Recommendation methods, systems, devices, and media for user-generated content
By constructing features based on user behavior data and categories, and training a user identification model using a weakly supervised method, the problems of low accuracy and high computational cost in high-quality content recommendation in online communities are solved, achieving low-cost and high-accuracy user-generated content recommendation.
Patent Information
- Application Number
- CN202111568980.1
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- Filing Date
- 2021-12-21
- Publication Date
- 2026-11-17
- Estimated Expiration
- 2041-12-21
AI Technical Summary
Existing technologies cannot effectively filter and recommend high-quality user-generated content in online communities, resulting in low recommendation quality and high computational cost. Furthermore, the model training data requires manual annotation, which is costly.
By constructing user features based on user behavior data and categories, a user identification model is trained using a weakly supervised method to determine the user's recommendation value, and content recommendations are made based on the user's recommendation value, thus reducing the reliance on content features.
It achieves high-quality content recommendation with low computational cost, reduces the cost of model training, improves recommendation accuracy, and can identify different levels of recommendation for users.
Smart Images

Figure CN114218492B_ABST
Abstract
Description
Technical Field
[0001] This invention relates to the field of machine learning technology, and in particular to methods, systems, devices, and media for recommending user-generated content. Background Technology
[0002] Currently, the content posted by users in online communities is generally of inconsistent quality, with low-quality, insincere sharing mixed together with high-quality UGC (User Generated Content). This makes it impossible to filter and push high-quality content to users on recommendation pages or search results pages.
[0003] To address this issue, common methods include content feature-based identification and detection, such as spam detection and rumor detection, to determine the quality of user-posted content for filtering and recommendation. However, even with these content-based identification and detection methods, the recommended content remains inconsistent in quality, and these methods are computationally intensive, requiring manually labeled training data for the model. Therefore, there is an urgent need for a content recommendation solution that is computationally efficient and does not require manually labeled training data. Summary of the Invention
[0004] The purpose of this invention is to provide a method, system, device, and medium for recommending user-generated content. This method can train a model to determine a user's recommendation score based on user behavior data and user category, and then use this model to recommend user-generated content. It requires minimal computation and eliminates the need for manually labeled training data. This addresses the technical problems of low recommendation quality, high computational cost, and high training data labeling costs associated with existing methods.
[0005] An embodiment of the present invention discloses a method for recommending user-generated content for electronic devices, the method comprising:
[0006] Determining user recommendation values based on user identification models, and
[0007] Recommend user-generated content based on the user's recommendation score;
[0008] The user identification model is obtained through a weakly supervised training method, which includes:
[0009] Obtain user information, including user categories and user behavior data;
[0010] The recommended value for a user is determined based on the user's category, and this recommended value includes multiple discrete numerical values.
[0011] Based on user behavior data, construct user characteristics;
[0012] The user's recommendation value and user features are combined to form a sample set, where each user's user features are used as sample features and the recommendation value is used as the sample label to form a sample.
[0013] Use this sample set to train the user identification model.
[0014] Optionally, recommending user-generated content based on the user's recommendation score includes:
[0015] Reduce the recommendation weight of user-generated content whose recommendation value is below the threshold.
[0016] Optionally, recommending user-generated content based on the user's recommendation score includes:
[0017] The recommendation weight of user-generated content is determined based on the recommendation value.
[0018] Optionally, a user's recommendation value is determined based on the user's category, and this recommendation value includes multiple discrete numerical values, including:
[0019] The first value is determined by assigning a referral value to a user marked as non-compliant, the second value is determined by assigning a referral value to a new user, and the third value is determined by assigning a referral value to other users, wherein the first value < the second value < the third value.
[0020] Optionally, the user identification model is a regression model of the predicted value between the first and third values.
[0021] Optionally, behavioral data includes publishing data, consumption data, and activity data. Publishing data includes the number of pieces of content published in the past week, whether the published content is plagiarized content, and whether the published content promotes sales. Activity data includes the active duration and search frequency in the past week.
[0022] An embodiment of the present invention discloses a recommendation system for user-generated content, including...
[0023] The determination module is used to determine the user's recommendation value based on the user identification model, and
[0024] The recommendation module is used to recommend user-generated content based on the user's recommendation score.
[0025] The training module is used to train the user recognition model based on a weakly supervised training method. This training module includes:
[0026] The acquisition unit is used to acquire user information, which includes the user's category and user behavior data.
[0027] The determining unit is used to determine the user's recommendation value based on the user's category. The recommendation value includes multiple discrete numerical values.
[0028] The building unit is used to construct user characteristics based on user behavior data;
[0029] The constituent unit is used to assemble a sample set from the user's recommendation value and user features, wherein each user's user features are used as sample features and the recommendation value is used as the sample label to form a sample;
[0030] Training unit, used to train the user recognition model using a sample set composed of constituent units.
[0031] An embodiment of the present invention discloses a user-generated content recommendation device, the device including a memory storing computer-executable instructions and a processor, wherein when the instructions are executed by the processor, the device performs any of the above-described user-generated content recommendation methods.
[0032] An embodiment of the present invention discloses a computer storage medium storing instructions that, when executed on a computer, cause the computer to perform any of the above-described methods for recommending user-generated content.
[0033] The main differences and effects of the embodiments of the present invention compared with the prior art are as follows:
[0034] In this invention, a user's recommendation value is determined based on a user identification model, and user-generated content is recommended based on that recommendation value. Recommendations are made based on the user, not on the content the user publishes. Recommendations based on published content are computationally intensive and have low accuracy. Recommendations based on the user, however, have lower model complexity and higher accuracy.
[0035] In this invention, a user's recommendation value is determined based on the user's category, and this recommendation value includes multiple discrete numerical values. User features are constructed based on the user's behavioral data. A sample set is formed by combining the user's recommendation value and user features, where each user's user features are used as sample features and the recommendation value is used as a sample label. This sample set is then used to train the user recognition model. A weakly supervised method is used to train the model, obtaining labels for the training data by identifying different levels of user recommendation, without requiring manual labeling of the training data. By constructing training data features based on user behavioral data, the trained model can directly predict the user's recommendation score based on these features.
[0036] In this invention, the recommendation value for users marked as non-compliant is determined as a first value, the recommendation value for new users is determined as a second value, and the recommendation value for other users is determined as a third value, wherein the first value < the second value < the third value. Based on the user category, the three levels of recommendation scores can be accurately identified as labels for the training data, enabling the training of a reliable user identification model in a weakly supervised manner without manually labeled training samples. Attached Figure Description
[0037] Figure 1 A schematic diagram illustrating an implementation scenario of a method for recommending user-generated content according to an embodiment of this application.
[0038] Figure 2A A flowchart illustrating a method for recommending user-generated content according to an embodiment of this application is provided.
[0039] Figure 2B A flowchart illustrating a method for training a user identification model according to an embodiment of this application is shown.
[0040] Figure 3A A schematic diagram illustrating a user classifier and user identification model according to an embodiment of this application is shown.
[0041] Figure 3B A schematic diagram of a user identification model according to an embodiment of this application is shown.
[0042] Figure 4 A block diagram of a user-generated content recommendation system according to an embodiment of this application is shown.
[0043] Figure 5 A block diagram illustrating a device for recommending user-generated content according to an embodiment of this application. Detailed Implementation
[0044] The present application will be further described below with reference to specific embodiments and accompanying drawings. It is to be understood that the specific embodiments described herein are merely for explaining the present application and not for limiting it. Furthermore, for ease of description, the accompanying drawings show only the parts relevant to the present application, and not all of the structures or processes. It should be noted that in this specification, similar reference numerals and letters denote similar items in the following drawings.
[0045] It should be understood that although the terms "first," "second," etc., may be used herein to describe various features, these features should not be limited by these terms. The use of these terms is merely for distinction and should not be construed as indicating or implying relative importance. For example, without departing from the scope of the exemplary embodiments, a first feature may be referred to as a second feature, and similarly, a second feature may be referred to as a first feature.
[0046] In the description of this application, it should also be noted that, unless otherwise explicitly specified and limited, the terms "set up," "connected," and "linked" should be interpreted broadly. For example, they can refer to a fixed connection, a detachable connection, or an integral connection; they can refer to a mechanical connection or an electrical connection; they can refer to a direct connection or an indirect connection through an intermediate medium; and they can refer to the internal connection of two components. Those skilled in the art can understand the specific meaning of the above terms in this embodiment based on the specific circumstances.
[0047] The illustrative embodiments of this application include, but are not limited to, methods, systems, devices, and media for recommending user-generated content.
[0048] Various aspects of the illustrative embodiments will be described using terminology commonly employed by those skilled in the art to convey the essence of their work to others skilled in the art. However, it will be apparent to those skilled in the art that some alternative embodiments will be practiced using the features partially described. Specific figures and configurations are set forth for purposes of explanation in order to provide a more thorough understanding of the illustrative embodiments. However, it will be apparent to those skilled in the art that alternative embodiments may be practiced without specific details. In some other instances, well-known features have been omitted or simplified herein to avoid obscuring the illustrative embodiments of this application.
[0049] Furthermore, the various operations will be described as multiple separate operations in a manner most conducive to understanding the illustrative embodiments; however, the order of description should not be construed as implying that these operations must depend on the order of description, and many of these operations may be performed in parallel, concurrently, or simultaneously. Moreover, the order of the operations may also be rearranged. The process may be terminated when the described operations are completed, but may also include additional steps not included in the figures. The process may correspond to a method, function, procedure, subroutine, subroutine, etc.
[0050] References to "an embodiment," "embodiment," "illustrative embodiment," etc., in this specification indicate that the described embodiment may include specific features, structures, or properties; however, each embodiment may or may not necessarily include specific features, structures, or properties. Furthermore, these phrases are not necessarily directed to the same embodiment. Additionally, when specific features are described in conjunction with specific embodiments, the knowledge of those skilled in the art can influence the combination of these features with other embodiments, whether or not those embodiments are explicitly described.
[0051] Unless the context otherwise specifies, the terms “comprising,” “having,” and “including” are synonyms. The phrase “A and / or B” means “(A), (B), or (A and B).”
[0052] As used herein, the term "module" may refer to, as part of, or include: a memory (shared, dedicated, or grouped), an application-specific integrated circuit (ASIC), electronic circuitry and / or a processor (shared, dedicated, or grouped), combinational logic circuitry, and / or other suitable components that provide the said functionality for running one or more software or firmware programs.
[0053] In the accompanying drawings, some structural or methodological features may be shown in a specific arrangement and / or order. However, it should be understood that such a specific arrangement and / or order is not necessary. Rather, in some embodiments, these features may be illustrated in a manner and / or order different from that shown in the illustrative drawings. Furthermore, the inclusion of structural or methodological features in a particular drawing does not mean that all embodiments need to include such features; in some embodiments, these features may be omitted or may be combined with other features.
[0054] To make the objectives, technical solutions, and advantages of this application clearer, the embodiments of this application will be described in further detail below with reference to the accompanying drawings.
[0055] Figure 1 A schematic diagram illustrating an implementation scenario of a method for recommending user-generated content according to an embodiment of this application.
[0056] like Figure 1 As shown, the scenario includes terminal 103, which can be a desktop terminal or a mobile terminal. The mobile terminal can be, but is not limited to, various personal computers, laptops, smartphones, tablets, and portable wearable devices. Terminal 103 can have an application installed that can access remote services, such as a browser or client, and can register accounts, browse and publish content on remote services through this browser or client. The application involved in this embodiment can be a software client, or a webpage, mini-program, or other client. If the application is a webpage, mini-program, or other client, the backend server is the backend server corresponding to the software, webpage, or mini-program, without limiting the specific type of client.
[0057] The scenario also includes server 101, which can be the backend server corresponding to the application installed on terminal 103. For example, it can be a server that provides network community services. It can be an independent physical server or a server cluster or distributed system composed of multiple servers. 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 communication, middleware services, domain name services, security services, CDN, and big data and artificial intelligence platforms, but it is not limited to these.
[0058] Terminal 103 and server 101 are connected via network 105. Network 105 may include one or more and may include various connection types, such as wired, wireless communication links, cloud or fiber optic cables, etc. Specific examples of the network mentioned above may include the Internet provided by the communication provider of terminal 103.
[0059] In online communities, the quality of user-generated content varies greatly. Some content is genuine, while others is insincere. For instance, content created from a user's own experience or creation can be considered genuine; conversely, content created by borrowing unauthorized material or reposting content without permission or attribution can be considered insincere. Insincere content creation may also include: receiving sponsorship or assistance from businesses during sharing or creation without disclosing this benefit; posting content to flaunt extraordinary spending power; posting unscientific content (especially pseudoscience in health advice and commercial promotion); posting excessively nude or sexually suggestive content; impersonating others; using personal influence to make accusations or post inflammatory content; using exaggeration or sensationalism to attract users; providing medical or investment advice without proper qualifications; and posting other content that disrespects the rights of others.
[0060] For online community operators, public traffic such as recommendations and search engines are crucial scenarios for reflecting community values. To ensure and enhance these values, it's necessary to reduce the visibility of low-quality (e.g., insincere) content in search and recommendation scenarios. One possible approach is to manually review user-posted content and assess its quality. Then, when recommending content to other users, the exposure is adjusted based on the content's quality. This achieves the desired goal; however, in scenarios with large online communities, a large number of users, and a high volume of posts, manually reviewing every post by every user would be extremely costly, if not impossible.
[0061] To address the above problems, embodiments of this application provide a method for recommending user-generated content in electronic devices, which uses a machine learning model to recommend user-generated content. As an alternative, the quality of user-posted content can be manually judged, a recommendation value label can be assigned to the content, and then the content's features can be extracted. The recommendation value label and the features can then be combined to form training samples to train a machine learning model. This machine learning model can then predict the recommendation value of the user-posted content based on its features, allowing for content recommendations based on the recommendation value. However, this approach requires manual labeling of training samples, resulting in high costs and low efficiency. Furthermore, due to the complexity of the posted content's features, the accuracy of predicting the recommendation value of posted content is limited. Embodiments of this application further improve upon this approach. Figure 2A A flowchart illustrating a method for recommending user-generated content according to an embodiment of this application is provided. Figure 2B A flowchart illustrating a method for training a user identification model according to an embodiment of this application is shown.
[0062] like Figure 2A As shown, the user-generated content recommendation method 200 includes:
[0063] Step 202, determine the user's recommendation value based on the user identification model, and
[0064] Step 204: Recommend user-generated content based on the user's recommendation score;
[0065] The user identification model is obtained through a weakly supervised training method, which includes:
[0066] Step 206: Obtain user information, which includes the user's category and user behavior data;
[0067] Step 208: Determine the user's recommendation value based on the user's category. The recommendation value includes multiple discrete numerical values.
[0068] Step 210: Construct user characteristics based on user behavior data;
[0069] Step 212: Combine the user's recommendation value and user features into a sample set, wherein each user's user features are used as sample features and the recommendation value is used as the sample label to form a sample.
[0070] Step 214: Use the sample set to train the user identification model.
[0071] According to some embodiments of this application, recommending user-generated content based on the user's recommendation value includes:
[0072] The recommendation weight of user-generated content is determined based on the recommendation value.
[0073] According to some embodiments of this application, recommending user-generated content based on the user's recommendation value includes: reducing the recommendation weight of user-generated content whose recommendation value is below a threshold.
[0074] According to some embodiments of this application, a user's recommendation value is determined based on the user's category, and the recommendation value includes multiple discrete numerical values, including:
[0075] The first value is determined by assigning a referral value to a user marked as non-compliant, the second value is determined by assigning a referral value to a new user, and the third value is determined by assigning a referral value to other users, wherein the first value < the second value < the third value.
[0076] According to some embodiments of this application, behavioral data includes publishing data, consumption data, and activity data. Publishing data includes the number of contents published in the past week, whether the published contents are copied content, and whether the published contents promote sales. Activity data includes the active duration and search frequency in the past week.
[0077] Figure 3A A schematic diagram illustrating a user classifier and a user identification model according to an embodiment of this application is shown. Figure 3B A schematic diagram of a user identification model according to an embodiment of this application is shown below. Figure 3A and 3B Here is an example illustrating the user-generated content recommendation method described above.
[0078] As mentioned earlier, the quality of content posted by users in online communities varies greatly; for example, some content is sincere, while some is not. Accordingly, users who post a higher proportion of sincere content or who are more likely to post insincere content can be considered to have higher sincerity levels. Conversely, users who post a lower proportion of sincere content or who are less likely to post insincere content can be considered to have lower sincerity levels. Based on this, machine learning models can be trained to predict a user's sincerity level based on their characteristics and then recommend user-generated content based on that sincerity level.
[0079] Typically, training machine learning models requires labeled training sets. However, in large online communities, the amount of user information and posted data is enormous, making manual collection and labeling of training sets extremely costly. Therefore, as... Figure 3AAs shown, embodiments of this application use a user classifier 301 to label the collected samples for use as a training set. Users are divided into three categories for labeling. The first category consists of users historically classified as level 0, meaning users manually flagged as non-compliant, such as those banned for repeatedly posting content violating online community rules. These users are considered highly insincere and are assigned a score of -100, representing low sincerity. The second category consists of new users, such as newly registered users or users who have not posted anything. Since it's impossible to determine their sincerity, they are assigned a score of 0, representing medium sincerity. The third category consists of UGC users who have posted content. These users are considered sincere and are assigned a score of 100, representing high sincerity. In reality, since most users in the online community are considered sincere, all users other than those in the first and second categories can be considered UGC users. For example, as shown in Table 1, users 1-n are assigned one of the three categories.
[0080] Table 1
[0081]
[0082]
[0083] After user classifier 302 labels the sample users, user characteristics can be constructed based on user behavior data, such as posting data, consumption data, and activity data. Posting data includes the number of posts (e.g., notes) made in the past week, whether the posts are copied content, and whether the posts promote products. Activity data includes the duration of activity in the past week (e.g., time spent in the recommendation section) and search frequency. Examples of user behavior data 1-n are shown in Table 2.
[0084] Table 2
[0085]
[0086] Once the samples have the above features and labels, they can be used as a training set to train a user identification model 302 for predicting user sincerity scores. The trained user identification model 302 performs the following functions: it takes user features as input and outputs the user's sincerity score (as a recommendation value), with the output sincerity score being a continuous value within the range [-100, 100]. For example, as... Figure 3B As shown, the user's characteristics are input, and the user identification model 302 outputs 27 points.
[0087] After training the user identification model 302 using the above method, it is possible to predict the sincerity score of users in the online community, and adjust the exposure weight of the content generated and published by the user in search and recommendation scenarios based on the sincerity score.
[0088] Those skilled in the art will understand that the above example uses a sincerity score as the recommendation value, but this application is not limited to this. The recommendation value can also be determined from any other user evaluation dimension related to the recommendation of user-generated content. The above example divides users into three categories, assigning -100 points to low sincerity users, 0 points to medium sincerity users, and 100 points to high sincerity users. However, this application is not limited to this. Users can be divided into any number of categories based on different user evaluation dimensions, and each category can be assigned any appropriate label or value that can distinguish the user category.
[0089] In this application, recommendations are made based on user characteristics, rather than on the characteristics of the content posted by the user. Recommendations based on the characteristics of posted content are computationally intensive and have low accuracy. In contrast, recommendations based on user characteristics have lower model complexity and higher accuracy.
[0090] In this application, a weakly supervised method is used to train the model. Labels for the training data are obtained by identifying different levels of user recommendation scores, eliminating the need for manual annotation. Training data features are constructed based on user behavior data, and the trained model can directly predict user recommendation scores based on these features.
[0091] According to some embodiments of this application, the user identification model is a regression model for predicting values between a first value and a third value. For example, the user identification model can be a Gradient Boosting Regressor model, which is a strong classifier composed of multiple weak classifiers. Even without manually labeled training data, a weakly supervised method can be used to train a user identification model that accurately predicts continuous values.
[0092] Now for reference Figure 4 The diagram shown is a block diagram of a user-generated content recommendation system according to an embodiment of this application.
[0093] like Figure 4 As shown, the user-generated content recommendation system 400 includes:
[0094] Module 402 is used to determine the user's recommendation value based on the user identification model, and
[0095] Recommendation module 404 is used to recommend user-generated content based on the user's recommendation score.
[0096] Training module 406 is used to train a user recognition model based on a weakly supervised training method. Training module 406 includes:
[0097] Acquisition unit 406a is used to acquire user information, which includes the user's category and user behavior data;
[0098] The determining unit 406b is used to determine the user's recommendation value based on the user's category, and the recommendation value includes multiple discrete numerical values;
[0099] Building unit 406c is used to build user characteristics based on user behavior data;
[0100] Unit 406d is used to assemble a sample set by combining the user's recommendation value and user features, wherein each user's user features are used as sample features and the recommendation value is used as the sample label to form a sample.
[0101] Training unit 406e is used to train the user recognition model using a sample set composed of constituent units 406d.
[0102] The first embodiment is a method embodiment corresponding to this embodiment, and this embodiment can be implemented in conjunction with the first embodiment. The relevant technical details mentioned in the first embodiment remain valid in this embodiment, and will not be repeated here to avoid repetition. Correspondingly, the relevant technical details mentioned in this embodiment can also be applied to the first embodiment.
[0103] Now for reference Figure 5 The diagram shows a block diagram of a user-generated content recommendation device 500 according to an embodiment of this application. Device 500 may include one or more processors 502, system control logic 508 connected to at least one of the processors 502, system memory 504 connected to the system control logic 508, non-volatile memory (NVM) 506 connected to the system control logic 508, and a network interface 510 connected to the system control logic 508.
[0104] Processor 502 may include one or more single-core or multi-core processors. Processor 502 may include any combination of general-purpose processors and special-purpose processors (e.g., graphics processors, application processors, baseband processors, etc.). In embodiments herein, processor 502 may be configured to perform operations according to... Figures 2A-3B One or more embodiments of the various embodiments shown.
[0105] In some embodiments, system control logic 508 may include any suitable interface controller to provide any suitable interface to at least one of the processors 502 and / or any suitable device or component communicating with system control logic 508.
[0106] In some embodiments, system control logic 508 may include one or more memory controllers to provide an interface to system memory 504. System memory 504 may be used to load and store data and / or instructions. In some embodiments, memory 504 of device 500 may include any suitable volatile memory, such as suitable dynamic random access memory (DRAM).
[0107] NVM / memory 506 may include one or more tangible, non-transitory computer-readable media for storing data and / or instructions. In some embodiments, NVM / memory 506 may include any suitable non-volatile memory such as flash memory and / or any suitable non-volatile storage device, such as at least one of HDD (Hard Disk Drive), CD (Compact Disc) drive, and DVD (Digital Versatile Disc) drive.
[0108] NVM / Storage 506 may include a portion of the storage resources mounted on the device 500, or it may be accessible by the device but is not necessarily part of the device. For example, NVM / Storage 506 may be accessed over a network via network interface 510.
[0109] Specifically, system memory 504 and NVM / memory 506 may each include a temporary copy and a permanent copy of instruction 520. Instruction 520 may include, when executed by at least one of processors 502, causing device 500 to perform, as Figures 2A-3B The instructions for the method shown. In some embodiments, the instructions 520, hardware, firmware and / or their software components may additionally / alternatively be located in system control logic 508, network interface 510 and / or processor 502.
[0110] Network interface 510 may include a transceiver for providing a radio interface to device 500, thereby enabling communication with any other suitable device (such as a front-end module, antenna, etc.) via one or more networks. In some embodiments, network interface 510 may be integrated into other components of device 500. For example, network interface 510 may be integrated into at least one of processor 502, system memory 504, NVM / memory 506, and firmware device (not shown) with instructions that, when at least one of processor 502 executes the instructions, enable device 500 to implement... Figures 2A-3B One or more embodiments of the various embodiments shown. Communication module.
[0111] The network interface 510 may further include any suitable hardware and / or firmware to provide a multiple-input multiple-output radio interface. For example, the network interface 510 may be a network adapter, a wireless network adapter, a telephone modem, and / or a wireless modem.
[0112] In one embodiment, at least one of the processors 502 may be packaged together with the logic of one or more controllers for system control logic 508 to form a system-in-package (SiP). In another embodiment, at least one of the processors 502 may be integrated on the same die with the logic of one or more controllers for system control logic 508 to form a system-on-a-chip (SoC).
[0113] Device 500 may further include an input / output (I / O) device 512. The I / O device 512 may include a user interface enabling a user to interact with device 500; the peripheral component interface is designed to allow peripheral components to also interact with device 500. In some embodiments, device 500 may also include sensors for determining at least one of environmental conditions and location information related to device 500.
[0114] In some embodiments, the user interface may include, but is not limited to, a display (e.g., a liquid crystal display, a touch screen display, etc.), a speaker, a microphone, one or more cameras (e.g., a still image camera and / or a video camera), a flashlight (e.g., a light-emitting diode flash), and a keyboard.
[0115] In some embodiments, the peripheral component interface may include, but is not limited to, a non-volatile memory port, an audio jack, and a power interface.
[0116] In some embodiments, the sensor may include, but is not limited to, a gyroscope sensor, an accelerometer, a proximity sensor, an ambient light sensor, and a positioning unit. The positioning unit may also be part of or interact with the network interface 510 to communicate with components of the positioning network (e.g., Global Positioning System (GPS) satellites).
[0117] It is understood that the structure illustrated in the embodiments of the present invention does not constitute a specific limitation on the user-generated content recommendation device 500. In other embodiments of this application, the user-generated content recommendation device 500 may include more or fewer components than illustrated, or combine some components, or split some components, or have different component arrangements. The illustrated components may be implemented in hardware, software, or a combination of software and hardware.
[0118] Program code can be applied to input instructions to perform the functions described herein and generate output information. The output information can be applied to one or more output devices in a known manner. For the purposes of this application, the processing system includes any system having a processor such as, for example, a digital signal processor (DSP), a microcontroller, an application-specific integrated circuit (ASIC), or a microprocessor.
[0119] The program code can be implemented using a high-level procedural language or an object-oriented programming language to communicate with the processing system. Assembly language or machine language can also be used when needed. In fact, the mechanisms described in this paper are not limited to any particular programming language. In either case, the language can be a compiled language or an interpreted language.
[0120] One or more aspects of at least one embodiment can be implemented by representational instructions stored on a computer-readable storage medium, the instructions representing various logics in a processor, which, when read by a machine, cause the machine to create logic for performing the techniques described herein. These representations, referred to as “IP cores,” can be stored on a tangible computer-readable storage medium and provided to multiple customers or production facilities for loading into manufacturing machines that actually manufacture the logic or processor.
[0121] According to some embodiments of this application, a computer storage medium is disclosed, on which instructions are stored, which, when executed on a computer, cause the computer to perform a method for recommending user-generated content as described in any embodiment of this application.
[0122] The first embodiment is a method embodiment corresponding to this embodiment, and this embodiment can be implemented in conjunction with the first embodiment. The relevant technical details mentioned in the first embodiment remain valid in this embodiment, and will not be repeated here to avoid repetition. Correspondingly, the relevant technical details mentioned in this embodiment can also be applied to the first embodiment.
[0123] According to some embodiments of this application, a computer program product is disclosed, including a computer program / instructions that, when executed by a processor, implement the steps of a method for recommending user-generated content as described in any embodiment of this application.
[0124] The first embodiment is a method embodiment corresponding to this embodiment, and this embodiment can be implemented in conjunction with the first embodiment. The relevant technical details mentioned in the first embodiment remain valid in this embodiment, and will not be repeated here to avoid repetition. Correspondingly, the relevant technical details mentioned in this embodiment can also be applied to the first embodiment.
[0125] In some cases, the disclosed embodiments may be implemented in hardware, firmware, software, or any combination thereof. The disclosed embodiments may also be implemented as instructions or programs carried or stored on one or more transient or non-transient machine-readable (e.g., computer-readable) storage media, which may be read and executed by one or more processors, etc. When the instructions or program are run by a machine, the machine may perform the various methods described above. For example, the instructions may be distributed via a network or other computer-readable media. Therefore, machine-readable media may include, but are not limited to, any mechanism for storing or transmitting information in a machine-readable (e.g., computer-readable) form, such as floppy disks, optical disks, optical disc read-only memories (CD-ROMs), magneto-optical disks, read-only memories (ROMs), random access memories (RAMs), erasable programmable read-only memories (EPROMs), electronically erasable programmable read-only memories (EEPROMs), magnetic cards or optical cards, or flash memory or tangible machine-readable storage for transmitting network information via electrical, optical, acoustic, or other forms of signals (e.g., carrier waves, infrared signals, digital signals, etc.). Therefore, machine-readable media includes any form of machine-readable medium suitable for storing or transmitting electronic instructions or machine (e.g., computer) readable information.
[0126] The embodiments of this application have been described in detail above with reference to the accompanying drawings. However, the use of the technical solutions of this application is not limited to the various applications mentioned in the embodiments of this patent. Various structures and modifications can be easily implemented with reference to the technical solutions of this application to achieve the various beneficial effects mentioned herein. Within the scope of knowledge possessed by those skilled in the art, all changes made without departing from the spirit of this application should be considered within the scope of this patent application.
Claims
1. A method for recommending user-generated content, used in electronic devices, characterized in that, The method includes: Determining user recommendation values based on user identification models, and Recommend user-generated content based on the user's recommendation score; The user identification model is obtained through a weakly supervised training method, which includes: Obtain user information, which includes user categories and user behavior data, wherein the user behavior data includes posting data, consumption data, and activity data; The recommendation value for the user is determined based on the user's category, and the recommendation value includes multiple discrete numerical values; Based on the user's behavioral data, the user's characteristics are constructed; The user's recommendation value and user features are combined to form a sample set, wherein each user's user features are used as sample features and the recommendation value is used as a sample label to form a sample; The user identification model is trained using the sample set.
2. The method of claim 1, wherein, Recommending user-generated content based on the user's recommendation score includes: Reduce the recommendation weight of user-generated content whose recommendation value is below a threshold.
3. The method of claim 1, wherein, Recommending user-generated content based on the user's recommendation score includes: The recommendation weight of the user-generated content is determined based on the magnitude of the recommendation value.
4. The method of claim 1, wherein, The recommendation value for the user is determined based on the user's category. The recommendation value includes multiple discrete numerical values, including: The recommendation value of the user marked as non-compliant is determined to be a first value, the recommendation value of the new user is determined to be a second value, and the recommendation value of other users is determined to be a third value, wherein the first value < the second value < the third value.
5. The method of claim 4, wherein, The model is a regression model where the predicted value is between the first value and the third value.
6. The method of claim 1, wherein, The behavioral data includes posting data, consumption data, and activity data. The posting data includes the number of posts published in the past week, whether the posted content is plagiarized content, and whether the posted content promotes sales. The activity data includes the active duration and search frequency in the past week.
7. A recommendation system for user-generated content, characterized in that include The determination module is used to determine the user's recommendation value based on the user identification model, and The recommendation module is used to recommend user-generated content based on the user's recommendation score. A training module is used to train the user identification model based on a weakly supervised training method. The training module includes: An acquisition unit is used to acquire user information, which includes the user's category and user behavior data, wherein the user behavior data includes posting data, consumption data, and activity data; a determination unit is used to determine the user's recommendation value based on the user's category, wherein the recommendation value includes multiple discrete values; A construction unit is used to construct the user's user characteristics based on the user's behavioral data; A component unit is used to assemble the user's recommendation value and user features into a sample set, wherein each user's user features are used as sample features and the recommendation value is used as a sample label to form a sample; The training unit uses the sample set to train the user identification model.
8. A recommendation device for user-generated content, characterized in that, The device includes a memory that stores computer-executable instructions and a processor that, when the instructions are executed by the processor, cause the device to implement the method of any one of claims 1 6. The method of recommending user-generated content according to any one of claims 1-5.
9. A computer storage medium, characterized in that, The computer storage medium stores instructions which, when executed on a computer, cause the computer to perform the method according to any one of claims 1 6. The method of recommending user-generated content according to any one of claims 1 to 5.
Citation Information
Patent Citations
Text end related reading module individuation method and device, medium and electronic equipment
CN110377723A
Method, device and system for matching and recommending human resource based on artificial intelligence using video content platform service
KR102298722B1
