Two-dimensional user system construction method and system oriented to carbon-Plevatory platform

Abstract

The invention provides a two-dimensional user system construction method and system oriented to a carbon general benefit platform, and relates to the technical field of computers. The method comprises the following steps: acquiring multi-source data related to green behaviors of a user in a carbon general benefit platform; extracting behavior features and psychological features in the multi-source data; performing weighted fusion on the behavior features and the psychological features through a pre-constructed deep learning model to generate psychological-behavior portrait feature vectors; according to the psychological-behavioral portrait feature vectors, clustering and labeling the users to form user labels; according to the user label, personalized strategy pushing is carried out through a carbon general benefit platform; and updating the fusion weight of the behavior features and the psychological features according to the personalized strategy pushing result. According to the method, multi-dimensional consideration is carried out on the psychological characteristics and the behavior characteristics of the user in the carbon-Popingness platform, so that the behavior habit and the psychological motivation of the user are more fully reflected, and a solid foundation is provided for strategy optimization of the carbon-Popingness platform.

Description

Technical Field

[0001] This invention relates to the field of computer technology, and specifically to a method and system for constructing a two-dimensional user system for a carbon benefit platform. Background Technology

[0002] The carbon inclusion mechanism aims to incorporate individual energy conservation and emission reduction behaviors into the carbon management system through market-based and social means, making public low-carbon behavior quantifiable and incentivized. With the development of digital technology, the carbon inclusion mechanism is gradually being implemented in the form of carbon inclusion platforms. These platforms encourage users to proactively reduce carbon emissions in their daily activities such as travel, electricity use, and consumption through incentive points, task guidance, and reward redemption, thereby promoting the socialization and routine implementation of energy conservation and emission reduction actions.

[0003] Existing platforms generally adopt data analysis methods based on behavioral outcomes. By recording indicators such as users' check-in frequency, travel mileage, or energy-saving electricity consumption, they quantify and evaluate low-carbon behaviors and generate points and levels based on carbon emission reductions.

[0004] However, these methods can only reflect users' overt behavioral characteristics and are difficult to reveal their underlying psychological motivations and behavioral intentions. Summary of the Invention

[0005] (a) Technical problems to be solved To address the shortcomings of existing technologies, this invention provides a method and system for constructing a two-dimensional user profile, which solves the problem of narrow user identification dimensions in existing carbon benefit platforms.

[0006] (II) Technical Solution To achieve the above objectives, the present invention provides the following technical solution: Firstly, this application provides a method for constructing a two-dimensional user system for a carbon benefit platform, including: Obtain multi-source data related to users' green behaviors from the carbon benefit platform; Extract behavioral and psychological features from the multi-source data; The behavioral features and psychological features are weighted and fused together by a pre-built deep learning model to generate a psychological-behavioral profile feature vector; the deep learning model includes a behavioral channel and a psychological channel. Based on the psychological-behavioral profile feature vectors, users are clustered and labeled to form user tags; Personalized strategies are pushed through the carbon benefit platform based on the user tags; The fusion weights of the behavioral features and psychological features are updated based on the results of the personalized strategy push.

[0007] Preferably, the acquisition of multi-source data related to users' green behaviors in the carbon benefit platform includes: Establish cross-source data mapping relationships related to users' green behaviors in the carbon benefit platform based on users' unique identifiers and timestamps; Based on the cross-source data mapping relationship, multi-source data related to users' green behaviors are extracted from the carbon benefit platform. The multi-source data includes behavior log data, psychological questionnaire data, and text comment data.

[0008] Preferably, the step of weightedly fusing the behavioral features and psychological features through a pre-built deep learning model to generate a psychological-behavioral profile feature vector includes: The behavioral features are processed through the behavioral channel to obtain a behavioral feature vector; the psychological features are processed through the psychological channel to obtain a psychological feature vector. The attention mechanism of the deep learning model is used to weight and fuse the behavioral feature vector and the psychological feature vector to obtain a psychological-behavioral profile feature vector.

[0009] Preferably, the personalized strategy push based on the user tag through the carbon credit platform includes: The mapping relationship between the user tags and the platform policies is established through the business rules of the carbon benefit platform; A multi-dimensional strategy matching rule base for the carbon benefit platform is established based on the aforementioned mapping relationship; The platform pushes personalized strategies based on the multi-dimensional strategy matching rule base and performs a quantitative evaluation of the execution results of the platform's strategy push.

[0010] Preferably, the behavioral pathway includes a long short-term memory network, and the psychological pathway includes a multilayer perceptron.

[0011] Preferably, the user clustering and tagging based on the psychological-behavioral profile feature vector to form user tags includes: Cluster analysis is performed on the aforementioned psychological-behavioral profile feature vectors; Based on the results of the cluster analysis, the user's profile feature indicators are calculated, including the mean of behavioral features, the mean of psychological features, and the distribution of emotional features. The user tags are generated based on the profile feature indicators and platform business rules.

[0012] Preferably, the user tag includes: “High participation-high motivation type”, “Medium participation-medium motivation type”, “Low participation-potential type” and “Low participation-low motivation type”.

[0013] Secondly, this application also provides a two-dimensional user system construction system for carbon benefit platforms, including: The acquisition module retrieves multi-source data related to users' green behaviors from the carbon benefit platform; The feature extraction module extracts behavioral and psychological features from the multi-source data; The first processing module uses a pre-built deep learning model to weightedly fuse the behavioral features and psychological features to generate a psychological-behavioral profile feature vector; the deep learning model includes a behavioral channel and a psychological channel. The second processing module performs user clustering and tagging based on the psychological-behavioral profile feature vector to form user tags. The strategy push module pushes personalized strategies through the carbon benefit platform based on the user tags. The update module updates the fusion weights of the behavioral features and the psychological features based on the results of the personalized strategy push.

[0014] Thirdly, this application also provides a computer-readable storage medium storing a computer program for constructing a two-dimensional user system for a carbon benefit platform, wherein the computer program causes a computer to execute the two-dimensional user system construction method for a carbon benefit platform as described above.

[0015] Fourthly, this application also provides an electronic device, comprising: One or more processors, a memory, and one or more programs, wherein the one or more programs are stored in the memory and configured to be executed by the one or more processors, the programs including methods for executing a two-dimensional user system construction method for a carbon inclusive platform as described above.

[0016] (III) Beneficial Effects This invention provides a method and system for constructing a two-dimensional user system for carbon benefit platforms. Compared with existing technologies, it has the following advantages: This invention takes into account the psychological and behavioral characteristics of users on the carbon benefit platform from multiple dimensions. By integrating features through a deep learning model that incorporates attention mechanisms, it generates a comprehensive user profile that can more fully reflect users' behavioral habits and psychological motivations, providing a solid foundation for the optimization of carbon benefit platform strategies. Attached Figure Description

[0017] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0018] Figure 1 This is a flowchart illustrating a method for constructing a two-dimensional user system for a carbon benefit platform, as provided in an embodiment of this application. Detailed Implementation

[0019] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, the technical solutions in the embodiments of the present invention are described clearly and completely. Obviously, the described embodiments are only some embodiments of the present invention, not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0020] This application provides a method and system for constructing a two-dimensional user system for carbon benefit platforms, which solves the problem of narrow user identification dimensions in existing carbon benefit platforms and constructs a user system that simultaneously considers user behavior habits and psychological motivations.

[0021] The technical solution in this application is to solve the above-mentioned technical problems, and the general idea is as follows: Existing platforms generally employ data analysis methods based on behavioral outcomes. They quantify low-carbon behavior by recording indicators such as user check-in frequency, travel mileage, or energy-saving electricity consumption, generating points and levels based on carbon emission reductions. However, these methods have limitations: first, they only reflect users' explicit behavioral characteristics, failing to reveal their underlying psychological motivations and behavioral intentions; second, they focus primarily on building the overall operational process, neglecting the integration and utilization of multi-source data, making it difficult to form a comprehensive understanding of users and limiting the platform's ability to develop personalized intervention strategies. Current technologies cannot yet achieve collaborative modeling of the psychological motivations and behavioral characteristics of carbon credit platform users, and there is a lack of effective methods to integrate multi-source data and construct comprehensive user profiles.

[0022] To address the problems of single data sources, narrow user identification dimensions, and homogeneous intervention methods in existing carbon credit platforms, this invention provides a method and system for constructing a two-dimensional user system for carbon credit platforms. This system enables accurate identification of users' low-carbon behaviors and effective characterization of their psychological intentions, thereby providing differentiated user segmentation criteria and personalized intervention strategies for carbon credit platforms, and enhancing the public's sustained participation in low-carbon behaviors.

[0023] To better understand the above technical solutions, the following will provide a detailed explanation of the technical solutions in conjunction with the accompanying drawings and specific implementation methods.

[0024] like Figure 1 As shown in the embodiments of this application, a method for constructing a two-dimensional user system for a carbon benefit platform is provided. Specific steps include: Step S110: Obtain multi-source data related to users' green behaviors from the carbon benefit platform.

[0025] Step S120: Extract behavioral and psychological features from multi-source data.

[0026] Step S130: The behavioral features and psychological features are weighted and fused through a pre-built deep learning model to generate a psychological-behavioral profile feature vector; the deep learning model includes a behavioral channel and a psychological channel.

[0027] Step S140: Cluster and label users based on psychological-behavioral profile feature vectors to form user tags.

[0028] Step S150: Personalized strategy push is carried out through the carbon credit platform based on user tags.

[0029] Step S160: Update the fusion weights of behavioral features and psychological features based on the results of personalized strategy push.

[0030] In this embodiment, by considering the psychological and behavioral characteristics of users on the carbon benefit platform from multiple dimensions, and by using a deep learning model that incorporates attention mechanisms to fuse the features and generate a comprehensive user profile, a more complete reflection of user behavior habits and psychological motivations can be achieved, providing a solid foundation for the strategy optimization of the carbon benefit platform.

[0031] Step S110: Obtain multi-source data related to users' green behaviors from the carbon benefit platform. Specifically, this embodiment includes the following steps: Step S111: Establish cross-source data mapping relationships for users in the carbon benefit platform based on user unique identifiers and timestamps.

[0032] Step S112: Based on the cross-source data mapping relationship, extract multi-source data related to users' green behavior from the carbon benefit platform. The multi-source data includes behavior log data, psychological questionnaire data, and text comment data.

[0033] Specifically, the system retrieves user behavior data from the platform's log database, including information on green travel, task check-ins, and energy-saving electricity consumption; it also retrieves user scores on the low-carbon attitude scale and environmental belief scale from the platform's psychological questionnaire system; and it retrieves user text comments, such as task evaluations and policy discussion messages, from the platform's community and task feedback system. For example, taking "User B" as an example, the platform log records their "cycling trip" data at 07:55 on September 1, 2024; the psychological questionnaire system records their low-carbon attitude score of 4.5 and their environmental belief score of 5; and they post a comment in the task feedback area: "I find walking to and from get off work very meaningful." Through matching rules based on UID consistency and time intervals of less than 5 minutes, the system automatically associates the three records—behavior logs, psychological scores, and text comments—to form a multi-source dataset.

[0034] In this embodiment, multi-source data from users in the carbon benefit platform database is collected, and a cross-source mapping relationship is established based on anonymous user IDs and timestamps, enabling automatic association of multi-modal data. This unified association mechanism for multi-source data based on anonymous user IDs and timestamps can achieve automatic matching and integrated management of behavioral data, psychological data, and text data.

[0035] Step S120: Extract behavioral and psychological features from multi-source data.

[0036] Specifically, behavioral feature extraction involves time-series modeling of user behavior logs to extract dynamic features such as travel frequency, task responsiveness, and energy-saving sustainability. For example, User B cycled 5 times, walked 2 times, completed 4 low-carbon tasks, and saved a total of 30 kWh of electricity in a week, resulting in behavioral feature vectors of 5, 2, 4, and 30. Psychological feature extraction involves quantifying scores on psychological questionnaires, calculating average scores, and normalizing them. For example, User B's average score on the low-carbon attitude scale is 4.5, and their score on the environmental belief scale is 5, resulting in psychological features of 0.85 and 0.92 after standardization. Thematic identification and sentiment analysis are performed on user comment texts to extract theme weights and sentiment polarities. For example, User B's comment, "Walking to and from get off work makes me feel very meaningful," is identified by NLP analysis as having the themes of "active participation" and "environmental awareness," with a positive sentiment polarity of 1.0, corresponding to text features of 1, 1, and 1.0.

[0037] The specific implementation of the present invention also includes step S121, which standardizes the behavioral features and psychological features and aligns them by user ID to obtain the feature matrix of the user sample.

[0038] In this embodiment, the extracted features are Z-score standardized to eliminate dimensional differences and aligned by user ID to obtain the feature vector of a single user sample. The feature vectors of all user samples are then stacked in batches to form a unified feature matrix, which is used as the input to the subsequent deep learning model.

[0039] Step S130 involves weighted fusion of behavioral and psychological features using a pre-built deep learning model to generate a psychological-behavioral profile feature vector. The deep learning model includes both behavioral and psychological channels. The specific implementation of this step includes the following steps: Step S131: Process behavioral features according to the behavioral channel to obtain a behavioral feature vector; process psychological features according to the psychological channel to obtain a psychological feature vector.

[0040] Step S132: The behavioral feature vector and the psychological feature vector are weighted and fused through the attention mechanism of the deep learning model to obtain the psychological-behavioral profile vector.

[0041] Specifically, the pre-built deep learning model has a behavioral channel for processing behavioral features and a psychological channel for processing psychological features. Optionally, the behavioral channel is a recurrent neural network (RNN) or a long short-term memory network (LSTM); user behavioral feature vectors are input into the behavioral channel to capture low-carbon behavior patterns and trends; for example, after processing the behavioral feature vector of "User B", the system identifies it as a pattern of "high-frequency travel + regular energy saving". Optionally, the psychological channel is a multilayer perceptron (MLP) or a convolutional neural network (CNN); user psychological feature vectors and text feature vectors are input into the psychological channel to extract potential psychological motivations and emotional features; for example, "User B's" psychological score and positive comments together reflect its high environmental awareness and high willingness to act, and the psychological channel outputs potential vectors of 0.72, 0.88, and 0.65.

[0042] The deep learning model in this embodiment also incorporates an attention mechanism, which dynamically assigns weights based on the importance of features, and then weights and fuses behavioral and psychological features. For example, the behavioral weights of user B... α b =0.45, psychological weight α m =0.55 is used to perform weighted fusion of the two types of features. The weight is calculated using the following expression: in, h b and h m These represent the output vectors of the behavioral channel and the psychological channel, respectively, both with dimension [missing information]. d ; W b andW m These represent trainable weight vectors that linearly score the outputs of the behavioral and psychological channels, respectively.

[0043] User B's behavior channel output vector is h b =[0.5,-0.2,0.1], the mental channel output vector is h m =[0.3,0.7,-0.1], the weight matrix is ​​obtained through training. W b =[0.2,0.5,0.3], W m If the expression is [0.4, 0.4, 0.2], then the attention weights are calculated. α b ≈0.413, α m ≈0.587.

[0044] Psychological-behavioral profiling vector H The calculation expression is: In this embodiment, the final fusion vector can be calculated according to the formula as follows: H =[0.383, 0.328, [0.017], which simultaneously represents the user's explicit behavioral patterns and intrinsic psychological motivations.

[0045] By processing the feature vectors through the behavioral and psychological channels of the deep learning model and then weighting and fusing them, a user's psychological-behavioral fusion profile vector is obtained. This vector comprehensively translates the user's explicit behavioral patterns and psychological motivational tendencies.

[0046] In this embodiment, a dual-channel deep learning architecture with parallel behavioral and mental channels is designed to model the user's explicit behavioral patterns and internal psychological motivations, respectively, thereby enhancing the expressive dimensions of the user profile. This embodiment introduces an attention mechanism in the feature fusion stage, which dynamically weights features according to their importance, generating a high-precision mind-behavior fusion profile vector.

[0047] Step S140: Cluster and label users based on psychological-behavioral profile feature vectors to form user tags.

[0048] Specifically, K-Means clustering analysis (K=4) was performed on the fused profile vectors of all users to identify four typical user groups, and corresponding user tags were generated accordingly, including "high engagement-high motivation", "medium engagement-medium motivation", "low engagement-potential", and "low engagement-low motivation".

[0049] Based on the clustering results and the business rules of the carbon benefit platform, a corresponding user tagging system is generated. This system groups user profile vectors according to a combination of user behavioral and psychological characteristics. For example, the profile vector of "User B" belongs to the first group and is labeled as "High Engagement-High Motivation".

[0050] This application proposes a user clustering and tag generation method based on fused profile vectors, which can achieve refined identification of different types of user groups and provide data support for personalized intervention of the carbon benefit platform.

[0051] Step S150: Personalized strategy recommendations are pushed through the carbon credit platform based on user tags. The specific implementation of this step includes the following steps: Step S151: Establish a mapping relationship between user tags and platform policies through the carbon benefit platform business rules.

[0052] Step S152: Establish a multi-dimensional strategy matching rule base for the carbon benefit platform through mapping relationships.

[0053] Step S153: Execute personalized strategy push for the platform based on the multi-dimensional strategy matching rule base, and quantitatively evaluate the execution result of the platform strategy push.

[0054] Step S160: Update the fusion weight of behavioral features and psychological features based on the personalized strategy push results.

[0055] Specifically, a multi-dimensional strategy matching rule base is generated based on the user system to conduct personalized interventions and recommendations for the carbon benefit platform. Corresponding intervention strategies and recommended content are matched according to user tags. A mapping relationship between tags and strategies is established based on user segmentation results to achieve precise strategy matching. For example, for "low-participation-potential" users, a composite intervention strategy of "new user guidance tasks + tiered points rewards" is automatically matched; for "high-participation-high-motivation" users, a strategy of "challenging tasks + honor badges" is matched to meet their achievement needs. The strategy matching unit monitors changes in user tag status in real time and dynamically adjusts the applicable intervention strategies to ensure the timeliness and relevance of the strategies.

[0056] This embodiment also includes establishing a multi-dimensional evaluation index system to quantitatively evaluate the effectiveness of the intervention strategy. By tracking short-term indicators such as task click-through rate and completion rate, medium-term indicators such as user activity and behavior frequency, and long-term indicators such as user retention rate and total carbon emission reduction, the effectiveness of the strategy is comprehensively evaluated.

[0057] The specific implementation of the present invention also includes updating the weights of behavioral and psychological features in the attention mechanism according to the strategy effect, while supporting dynamic adjustment of the multi-dimensional strategy matching rule base.

[0058] The specific implementation of this invention also includes conducting parallel testing and effect comparison of candidate intervention strategies through randomized group experiments. During the testing process, the target user group is randomly divided into multiple test groups, and different intervention strategies are applied to each group. User behavior data from each group is collected and statistically analyzed. Based on the test results, the optimal strategy is selected for full-scale promotion, and effective strategy elements are integrated and innovated to continuously optimize the strategy library.

[0059] This invention also provides a two-dimensional user system construction system for carbon benefit platforms, comprising: The acquisition module retrieves multi-source data related to users' green behaviors from the carbon benefit platform.

[0060] The feature extraction module extracts behavioral and psychological features from multi-source data.

[0061] The first processing module uses a pre-built deep learning model to weightedly fuse behavioral and psychological features to generate a psychological-behavioral profile feature vector; the deep learning model includes behavioral and psychological channels.

[0062] The second processing module clusters and labels users based on their psychological-behavioral profile feature vectors to form user tags.

[0063] The strategy push module pushes personalized strategies through the carbon benefit platform based on user tags.

[0064] The update module updates the fusion weights of behavioral and psychological characteristics based on the results of personalized strategy pushes.

[0065] It is understood that the dual-dimensional user system construction system for carbon benefit platforms provided in this embodiment of the invention corresponds to the dual-dimensional user system construction method for carbon benefit platforms described above. The explanations, examples, and beneficial effects of the relevant content can be referred to the corresponding content in the dual-dimensional user system construction method for carbon benefit platforms, and will not be repeated here.

[0066] This invention also provides a computer-readable storage medium storing a computer program for constructing a two-dimensional user system for a carbon inclusive platform, wherein the computer program causes a computer to execute the two-dimensional user system construction method for a carbon inclusive platform as described above.

[0067] This application also provides an electronic device, including: one or more processors; a memory; and one or more programs, wherein the one or more programs are stored in the memory and configured to be executed by the one or more processors, the programs including methods for performing a two-dimensional user system construction method for a carbon inclusive platform as described above.

[0068] In summary, compared with existing technologies, it has the following beneficial effects: 1. This invention, based on traditional methods that only consider behavioral characteristics, incorporates psychological features into the user modeling process by extracting psychological questionnaire data and text comment data. Furthermore, it fuses multi-dimensional features through an attention mechanism, enabling the generated profile vector to more fully reflect user behavior habits and psychological motivations in a high-dimensional space. This significantly improves the accuracy and expressive power of user profiles, solving the problem that traditional methods, which rely solely on user behavior outcomes, struggle to capture users' psychological motivations and intentions, resulting in profiles with limited dimensions and low discriminative power.

[0069] 2. This invention, through multi-source data collection and unified association mechanisms, standardized feature extraction processes, and automatic clustering label generation methods, constitutes a complete automated modeling system. Compared to traditional methods that rely on manual rule segmentation and single-modal data processing, this invention can achieve efficient construction and cluster labeling of user profiles without manual intervention, reducing operating costs and improving the platform's intelligence level.

[0070] 3. This invention introduces an attention mechanism in the fusion stage, allowing the weight distribution of each modality feature to be explicitly calculated during training, thus providing better interpretability. For different types of users, the differences in the contributions of behavioral and psychological features to the final profile generation can clearly reflect the reasons for profile formation, facilitating subsequent strategy formulation and optimization. Simultaneously, this dual-channel modeling structure and feature fusion mechanism features a modular design, allowing for flexible expansion to other modalities (such as sensor data, social interaction data, etc.) according to actual business needs, further enhancing the richness and applicability of user profiles.

[0071] 4. This invention, through psychological-behavioral dual-dimensional modeling, can further differentiate users within groups with similar participation frequencies based on psychological characteristics and emotional expression. It can identify potentially active but behaviorally deficient user groups, as well as groups with outwardly active behavior but weak psychological motivation. This segmentation method is more targeted and provides a theoretical basis for subsequent personalized operations. It solves the problem that traditional user segmentation based on behavioral data often only roughly distinguishes between active and inactive users, making it difficult to identify differences in users' psychological intentions.

[0072] 5. The dual-dimensional user system provided in this embodiment of the invention allows the platform to implement differentiated interventions based on users' behavioral patterns and psychological motivations in task push and incentive strategies. For example, for users with "low participation – high motivation," customized incentive tasks can be set to stimulate their potential behavioral intentions; for users with "high participation – low motivation," their long-term participation can be enhanced through community interaction, emotional resonance, and other methods.

[0073] 6. The dual-dimensional user system construction method provided in this embodiment of the invention not only enables more accurate user identification but also provides a reliable basis for differentiated incentive strategies and personalized task pushes, thereby helping to improve public participation and sustainability in carbon emission reduction. By identifying and effectively intervening in potentially high-willed users, this invention can theoretically improve the overall low-carbon behavior response rate and activity level of the platform, providing technical support for achieving carbon peaking and carbon neutrality goals.

[0074] It should be noted that, in this document, relational terms such as "first" and "second" are used only to distinguish one entity or operation from another, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Furthermore, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such a process, method, article, or apparatus. Without further limitations, an element defined by the phrase "comprising one..." does not exclude the presence of other identical elements in the process, method, article, or apparatus that includes said element.

[0075] The above embodiments are only used to illustrate the technical solutions of the present invention, and are not intended to limit it. Although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features. Such modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. A method for constructing a two-dimensional user system for a carbon benefit platform, characterized in that, include: Obtain multi-source data related to users' green behaviors from the carbon benefit platform; Extract behavioral and psychological features from the multi-source data; The behavioral features and psychological features are weighted and fused together by a pre-built deep learning model to generate a psychological-behavioral profile feature vector; the deep learning model includes a behavioral channel and a psychological channel. Based on the psychological-behavioral profile feature vectors, users are clustered and labeled to form user tags; Personalized strategies are pushed through the carbon benefit platform based on the user tags; The fusion weights of the behavioral features and psychological features are updated based on the results of the personalized strategy push.

2. The method for constructing a two-dimensional user system according to claim 1, characterized in that, The acquisition of multi-source data related to users' green behaviors in the carbon benefit platform includes: Establish cross-source data mapping relationships related to users' green behaviors in the carbon benefit platform based on users' unique identifiers and timestamps; Based on the cross-source data mapping relationship, multi-source data related to users' green behaviors are extracted from the carbon benefit platform. The multi-source data includes behavior log data, psychological questionnaire data, and text comment data.

3. The method for constructing a two-dimensional user system according to claim 1, characterized in that, The step of generating a psychological-behavioral profile feature vector by weighted fusion of the behavioral features and psychological features through a pre-built deep learning model includes: The behavioral features are processed through the behavioral channel to obtain a behavioral feature vector; the psychological features are processed through the psychological channel to obtain a psychological feature vector. The attention mechanism of the deep learning model is used to weight and fuse the behavioral feature vector and the psychological feature vector to obtain a psychological-behavioral profile feature vector.

4. The method for constructing a two-dimensional user system according to claim 3, characterized in that, The personalized strategy push based on the user tag through the carbon benefit platform includes: The mapping relationship between the user tags and the platform policies is established through the business rules of the carbon benefit platform; A multi-dimensional strategy matching rule base for the carbon benefit platform is established based on the aforementioned mapping relationship; The platform pushes personalized strategies based on the multi-dimensional strategy matching rule base and performs a quantitative evaluation of the execution results of the platform's strategy push.

5. The method for constructing a two-dimensional user system according to claim 3, characterized in that, The behavioral pathway includes a long short-term memory network, and the psychological pathway includes a multilayer perceptron.

6. The method for constructing a two-dimensional user system according to claim 1, characterized in that, Based on the aforementioned psychological-behavioral profile feature vectors, user clustering and tagging are performed to form user tags, including: Cluster analysis is performed on the aforementioned psychological-behavioral profile feature vectors; Based on the results of the cluster analysis, the user's profile feature indicators are calculated, including the mean of behavioral features, the mean of psychological features, and the distribution of emotional features. The user tags are generated based on the profile feature indicators and platform business rules.

7. The method for constructing a two-dimensional user system according to claim 6, characterized in that, The user tags include: "High participation - high motivation type", "medium participation - medium motivation type", "low participation - potential type" and "low participation - low motivation type".

8. A two-dimensional user system construction system for a carbon benefit platform, characterized in that, include: The acquisition module retrieves multi-source data related to users' green behaviors from the carbon benefit platform; The feature extraction module extracts behavioral and psychological features from the multi-source data; The first processing module uses a pre-built deep learning model to weightedly fuse the behavioral features and psychological features to generate a psychological-behavioral profile feature vector; the deep learning model includes a behavioral channel and a psychological channel. The second processing module performs user clustering and tagging based on the psychological-behavioral profile feature vector to form user tags. The strategy push module pushes personalized strategies through the carbon benefit platform based on the user tags. The update module updates the fusion weights of the behavioral features and the psychological features based on the results of the personalized strategy push.

9. A computer-readable storage medium, characterized in that, It stores a computer program for constructing a two-dimensional user system for a carbon benefit platform, wherein the computer program causes a computer to execute the two-dimensional user system construction method for a carbon benefit platform as described in any one of claims 1 to 7.

10. An electronic device, characterized in that, include: One or more processors, a memory, and one or more programs, wherein the one or more programs are stored in the memory and configured to be executed by the one or more processors, the programs including methods for performing a two-dimensional user system construction method for a carbon inclusive platform as described in any one of claims 1 to 7.

Images