User classification information generation methods, devices, equipment, media, and program products

Abstract

This disclosure presents embodiments of a method, apparatus, device, medium, and program product for generating user classification information. One specific implementation of the method includes: generating real-time feature representations and offline feature representations of a target user; inputting the real-time and offline feature representations into a pre-trained user classification model to obtain first user classification information, the user classification model including a real-time classification sub-model, an offline classification sub-model, and a fusion classification sub-model; inputting the offline feature representation, the pseudo-real-time feature representation corresponding to the real-time feature representation, and the pseudo-offline feature representation corresponding to the offline feature representation into the user classification model to obtain second user classification information; and generating user classification information for the target user based on the first and second user classification information. This implementation is related to artificial intelligence and improves the accuracy of the generated user classification information.

Description

Technical Field

[0001] The embodiments disclosed herein relate to the field of computer technology, and specifically to user classification information generation methods, apparatus, devices, media, and program products. Background Technology

[0002] With the continuous development of internet technology, personalized display and targeted content recommendation for different users have become important mechanisms for various platforms. In this process, user classification information needs to be generated in real time for each user request. Generally, the generation of user classification information requires the simultaneous use of both offline and real-time user features. Specifically, offline and real-time features can be concatenated and then fed into a user classification model to obtain user classification information.

[0003] However, the inventors discovered that when generating user classification information using the above method, the following technical problems often arise:

[0004] The prediction accuracy is low.

[0005] The information disclosed in this background section is only intended to enhance the understanding of the background of the inventive concept, and therefore may contain information that does not constitute prior art known to those skilled in the art. Summary of the Invention

[0006] The summary portion of this disclosure is intended to provide a brief overview of the concepts, which will be described in detail in the detailed description portion. This summary portion is not intended to identify key or essential features of the claimed technical solutions, nor is it intended to limit the scope of the claimed technical solutions.

[0007] Some embodiments of this disclosure provide methods, apparatus, devices, computer-readable media, and program products for generating user classification information to address one or more of the technical problems mentioned in the background section above.

[0008] In a first aspect, some embodiments of this disclosure provide a method for generating user classification information. The method includes: generating real-time feature representations and offline feature representations of a target user; inputting the real-time feature representations and offline feature representations into a pre-trained user classification model to obtain first user classification information, wherein the user classification model includes a real-time classification sub-model, an offline classification sub-model, and a fusion classification sub-model; inputting the offline feature representations, pseudo-real-time feature representations corresponding to the real-time feature representations, and pseudo-offline feature representations corresponding to the offline feature representations into the user classification model to obtain second user classification information; and generating user classification information of the target user based on the first user classification information and the second user classification information.

[0009] Secondly, some embodiments of this disclosure provide a user classification information generation apparatus, the apparatus comprising: a generation unit configured to generate real-time feature representations and offline feature representations of a target user respectively; a first generation unit configured to input the real-time feature representations and offline feature representations into a pre-trained user classification model to obtain first user classification information, the user classification model including a real-time classification sub-model, an offline classification sub-model, and a fusion classification sub-model; a second generation unit configured to input the offline feature representations, pseudo-real-time feature representations corresponding to the real-time feature representations, and pseudo-offline feature representations corresponding to the offline feature representations into the user classification model to obtain second user classification information; and a third generation unit configured to generate user classification information of the target user based on the first user classification information and the second user classification information.

[0010] Thirdly, some embodiments of this disclosure provide an electronic device, including: one or more processors; and a storage device having one or more programs stored thereon, wherein when the one or more programs are executed by the one or more processors, the one or more processors implement the method described in any implementation of the first aspect above.

[0011] Fourthly, some embodiments of this disclosure provide a computer-readable medium having a computer program stored thereon, wherein the program, when executed by a processor, implements the method described in any of the implementations of the first aspect above.

[0012] Fifthly, some embodiments of this disclosure provide a computer program product, including a computer program that, when executed by a processor, implements the method described in any of the implementations of the first aspect above.

[0013] The embodiments disclosed above have the following beneficial effects: they improve the accuracy of the generated user classification information. Specifically, the reason why the prediction accuracy of related methods for generating user classification information is low is that, in practice, offline feature data and real-time feature data have different distributions. Offline feature data is often more concentrated. When prediction is made by directly concatenating real-time features and offline features, the model is more likely to identify spurious correlations between offline features and labels, resulting in offline feature bias. Real-time features are overwhelmed by offline feature bias, ultimately leading to low accuracy of the generated user classification information. Based on this, some embodiments of this disclosure introduce pseudo-real-time feature representation and pseudo-offline feature representation to remove offline feature bias and improve the accuracy of the generated user classification information. Attached Figure Description

[0014] The above and other features, advantages, and aspects of the embodiments of this disclosure will become more apparent from the accompanying drawings and the following detailed description. Throughout the drawings, the same or similar reference numerals denote the same or similar elements. It should be understood that the drawings are schematic, and elements are not necessarily drawn to scale.

[0015] Figures 1-2 This is a schematic diagram of an application scenario of a user classification information generation method according to some embodiments of the present disclosure;

[0016] Figure 3 This is a flowchart of some embodiments of the user classification information generation method according to the present disclosure;

[0017] Figure 4 This is a flowchart of some other embodiments of the user classification information generation method according to the present disclosure;

[0018] Figure 5 These are schematic diagrams illustrating the structure of some embodiments of the user classification information generation apparatus according to this disclosure;

[0019] Figure 6 This is a schematic diagram of the structure of an electronic device suitable for implementing some embodiments of the present disclosure. Detailed Implementation

[0020] Embodiments of this disclosure will now be described in more detail with reference to the accompanying drawings. While some embodiments of this disclosure are shown in the drawings, it should be understood that this disclosure can be implemented in various forms and should not be construed as limited to the embodiments set forth herein. Rather, these embodiments are provided to provide a more thorough and complete understanding of this disclosure. It should be understood that the accompanying drawings and embodiments of this disclosure are for illustrative purposes only and are not intended to limit the scope of protection of this disclosure.

[0021] It should also be noted that, for ease of description, only the parts relevant to the invention are shown in the accompanying drawings. Unless otherwise specified, the embodiments and features described in this disclosure can be combined with each other.

[0022] It should be noted that the concepts of "first" and "second" mentioned in this disclosure are used only to distinguish different devices, modules or units, and are not used to limit the order of functions performed by these devices, modules or units or their interdependencies.

[0023] It should be noted that the terms "a" and "a plurality of" used in this disclosure are illustrative rather than restrictive, and those skilled in the art should understand that, unless otherwise expressly indicated in the context, they should be understood as "one or more".

[0024] The names of messages or information exchanged between multiple devices in the embodiments of this disclosure are for illustrative purposes only and are not intended to limit the scope of such messages or information.

[0025] Before performing any of the operations involving the collection, storage, or use of user personal information disclosed in this disclosure, the relevant organizations or individuals have fulfilled their obligations, including conducting personal information security impact assessments, informing personal information subjects, and obtaining prior authorization and consent from personal information subjects, and have complied with the provisions of relevant laws and regulations.

[0026] This disclosure will now be described in detail with reference to the accompanying drawings and embodiments.

[0027] Figures 1-2 This is a schematic diagram illustrating an application scenario of a user classification information generation method according to some embodiments of the present disclosure.

[0028] like Figure 1 As shown, the execution entity of the user classification information generation method can obtain the real-time and offline features of the target user, where the target user can be the user currently requesting data. Furthermore, the real-time features include real-time categorical features and real-time numerical features, and the offline features include offline categorical features and offline numerical features. Based on this, feature representations of categorical features are generated through an embedding layer, and feature representations of numerical features are generated through a multilayer perceptron. Then, these are cross-fused through MLP layers to obtain real-time feature representations (i.e., real-time representation information) and offline feature representations (i.e., offline representation information).

[0029] Based on this, real-time and offline feature representations are input into a pre-trained user classification model to obtain the first user classification information. The user classification model includes a real-time classification sub-model, an offline classification sub-model, and a fusion classification sub-model. For example... Figure 1 As shown, the real-time classification sub-model includes a real-time classification layer, the offline classification sub-model includes an offline classification layer, and the fusion classification sub-model includes an MLP layer group and a fusion classification layer. The MLP layer group is used to fuse the real-time feature representation and the offline feature representation to obtain a fused feature representation (i.e., fused representation information).

[0030] Based on this, the aforementioned execution entity can input the offline feature representation, the pseudo-real-time feature representation corresponding to the real-time feature representation, and the pseudo-offline feature representation corresponding to the offline feature representation into the user classification model to obtain the second user classification information. As an example, such as... Figure 2In the process indicated by the dashed line and the "×" in the graphic, pseudo-real-time feature representation and pseudo-offline feature representation are used instead of the true feature representation. Specifically, the pseudo-real-time feature representation is input into the real-time classification layer, and the pseudo-real-time feature representation and pseudo-offline feature representation are input into the fusion classification layer; while the input to the offline classification layer remains the offline feature representation. Based on this, the results of the three classification layers are fused to obtain the second user classification information. Then, based on the first user classification information and the second user classification information, the user classification information of the target user is generated.

[0031] It is understandable that the execution entity of the user classification information generation method can be either hardware or software. When it is hardware, it can be various electronic devices with information processing capabilities, including but not limited to smartphones, tablets, e-book readers, laptops, and desktop computers. When it is software, it can be installed in the aforementioned electronic devices. It can be implemented as multiple software programs or software modules, for example, to provide distributed services, or it can be implemented as a single software program or software module. No specific limitations are made here.

[0032] Continue to refer to Figure 3 The flowchart 300 illustrates some embodiments of a user classification information generation method according to the present disclosure. The user classification information generation method includes the following steps:

[0033] Step 301: Generate real-time feature representations and offline feature representations for the target user, respectively.

[0034] In some embodiments, the entity executing the user classification information generation method can first generate real-time feature representations and offline feature representations of the target user. The real-time feature representation characterizes real-time features, and the offline feature representation characterizes offline features. For example, the real-time and offline features can be input into an embedding layer or a multilayer perceptron (MLP) to obtain the real-time and offline feature representations, respectively. Real-time features are features that change over time, including relevant features within a short timeframe up to the request time. The short timeframe is relative to the long timeframe corresponding to the offline features, typically on the order of hours or days, such as 24 hours. A real-time feature could be, for example, the number of times a user added a certain product category to their cart within 24 hours up to the request time. Offline features are features that generally do not change in real-time over time, including relevant features within a long timeframe up to the day before the request time. The long timeframe can be on the order of days or months, and an offline feature could be, for example, the average amount spent on additional purchases within a month.

[0035] Optionally, offline features can be categorized into categorical features or numerical features. For example, gender is a categorical feature, while the average spending amount over the past month is a numerical feature. Similarly, real-time features can also be categorical or numerical. In practice, categorical features can be generated using an embedding layer, while numerical features can be generated using an MLP layer. Based on this, the categorical and numerical feature representations are then fused using methods such as concatenation to generate either a real-time or offline feature representation.

[0036] In some optional implementations of certain embodiments, before generating the real-time feature representation and offline feature representation of the target user respectively, the method further includes: in response to receiving a data acquisition request from the target user, acquiring the target user's real-time features and offline features respectively; generating a real-time feature representation corresponding to the real-time features; and generating an offline feature representation corresponding to the offline features. Thus, when a user acquires content data, user classification information can be triggered to achieve personalized and targeted content display.

[0037] Step 302: Input the real-time feature representation and the offline feature representation into the pre-trained user classification model to obtain the first user classification information. The user classification model includes a real-time classification sub-model, an offline classification sub-model, and a fusion classification sub-model.

[0038] In some embodiments, the aforementioned execution entity can input real-time feature representations and offline feature representations into a pre-trained user classification model to obtain first user classification information. The user classification model includes a real-time classification sub-model, an offline classification sub-model, and a fusion classification sub-model. In practice, these three sub-models can be various classification networks, such as a Softmax layer, to generate corresponding sub-classification results based on the input data. Furthermore, the three sub-classification results are fused to obtain the first user classification information. Additionally, the fusion classification sub-model is used to fuse the input data and can include a multilayer perceptron (MLP) or a recurrent neural network for feature fusion.

[0039] In some optional implementations of certain embodiments, inputting real-time feature representations and offline feature representations into a pre-trained user classification model to obtain first user classification information includes: inputting real-time feature representations into a real-time classification sub-model to obtain a first real-time classification result; inputting offline feature representations into an offline classification sub-model to obtain a first offline classification result; inputting real-time feature representations and offline feature representations into a fusion classification sub-model to obtain a first fusion classification result; and fusing the first real-time classification result, the first fusion classification result, and the first offline classification result to obtain first user classification information.

[0040] Step 303: Input the offline feature representation, the pseudo-real-time feature representation corresponding to the real-time feature representation, and the pseudo-offline feature representation corresponding to the offline feature representation into the user classification model to obtain the second user classification information.

[0041] In some embodiments, the execution entity can acquire or generate pseudo-real-time feature representations corresponding to real-time feature representations and pseudo-real-time feature representations corresponding to offline feature representations. The pseudo-real-time feature representations are not actual real-time features; similarly, the pseudo-offline feature representations are not actual offline features. For example, the pseudo-real-time or pseudo-offline feature representations can be random numbers. In this process, by replacing actual real-time features with pseudo-real-time feature representations and replacing actual offline features with pseudo-real-time feature representations, the resulting second user classification information can represent the natural direct causal effect of the offline features.

[0042] In some embodiments, as an example, pseudo-real-time feature representations can be input into a real-time classification sub-model to obtain a second real-time classification result; pseudo-real-time feature representations and pseudo-offline feature representations can be input into a fusion classification sub-model to obtain a second fusion classification result; offline feature representations can be input into an offline classification sub-model to obtain a second offline classification result; the second real-time classification result, the second fusion classification result, and the second offline classification result can be fused to obtain second user classification information. As an example, fusion can be performed using a weighted summation method.

[0043] Step 304: Generate user classification information for the target user based on the first user classification information and the second user classification information.

[0044] In some embodiments, the aforementioned executing entity can perform a weighted fusion of the first user classification information and the second user classification information by weighted summation, for example, by taking the difference, to obtain the target user's classification information. Since the first user classification information is obtained through the target user's real-time feature representation and offline feature representation, it can represent the total causal effect of the target user's real-time and offline feature representations. The second user classification information can represent the natural direct causal effect of offline features; by taking the difference, the offline feature bias mentioned above can be removed, thereby improving the accuracy of the generated user classification information.

[0045] The method provided in some embodiments of this disclosure replaces real real-time features with pseudo-real-time feature representations and replaces real offline features with pseudo-real-time feature representations, so that the obtained second user classification information can represent the natural direct causal effect of offline features, thereby removing offline feature bias and improving the accuracy of the generated user classification information.

[0046] Further reference Figure 4This illustrates a flow 400 of another embodiment of the user classification information generation method. Flow 400 of this user classification information generation method includes the following steps:

[0047] Step 401: Generate real-time feature representation and offline feature representation for the target user, respectively.

[0048] Step 402: Input the real-time feature representation and the offline feature representation into the pre-trained user classification model to obtain the first user classification information. The user classification model includes a real-time classification sub-model, an offline classification sub-model, and a fusion classification sub-model.

[0049] In some embodiments, the specific implementation of steps 401-402 and the resulting technical effects can be found in [reference needed]. Figure 3 The corresponding implementation examples will not be described in detail here.

[0050] Step 403: Input the offline feature representation, the pseudo-real-time feature representation corresponding to the real-time feature representation, and the pseudo-offline feature representation corresponding to the offline feature representation into the user classification model to obtain the second user classification information.

[0051] In some embodiments, step 403 includes the following sub-steps 4031-4034:

[0052] Step 4031: Generate pseudo-real-time feature representations corresponding to real-time feature representations and pseudo-offline feature representations corresponding to offline feature representations.

[0053] In some embodiments, pseudo-real-time feature representations and pseudo-offline feature representations can be generated through the following steps:

[0054] Step 1: Obtain the training sample set. The training samples in the training sample set include real-time features, offline features, and user classification information.

[0055] Step 2: The mean of the real-time features corresponding to each training sample in the training sample set is determined as the pseudo-real-time feature representation.

[0056] Step 3: Determine the mean of the offline features corresponding to each training sample in the training sample set as the pseudo offline feature representation.

[0057] In some embodiments, by using the mean of each training sample in the training sample set to replace the real real-time features and offline features, the prediction process can remove the bias of offline features while increasing the credibility of pseudo-real-time feature representation and pseudo-offline feature representation, avoiding errors caused by the mismatch between random values ​​and the actual situation, thereby improving the accuracy of the generated user classification information.

[0058] Step 4032: Input the pseudo-real-time feature representation into the real-time classification sub-model to obtain the second real-time classification result.

[0059] Step 4033: Input the pseudo-real-time feature representation and pseudo-offline feature representation into the fusion classification sub-model to obtain the second fusion classification result.

[0060] Step 4034: Input the offline feature representation into the offline classification sub-model to obtain the second offline classification result.

[0061] Step 4035: The second real-time classification result, the second fused classification result, and the second offline classification result are merged to obtain the second user classification information.

[0062] In some embodiments, the second user classification information can be obtained by merging, for example, by splicing.

[0063] Step 404: Generate user classification information for the target user based on the first user classification information and the second user classification information.

[0064] In some embodiments, the aforementioned executing entity can perform a weighted fusion of the first user classification information and the second user classification information by weighted summation, for example, by taking the difference, to obtain the classification information of the target user.

[0065] from Figure 4 It can be seen from this that, with Figure 3 Compared to the description of some corresponding embodiments, Figure 4 In some corresponding embodiments, the user classification information generation method process 400 replaces the real real-time features and offline features with the mean of each training sample in the training sample set. This removes the offline feature bias during the prediction process and further improves the accuracy of the generated user classification information.

[0066] Further reference Figure 5 As an implementation of the methods shown in the above figures, this disclosure provides some embodiments of a user classification information generation device, which are similar to... Figure 3 Corresponding to the method embodiments shown, the device can be specifically applied to various electronic devices.

[0067] like Figure 5As shown, a user classification information generation apparatus 500 in some embodiments includes: a generation unit 501 configured to generate real-time feature representation and offline feature representation of a target user respectively; a first generation unit 502 configured to input the real-time feature representation and offline feature representation into a pre-trained user classification model to obtain first user classification information, wherein the user classification model includes a real-time classification sub-model, an offline classification sub-model, and a fusion classification sub-model; a second generation unit 503 configured to input the offline feature representation, the pseudo-real-time feature representation corresponding to the real-time feature representation, and the pseudo-offline feature representation corresponding to the offline feature representation into the user classification model to obtain second user classification information; and a third generation unit 504 configured to generate user classification information of the target user based on the first user classification information and the second user classification information.

[0068] In some optional implementations of embodiments, the second generation unit 503 is further configured to: generate pseudo-real-time feature representations corresponding to real-time feature representations and pseudo-offline feature representations corresponding to offline feature representations; input the pseudo-real-time feature representations into a real-time classification sub-model to obtain a second real-time classification result; input the pseudo-real-time feature representations and pseudo-offline feature representations into a fusion classification sub-model to obtain a second fusion classification result; input the offline feature representations into an offline classification sub-model to obtain a second offline classification result; and fuse the second real-time classification result, the second fusion classification result, and the second offline classification result to obtain second user classification information.

[0069] In some optional implementations of the embodiments, the second generation unit 503 is further configured to: acquire a training sample set, wherein the training samples in the training sample set include real-time features of the samples, offline features of the samples, and user classification information of the samples; determine the mean of the real-time features corresponding to each training sample in the training sample set as a pseudo-real-time feature representation; and determine the mean of the offline features corresponding to each training sample in the training sample set as a pseudo-offline feature representation.

[0070] In some optional implementations of embodiments, the first generation unit 502 is further configured to: input the real-time feature representation into a real-time classification sub-model to obtain a first real-time classification result; input the offline feature representation into an offline classification sub-model to obtain a first offline classification result; input the real-time feature representation and the offline feature representation into a fusion classification sub-model to obtain a first fusion classification result; and fuse the first real-time classification result, the first fusion classification result, and the first offline classification result to obtain first user classification information.

[0071] In some optional implementations of embodiments, the third generation unit 504 is further configured to determine the difference between the first user classification information and the second user classification information as the user classification information of the target user.

[0072] In some optional implementations of the embodiments, the user classification information generation device 500 further includes: an acquisition unit configured to: in response to receiving a data acquisition request from a target user, acquire the real-time features and offline features of the target user respectively.

[0073] It is understandable that the units described in the device 500 are related to the reference. Figure 3 The steps in the described method correspond accordingly. Therefore, the operations, features, and beneficial effects described above for the method also apply to the device 500 and the units contained therein, and will not be repeated here.

[0074] The following is for reference. Figure 6 It illustrates an electronic device 600 suitable for implementing some embodiments of the present disclosure (e.g., Figure 1 A structural diagram of the server or terminal device in the process. Figure 6 The electronic device shown is merely an example and should not be construed as limiting the functionality and scope of the embodiments of this disclosure.

[0075] like Figure 6 As shown, electronic device 600 may include processing unit 601 (e.g., central processing unit, graphics processor, etc.), which can perform various appropriate actions and processes according to a program stored in read-only memory (ROM) 602 or a program loaded from storage device 608 into random access memory (RAM) 603. RAM 603 also stores various programs and data required for the operation of electronic device 600. Processing unit 601, ROM 602, and RAM 603 are interconnected via bus 604. Input / output (I / O) interface 606 is also connected to bus 604.

[0076] Typically, the following devices can be connected to I / O interface 606: input devices 605 including, for example, touchscreens, touchpads, keyboards, mice, cameras, microphones, accelerometers, gyroscopes, etc.; output devices 607 including, for example, liquid crystal displays (LCDs), speakers, vibrators, etc.; storage devices 608 including, for example, magnetic tapes, hard disks, etc.; and communication devices 609. Communication device 609 allows electronic device 600 to communicate wirelessly or wiredly with other devices to exchange data. Although Figure 6 An electronic device 600 with various devices is shown; however, it should be understood that it is not required to implement or possess all of the devices shown. More or fewer devices may be implemented or possessed alternatively. Figure 6 Each box shown can represent a device or multiple devices as needed.

[0077] In particular, according to some embodiments of this disclosure, the processes described above with reference to the flowcharts can be implemented as computer software programs. For example, some embodiments of this disclosure include a computer program product comprising a computer program carried on a computer-readable medium, the computer program containing program code for performing the methods shown in the flowcharts. In such embodiments, the computer program can be downloaded and installed from a network via a communication device 609, or installed from a storage device 608, or installed from a ROM 602. When the computer program is executed by the processing device 601, it performs the functions defined above in the methods of some embodiments of this disclosure.

[0078] It should be noted that, in some embodiments of this disclosure, the computer-readable medium may be a computer-readable signal medium or a computer-readable storage medium, or any combination thereof. A computer-readable storage medium may be, for example,—but not limited to—an electrical, magnetic, optical, electromagnetic, infrared, or semiconductor system, apparatus, or device, or any combination thereof. More specific examples of a computer-readable storage medium may include, but are not limited to: an electrical connection having one or more wires, a portable computer disk, a hard disk, random access memory (RAM), read-only memory (ROM), erasable programmable read-only memory (EPROM or flash memory), optical fiber, portable compact disk read-only memory (CD-ROM), optical storage device, magnetic storage device, or any suitable combination thereof. In some embodiments of this disclosure, a computer-readable storage medium may be any tangible medium containing or storing a program that can be used by or in conjunction with an instruction execution system, apparatus, or device. In some embodiments of this disclosure, a computer-readable signal medium may include a data signal propagated in baseband or as part of a carrier wave, carrying computer-readable program code. Such propagated data signals may take various forms, including but not limited to electromagnetic signals, optical signals, or any suitable combination thereof. A computer-readable signal medium can be any computer-readable medium other than a computer-readable storage medium, which can send, propagate, or transmit a program for use by or in connection with an instruction execution system, apparatus, or device. The program code contained on the computer-readable medium can be transmitted using any suitable medium, including but not limited to: wires, optical fibers, RF (radio frequency), etc., or any suitable combination thereof.

[0079] In some implementations, clients and servers can communicate using any currently known or future-developed network protocol such as HTTP (Hypertext Transfer Protocol) and can interconnect with digital data communication (e.g., communication networks) of any form or medium. Examples of communication networks include local area networks (“LANs”), wide area networks (“WANs”), the Internet (e.g., the Internet of Things), and peer-to-peer networks (e.g., ad hoc peer-to-peer networks), as well as any currently known or future-developed networks.

[0080] The aforementioned computer-readable medium may be included in the aforementioned electronic device; or it may exist independently and not assembled into the electronic device. The aforementioned computer-readable medium carries one or more programs that, when executed by the electronic device, cause the electronic device to: generate real-time feature representations and offline feature representations of the target user; input the real-time feature representations and offline feature representations into a pre-trained user classification model to obtain first user classification information, the user classification model including a real-time classification sub-model, an offline classification sub-model, and a fusion classification sub-model; input the offline feature representations, the pseudo-real-time feature representations corresponding to the real-time feature representations, and the pseudo-offline feature representations corresponding to the offline feature representations into the user classification model to obtain second user classification information; and generate user classification information for the target user based on the first user classification information and the second user classification information.

[0081] Computer program code for performing operations of some embodiments of this disclosure can be written in one or more programming languages ​​or a combination thereof. Programming languages ​​include object-oriented programming languages—such as Java, Smalltalk, and C++—and conventional procedural programming languages—such as the "C" language or similar programming languages. The program code can be executed entirely on the user's computer, partially on the user's computer, as a standalone software package, partially on the user's computer and partially on a remote computer, or entirely on a remote computer or server. In cases involving remote computers, the remote computer can be connected to the user's computer via any type of network—including a local area network (LAN) or a wide area network (WAN)—or can be connected to an external computer (e.g., via the Internet using an Internet service provider).

[0082] The flowcharts and block diagrams in the accompanying drawings illustrate the architecture, functionality, and operation of possible implementations of systems, methods, and computer program products according to various embodiments of this disclosure. In this regard, each block in a flowchart or block diagram may represent a module, segment, or portion of code containing one or more executable instructions for implementing a specified logical function. It should also be noted that in some alternative implementations, the functions indicated in the blocks may occur in a different order than those indicated in the drawings. For example, two consecutively indicated blocks may actually be executed substantially in parallel, and they may sometimes be executed in reverse order, depending on the functions involved. It should also be noted that each block in the block diagrams and / or flowcharts, and combinations of blocks in the block diagrams and / or flowcharts, can be implemented using a dedicated hardware-based system that performs the specified function or operation, or using a combination of dedicated hardware and computer instructions.

[0083] The units described in some embodiments of this disclosure can be implemented in software or hardware. The described units can also be housed in a processor; for example, a processor may be described as including a generation unit, a first generation unit, a second generation unit, and a third generation unit. The names of these units do not necessarily limit the specific unit; for example, a generation unit may also be described as "a unit that generates real-time and offline feature representations of a target user."

[0084] The functions described above in this document can be performed, at least in part, by one or more hardware logic components. For example, exemplary types of hardware logic components that can be used, without limitation, include: Field Programmable Gate Arrays (FPGAs), Application-Specific Integrated Circuits (ASICs), Application Standard Products (ASSPs), System-on-Chip (SoCs), Complex Programmable Logic Devices (CPLDs), and so on.

[0085] Some embodiments of this disclosure also provide a computer program product, including a computer program that, when executed by a processor, implements any of the user classification information generation methods described above.

[0086] The above description is merely a selection of preferred embodiments of this disclosure and an explanation of the technical principles employed. Those skilled in the art should understand that the scope of the invention involved in the embodiments of this disclosure is not limited to technical solutions formed by specific combinations of the above-described technical features, but should also cover other technical solutions formed by arbitrary combinations of the above-described technical features or their equivalents without departing from the above-described inventive concept. For example, technical solutions formed by substituting the above-described features with (but not limited to) technical features with similar functions disclosed in the embodiments of this disclosure.

Claims

1. A method for generating user classification information, comprising: Generate real-time and offline feature representations for the target user, respectively; The real-time feature representation and the offline feature representation are input into a pre-trained user classification model to obtain first user classification information. The user classification model includes a real-time classification sub-model, an offline classification sub-model, and a fusion classification sub-model, and includes: inputting the real-time feature representation into the real-time classification sub-model to obtain a first real-time classification result; inputting the offline feature representation into the offline classification sub-model to obtain a first offline classification result; inputting the real-time feature representation and the offline feature representation into the fusion classification sub-model to obtain a first fusion classification result; and fusing the first real-time classification result, the first fusion classification result, and the first offline classification result to obtain first user classification information. The process of inputting the offline feature representation, the pseudo-real-time feature representation corresponding to the real-time feature representation, and the pseudo-offline feature representation corresponding to the offline feature representation into the user classification model to obtain second user classification information includes: inputting the pseudo-real-time feature representation into the real-time classification sub-model to obtain a second real-time classification result; inputting the pseudo-real-time feature representation and the pseudo-offline feature representation into the fusion classification sub-model to obtain a second fusion classification result; inputting the offline feature representation into the offline classification sub-model to obtain a second offline classification result; and fusing the second real-time classification result, the second fusion classification result, and the second offline classification result to obtain second user classification information. The pseudo-real-time feature representation is the mean of the real-time features corresponding to each training sample in the acquired training sample set, and the pseudo-offline feature representation is the mean of the offline features corresponding to each training sample in the acquired training sample set. Based on the first user classification information and the second user classification information, user classification information for the target user is generated.

2. The method according to claim 1, wherein, The pseudo-real-time feature representations corresponding to the real-time feature representations and the pseudo-offline feature representations corresponding to the offline feature representations are generated respectively.

3. The method according to claim 2, wherein, The step of generating pseudo-real-time feature representations corresponding to the real-time feature representations and pseudo-offline feature representations corresponding to the offline feature representations includes: Obtain a training sample set, wherein the training samples in the training sample set include real-time features of the samples, offline features of the samples, and user classification information of the samples; The mean of the real-time features corresponding to each training sample in the training sample set is determined as the pseudo-real-time feature representation. The mean of the offline features corresponding to each training sample in the training sample set is determined as the pseudo offline feature representation.

4. The method according to claim 1, wherein, The step of generating user classification information for the target user based on the first user classification information and the second user classification information includes: The difference between the first user classification information and the second user classification information is determined as the user classification information of the target user.

5. The method according to claim 1, wherein, Before generating the real-time and offline feature representations of the target user respectively, the method further includes: In response to receiving a data acquisition request from a target user, the real-time characteristics and offline characteristics of the target user are acquired respectively; and The generation of real-time and offline feature representations of the target user includes: Generate the real-time feature representation corresponding to the real-time feature, and generate the offline feature representation corresponding to the offline feature.

6. A user classification information generation device, comprising: The generation unit is configured to generate real-time and offline feature representations of the target user, respectively. A first generation unit is configured to input the real-time feature representation and the offline feature representation into a pre-trained user classification model to obtain first user classification information. The user classification model includes a real-time classification sub-model, an offline classification sub-model, and a fusion classification sub-model. The generation unit includes: inputting the real-time feature representation into the real-time classification sub-model to obtain a first real-time classification result; inputting the offline feature representation into the offline classification sub-model to obtain a first offline classification result; inputting the real-time feature representation and the offline feature representation into the fusion classification sub-model to obtain a first fusion classification result; and fusing the first real-time classification result, the first fusion classification result, and the first offline classification result to obtain the first user classification information. The second generation unit is configured to input the offline feature representation, the pseudo-real-time feature representation corresponding to the real-time feature representation, and the pseudo-offline feature representation corresponding to the offline feature representation into the user classification model to obtain second user classification information. This includes: inputting the pseudo-real-time feature representation into the real-time classification sub-model to obtain a second real-time classification result; inputting the pseudo-real-time feature representation and the pseudo-offline feature representation into the fusion classification sub-model to obtain a second fusion classification result; inputting the offline feature representation into the offline classification sub-model to obtain a second offline classification result; and fusing the second real-time classification result, the second fusion classification result, and the second offline classification result to obtain second user classification information. The pseudo-real-time feature representation is the mean of the real-time features corresponding to each training sample in the acquired training sample set, and the pseudo-offline feature representation is the mean of the offline features corresponding to each training sample in the acquired training sample set. The third generation unit is configured to generate user classification information for the target user based on the first user classification information and the second user classification information.

7. An electronic device, comprising: One or more processors; Storage device, on which one or more programs are stored, When the one or more programs are executed by the one or more processors, the one or more processors implement the method as described in any one of claims 1-5.

8. A computer-readable medium having a computer program stored thereon, wherein, When the computer program is executed by a processor, it implements the method as described in any one of claims 1-5.

9. A computer program product comprising a computer program that, when executed by a processor, implements the method according to any one of claims 1-5.

Images