognition model is trained based on the historical basic feature data of the historical account object and a second label of the historical account object. The second label represents whether the corresponding historical account object meets a first requirement. The first requirement is that the corresponding historical account object is a key account object and performs a preset operation on the application object.

6. The method according to claim 2, characterized in that The determining of the second probability component according to the basic feature data includes: The basic feature data is input into a third recognition model to obtain the second probability component. The third recognition model is trained based on the historical basic feature data of the historical account object and a third label of the historical account object. The third label represents whether the corresponding historical account object meets a second requirement. The second requirement is that the corresponding historical account object is not the key account object or is the key account object but does not perform the preset operation.

7. The method according to claim 4, characterized in that Determining the target basic probability according to the basic feature data includes: Extracting second feature data from the basic feature data, where the second feature data is data related to the registration operation; The second feature data is input into a fourth recognition model to obtain the target basic probability. The fourth recognition model is trained based on the historical second feature data of the historical account object and the fourth label of the historical account object. The fourth label represents whether the corresponding historical account object is a dyed account object.

8. The method according to claim 1, characterized in that The method further comprises: When the target probability is greater than a preset threshold, virtual resources are sent to the account object.

9. An object recognition device, characterized in that: The device comprises: Basic data acquisition module, used to obtain basic feature data corresponding to account objects; The object recognition module is used to perform the following operations: determining, based on the basic feature data, a first probability, the first probability being a probability that the account object becomes a dyed account object and performs a preset operation on an application object in the platform application, the dyed account object being an account object that performs a registration operation on the application object in the platform application; determining a second probability based on the basic feature data, where the second probability indicates a probability that the account object becomes a key account object, where the key account object is an account that satisfies the preset requirement when performing the preset operation; A target probability is determined based on the first probability and the second probability, where the target probability indicates a probability that the account object becomes a target object, where the target object refers to a colored account object whose situation of performing the preset operation on the application object meets the preset requirements.

10. A computer device, characterized in that: The computer device includes a processor and a memory, wherein the memory stores at least one instruction, at least one program, a code set, or an instruction set, and the at least one instruction, the at least one program, the code set, or the instruction set is loaded and executed by the processor to implement the object recognition method according to any one of claims 1 to 8.

11. A computer-readable storage medium, characterized in that The storage medium stores at least one instruction, at least one program, a code set, or an instruction set, and the at least one instruction, the at least one program, the code set, or the instruction set is loaded and executed by the processor to implement the object recognition method according to any one of claims 1 to 8.

12. A computer program product, characterized in that The computer program product includes computer instructions. A processor of a computer device reads the computer instructions, and the processor of the computer device executes the computer instructions to implement the object recognition method according to any one of claims 1 to 8.

Images