Internet enterprise social responsibility evaluation and treatment method and system based on artificial intelligence

Abstract

The invention discloses an Internet enterprise social responsibility evaluation and treatment method and system based on artificial intelligence. Comprising the steps of obtaining related data of a platform-type enterprise, constructing a social responsibility evaluation model of the platform-type enterprise based on a multi-objective and multi-layer probability fusion LSTM network optimized layer by layer, generating result data through fuzzy comprehensive evaluation and empirical analysis, and constructing a platform-type enterprise social responsibility supervision and governance mechanism based on a block chain. According to the invention, a platform type enterprise social responsibility evaluation system is established based on the long short-term memory network LSTM, supervision and governance are carried out through block chain updating and replacement data, and the evaluation system based on the ESG concept is combined, so that the relationship between targets and the relationship between targets and scenes can be accurately expressed and efficiently analyzed and processed, and the evaluation efficiency of the enterprise social responsibility evaluation system is improved. The monitoring and evaluation effect on the social responsibility of the platform-type enterprise is improved, and cooperative treatment of multiple subjects is realized.

Description

Technical Field

[0001] The present invention relates to the field of data processing technology, and in particular to an artificial intelligence-based Internet enterprise social responsibility assessment and governance method and system. Background Art

[0002] With the development of society, it is necessary to build a social responsibility evaluation mechanism for platform-based enterprises in the new era and new situation. The evaluation of corporate social responsibility generally adopts the stakeholder theory, but for borderless platform-based enterprises, their stakeholders are shareholders, suppliers, consumers, and even competitors, which is a complex network system that evolves over time. Therefore, the evaluation of social responsibility of platform-based enterprises must also be a long-term multi-dimensional time series model.

[0003] The long short-term memory network LSTM of the multi-layer neural network is a time recursive neural network, which is suitable for processing and predicting important events with relatively long intervals and delays in time series. However, at present, there is a lack of a social responsibility evaluation system for Internet platform companies. The supervision, evaluation and governance of Internet platform companies are still relatively chaotic and the level of intelligence is low. Therefore, in order to help promote the healthy and orderly development of the Internet industry and society and improve the level of intelligence in social governance, we propose an Internet enterprise social responsibility evaluation and governance method and system based on artificial intelligence. Summary of the invention

[0004] In view of the above problems existing in the existing corporate social responsibility supervision, the present invention is proposed.

[0005] Therefore, one of the purposes of the present invention is to provide an artificial intelligence-based Internet enterprise social responsibility assessment and governance method and system, which constructs an Internet platform enterprise social responsibility model, establishes a platform enterprise social responsibility evaluation system based on the long short-term memory network LSTM, and simultaneously conducts supervision and governance by updating and replacing data through blockchain, which not only improves the monitoring and evaluation effect of the social responsibility of platform enterprises, but also realizes the collaborative governance of multiple subjects.

[0006] In order to solve the above technical problems, the present invention provides the following technical solutions:

[0007] On the one hand, the present invention provides an artificial intelligence-based Internet enterprise social responsibility assessment and governance method and system, including:

[0008] Obtain relevant data of platform-type enterprises, and based on the characteristics of corporate social responsibility, perform feature extraction, cluster analysis and pattern recognition on the relevant data of platform-type enterprises in turn, aggregate them into characteristic data to be evaluated for platform-type enterprises, perform data preprocessing, and generate corresponding multimodal data sets;

[0009] Based on the multi-objective and multi-layer probability fusion LSTM network optimized layer by layer, the social responsibility evaluation model of platform-type enterprises was constructed after dividing the training set and the test set for parameter optimization;

[0010] The big model calls the social responsibility evaluation model of platform-type enterprises and the multimodal data of the multimodal data set to classify, train, identify and associate the input multimodal data, and build the review objectives of the social responsibility evaluation of platform-type enterprises. According to the review objectives, an expert database and a corresponding expert group are generated. After determining the evaluation index system of the social responsibility evaluation of platform-type enterprises, the fuzzy comprehensive evaluation and empirical analysis of the multi-objective and multi-layer probability fusion LSTM model are combined to generate the result data of the social responsibility evaluation of the big model expert group, export the result data, and transmit the result data into the blockchain and visualize it;

[0011] Build a platform-type enterprise social responsibility supervision and governance mechanism based on blockchain, that is, build a social responsibility governance model for platform-type enterprises based on peer-to-peer networking, distributed consensus collaboration and contribution-based incentive mechanism. Based on the combination of smart contracts and the digital RMB system, different types of corporate social responsibility governance virtual experiments, governance scenario simulations and governance results evaluations are carried out through blockchain.

[0012] As a preferred solution of the present invention, the specific steps of constructing a social responsibility evaluation model for platform-based enterprises based on a multi-objective and multi-layer probability fusion LSTM model optimized layer by layer are as follows:

[0013] Use a shallow LSTM model for training until the training converges, and retain the classification layer and objective function in the original LSTM model;

[0014] Add a new layer of LSTM model and objective function to retrain the model and fine-tune the original parameters of the social responsibility evaluation model for platform-based enterprises;

[0015] The Softmax functions used in different classification layers are transformed to obtain the predicted probability scores of social responsibility assessment at each layer;

[0016] The probability scores of multiple layers are weighted and fused to calculate the comprehensive prediction score of social responsibility assessment and output it as result data.

[0017] As a preferred solution of the present invention, wherein: the review objectives include ESG review objectives; the corporate social responsibility characteristics include ESG evaluation indicator characteristics; the result data includes ESG evaluation result data, and the ESG evaluation result data is used for blockchain, ESG investment and smart contracts;

[0018] ESG evaluation indicator characteristics are divided into primary attributes and secondary attributes. The primary attributes of ESG evaluation indicator characteristics include environmental issues, social concerns and corporate governance. Environmental issues are climate change and sustainability category characteristics, corporate governance is diversity and consumer protection category characteristics, and corporate governance attributes are management structure, executive compensation and employee compensation category characteristics. The secondary attributes of ESG evaluation indicator characteristics include social public services, environmental protection, scientific and technological innovation, cyberspace security and expanding open cooperation.

[0019] Corporate social responsibility characteristics also include corporate social responsibility characteristics influencing characteristics, and corporate social responsibility characteristics influencing characteristics include economic benefit attributes and non-economic benefit attributes;

[0020] Economic benefit attributes include economic returns, cost reduction, subsidies and tax incentives; non-economic benefit attributes include corporate image, goodwill, corporate culture, administrative penalties and operating status.

[0021] As a preferred solution of the present invention, the input multimodal data is classified, trained, identified and associated, and an expert group for platform-based corporate social responsibility assessment and an expert database for storing expert group review data are constructed. Specifically:

[0022] The input multimodal data is classified, trained, identified and associated. After determining the expert group review data in the same large model, a tree table structure of the large model expert group review data is formed, and the autonomous learning model file is deployed. That is, the expert group review data in the large model is classified, trained, identified and associated, and the corresponding expert group review content library table is generated after determining the associated corporate social responsibility feature categories;

[0023] According to the review data associated with the expert database, that is, according to the current social responsibility evaluation model of platform-type enterprises and the multimodal data set, respond to the expert group review content library table, associate the expert group review data tree table structure and the corresponding expert group review data in the expert group review content library table, and determine the large model evaluation associated data with the expert group review data tree table structure and the corresponding expert group review data;

[0024] Based on the large model assessment of related data, combined with fuzzy comprehensive evaluation and empirical analysis, the result data of the social responsibility assessment of the large model expert group is generated;

[0025] The expert group includes a manual expert group and a virtual expert group.

[0026] As a preferred solution of the present invention, the multi-objective and multi-layer probability fusion LSTM model includes an input gate for determining the information that needs to be updated for the corporate social responsibility feature, an output gate for determining the corporate social responsibility feature information in the storage state for output, and a forget gate;

[0027] The above LSTM model with a new layer and the objective function are retrained. The objective function is expressed as follows:

[0028]

[0029] Where f(·) is the system function, x is the sample data of the training set, y is the sample data of the test set, θ1 is the parameter set of the shallow LSTM model, and θ2 is the parameter set of the LSTM model network at the new level. is the loss function, R is the real number domain; it is to find a set of parameters θ1 of the shallow LSTM model and a set of parameters θ2 in the new level LSTM model network in the real number domain R, so that the loss function Minimum;

[0030] The loss function Divided into the loss function of the shallow LSTM model And the loss function of the new layer LSTM model network Then the loss function It is expressed as follows:

[0031]

[0032] Among them, E is the total number of layers used by the new level LSTM model, is the loss function of the LSTM model at the eth level in the LSTM model network at the new level;

[0033] The probability scores of multiple layers are weighted and fused, and the comprehensive weight vector data results of social responsibility assessment are calculated and output, as follows:

[0034]

[0035] Among them, X is the comprehensive prediction score of social responsibility assessment; TI e For the e The confidence data of the LSTM model at each level, i.e., the probability score; W e For the e The weight of the LSTM model at each level is determined by weighted conversion of probability scores; x e The data of the training set samples of the current e-th level LSTM model.

[0036] As a preferred solution of the present invention, the blockchain adopts a consortium chain framework, that is, based on peer-to-peer networking, distributed consensus collaboration and contribution-based incentive mechanisms, a social responsibility governance model for platform-based enterprises is constructed, including block analysis, smart contract design, and the construction of consortium chains and smart contracts;

[0037] Block analysis includes: chain configuration, block header key configuration, initial account asset configuration, loading configuration and installation block;

[0038] Smart contract design includes: alliance chain functions, account-related alliance chain functions, blockchain-related alliance chain functions, and transaction-related alliance chain functions;

[0039] The establishment of alliance chain and smart contract includes: node content, block configuration, configuration organization startup, alliance chain intercommunication and smart contract function design details. The smart contract function design details include smart contract startup, wallet creation request, startup request and transaction request.

[0040] As a preferred solution of the present invention, the result data of social responsibility assessment of the large model expert group is generated by combining fuzzy comprehensive evaluation and empirical analysis, and the result data of social responsibility assessment of the large model expert group is evaluated by hesitant fuzzy sets, as follows:

[0041] The hesitant fuzzy comprehensive evaluation clearly defines the target object of the evaluation as platform-type enterprises. According to the social responsibility characteristics of platform-type enterprises, a cluster analysis is conducted on the factors and factor sets that affect platform-type enterprises.

[0042] After determining the factor set of platform enterprises, determine the corresponding evaluation level based on the characteristics of ESG evaluation indicators and the influencing characteristics of corporate social responsibility characteristics;

[0043] Based on different evaluation levels and corresponding factors, a hesitant fuzzy evaluation is performed based on the evaluation level to obtain the comprehensive hesitant fuzzy evaluation matrix of corporate social responsibility of platform-type enterprises; specifically, let X = {x1, x2, …, x m} is the set of platform enterprises to be evaluated, U = {u1,u2,…,u n} is a set of evaluation indicators for platform enterprises. According to the definitions of hesitant fuzzy sets and intuitionistic fuzzy sets, a comprehensive hesitant fuzzy matrix is ​​obtained through large-scale expert evaluation.

[0044] The comprehensive hesitant fuzzy evaluation matrix and the weight vector are fuzzy synthesized, and the membership set is specifically calculated to obtain the generalized comparison table of the hesitant fuzzy set. After comparison with the characteristic function, the row sum, column sum, and row sum and column sum difference of the generalized comparison table are calculated. After the comprehensive evaluation results of the platform-type enterprises are calculated and empirically analyzed, the result data is generated and output;

[0045] The characteristic function is defined as follows:

[0046] Among them, x and y are the row value and column value of the generalized comparison table respectively;

[0047] The calculation of the comprehensive evaluation results of platform-based enterprises is shown in the following formula:

[0048]

[0049] Among them, S i (X1), S i (X2) represents the membership score and non-membership score of the ith platform enterprise, respectively; The comprehensive evaluation score of platform-based enterprises is Sort and get the result data.

[0050] As a preferred solution of the present invention, in which: different types of virtual experiments on corporate social responsibility governance, governance scenario deductions and governance result evaluations are carried out through blockchain. Specifically, through the intuition of deep learning and the reward and punishment mechanism of reinforcement learning, combined with the valuation network of the Monte Carlo algorithm, the optimal decision is obtained and automatically or semi-automatically executed after large-scale calculation;

[0051] At the same time, when blockchain is combined with supervisory agencies, industry digital visas and public welfare funds, a collaborative optimization governance strategy is generated, and a collaborative optimization governance plan for the social responsibility of platform-based enterprises is implemented based on the collaborative optimization governance strategy.

[0052] As a preferred solution of the present invention, wherein: obtaining relevant data of platform-type enterprises, specifically, collecting relevant data of various aspects of platform-type enterprises by combining distributed web crawler Nutch and distributed search engine ElasticSearch;

[0053] The distributed web crawler Nutch and the distributed search engine ElasticSearch also include sentiment analysis on the acquired relevant data, and obtain neutral, positive and negative data through non-negative probability and negative probability.

[0054] On the one hand, the present invention provides a system for Internet platform corporate social responsibility supervision based on artificial intelligence, comprising:

[0055] The data acquisition and processing module acquires the relevant data of platform-type enterprises, and based on the characteristics of corporate social responsibility, performs feature extraction, cluster analysis and pattern recognition on the relevant data of platform-type enterprises in turn, collects the feature data to be evaluated of platform-type enterprises, performs data preprocessing, and generates the corresponding multimodal data set;

[0056] The data analysis module is used to build a social responsibility evaluation model for platform-type enterprises based on the multi-objective and multi-layer probability fusion LSTM network optimized layer by layer, dividing the training set and the test set for parameter optimization;

[0057] The data evaluation module uses the big model to call the social responsibility evaluation model of platform-type enterprises and the multimodal data of the multimodal data set to classify, train, identify and associate the input multimodal data, and build the review objectives of the social responsibility evaluation of platform-type enterprises. According to the review objectives, an expert database and a corresponding expert group are generated. After determining the evaluation index system of the social responsibility evaluation of platform-type enterprises, the fuzzy comprehensive evaluation and empirical analysis of the multi-objective and multi-layer probability fusion LSTM model are combined to generate the result data of the social responsibility evaluation of the big model expert group, export the result data, and transmit the result data into the blockchain and visualize it;

[0058] The distributed blockchain storage governance module is used to store the processed evaluation data of platform enterprises, and to build a social responsibility supervision and governance mechanism for platform enterprises based on blockchain. That is, based on peer-to-peer networking, distributed consensus collaboration, and contribution-based incentive mechanisms, a social responsibility governance model for platform enterprises is constructed. Based on the combination of smart contracts and the digital RMB system, different types of corporate social responsibility governance virtual experiments, governance scenario simulations, and governance results evaluations are carried out through blockchain.

[0059] Beneficial effects of the invention: The invention establishes a social responsibility evaluation system for platform-type enterprises based on the long short-term memory network LSTM, performs supervision and governance through blockchain updates and replacement of data, and designs multi-layer and multi-objective machine learning methods and constructs a social responsibility evaluation model for platform-type enterprises. It can accurately express and efficiently analyze and process the relationship between goals and between goals and scenarios, which is beneficial to subsequent evaluation and governance work. In summary, the invention not only improves the monitoring and evaluation effect of the social responsibility of platform-type enterprises, but also realizes the collaborative governance of multiple subjects. BRIEF DESCRIPTION OF THE DRAWINGS

[0060] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the following briefly introduces the drawings required for describing the embodiments. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without creative labor. Among them:

[0061] Figure 1 This is a flow chart of an artificial intelligence-based Internet enterprise social responsibility assessment and governance method in Embodiment 1 of the present invention;

[0062] Figure 2 It is a schematic diagram of the vertical layout lanes of the Internet enterprise social responsibility assessment and governance method based on artificial intelligence in the first embodiment of the present invention;

[0063] Figure 3This is a flow chart of constructing a social responsibility evaluation model for a platform-based enterprise in the first embodiment of the present invention;

[0064] Figure 4 This is a schematic diagram of the process of generating result data through fuzzy comprehensive evaluation and empirical analysis in the first embodiment of the present invention;

[0065] Figure 5 This is a technical route flow chart for building a platform-based corporate social responsibility supervision and governance mechanism based on blockchain in Embodiment 1 of the present invention;

[0066] Figure 6 This is a schematic diagram of the modular structure of a system for corporate social responsibility supervision on an Internet platform based on artificial intelligence in Embodiment 2 of the present invention;

[0067] Numbers in the figure: 10, data acquisition and processing module; 20, data analysis module; 30, data evaluation module; 40, distributed blockchain storage governance module. DETAILED DESCRIPTION

[0068] In order to make the purpose, technical solution and advantages of the embodiment of the present invention clearer, the technical solution of the embodiment of the present invention will be clearly and completely described below in conjunction with the drawings of the embodiment of the present invention. Obviously, the described embodiment is a part of the embodiment of the present invention, not all of the embodiments. Based on the described embodiment of the present invention, all other embodiments obtained by ordinary technicians in this field belong to the scope of protection of the present invention.

[0069] Reference Figure 1-Figure 5 , is an embodiment of the present invention, which provides an artificial intelligence-based Internet enterprise social responsibility assessment and governance method provided by the present invention. First, based on the international ESG, the review objectives are established, and then an expert database is established based on a large model formed by domestic and foreign experts and virtual experts. Then, the corresponding expert group is selected to determine the evaluation indicators and evaluation system of corporate social responsibility, and then the hesitant fuzzy comprehensive evaluation is combined to draw conclusions. At the same time, a long-term cyclic evaluation demonstration is carried out by combining multi-objective and multi-layer probabilistic neural networks. Finally, the alliance chain is used to build the corresponding smart contract governance, put forward corresponding suggestions and governance mechanisms, and combine with China's digital RMB system to further improve the incentive mechanism and effect evaluation of social responsibility contributions; refer to Figure 1 and Figure 2 As shown:

[0070] Step S1, obtaining relevant data of platform-type enterprises, and based on the characteristics of corporate social responsibility, sequentially performing feature extraction, cluster analysis and pattern recognition on the relevant data of platform-type enterprises, grouping them into feature data to be evaluated for platform-type enterprises, and performing data preprocessing to generate corresponding multimodal data sets;

[0071] Step S2, based on the multi-objective and multi-layer probability fusion LSTM network optimized layer by layer, the training set and the test set are divided for parameter optimization, and then a social responsibility evaluation model for platform-type enterprises is constructed;

[0072] Step S3, responding to the social responsibility evaluation model of the platform-type enterprise, generating result data through fuzzy comprehensive evaluation and empirical analysis, exporting the result data and transmitting the result data into the blockchain and visually presenting it. Specifically, the social responsibility evaluation model of the platform-type enterprise and the multimodal data set data are called by the big model, and the input multimodal data is classified, trained, identified and associated, and an expert group for social responsibility evaluation of the platform-type enterprise and an expert database for storing the expert group review data are constructed; based on the review data associated with the expert database, the result data of the social responsibility evaluation of the big model expert group is generated in combination with fuzzy comprehensive evaluation and empirical analysis, and the result data is exported and the result data is transmitted to the blockchain and visually presenting it;

[0073] Step S4, build a platform-type enterprise social responsibility supervision and governance mechanism based on blockchain, that is, build a social responsibility governance model for platform-type enterprises based on peer-to-peer networking, distributed consensus collaboration, and contribution-based incentive mechanisms. Based on the combination of smart contracts and the digital RMB system, different types of corporate social responsibility governance virtual experiments, governance scenario simulations, and governance results evaluations are carried out through blockchain.

[0074] In one implementation, preferably, the relevant data of platform-type enterprises are obtained. Specifically, the relevant data of various aspects of platform-type enterprises are collected by combining the distributed web crawler Nutch and the distributed search engine ElasticSearch; in addition, the distributed web crawler Nutch and the distributed search engine ElasticSearch also include sentiment analysis of the relevant data obtained, and neutral, positive and negative data are obtained through non-negative probability and negative probability. Based on the above, it can be seen that through the distributed web crawler Nutch and the distributed search engine ElasticSearch, with social responsibility evaluation CSR as the theme, the data of various aspects of platform-type enterprises are collected, and the idea of ​​combining enterprise data with social data, domestic data with international data, online and offline data, and statistical data with non-statistical data is adopted. The financial reports and social responsibility reports published by enterprises are important references for evaluation. For the parts that cannot be effectively covered, the official public social data and Internet data are used. Official statistical data are reliable, authoritative, quantitative and systematic, while non-statistical data represented by Internet data are effective and have complete samples. They are an effective supplement to traditional statistical data. In addition, the data is written into the blockchain after processing to prevent tampering.

[0075] In one implementation, the corporate social responsibility characteristics specifically include ESG evaluation indicator characteristics;

[0076] The ESG evaluation index characteristics are divided into primary attributes and secondary attributes. The primary attributes of ESG evaluation index characteristics include environmental issues, social concerns and corporate governance. Environmental issues are climate change and sustainability category characteristics. Corporate governance is diversity and consumer protection category characteristics. Corporate governance attributes are management structure, executive compensation and employee compensation category characteristics. The secondary attributes of ESG evaluation index characteristics include social public services, environmental protection, scientific and technological innovation, cyberspace security and expanding open cooperation.

[0077] Corporate social responsibility characteristics also include corporate social responsibility characteristics influencing characteristics, and corporate social responsibility characteristics influencing characteristics include economic benefit attributes and non-economic benefit attributes;

[0078] Economic benefit attributes include financial returns, cost reductions, subsidies, and tax benefits;

[0079] Non-economic benefit attributes include corporate image, goodwill, corporate culture, administrative penalties and operating status.

[0080] Based on the above, it can be seen that the embodiment uses deep learning to adaptively mark, classify and process behavioral big data such as daring technological innovation, maintaining public services, expanding open cooperation, promoting network security development, employee rights and interests, and green energy conservation, that is, adaptively marking, classifying and processing from secondary attributes to primary attributes in turn. That is, this implementation builds a platform-type enterprise social responsibility evaluation system based on the ESG concept, and conducts all-round and multi-level evaluation and empirical analysis based on big data, artificial intelligence, and blockchain, thereby being able to construct a platform-type enterprise social responsibility model.

[0081] To further illustrate, the model of this embodiment can not only describe the goals and scenarios in socially responsible investment, but also express the relationship between goals and goals, and between goals and scenarios, and can generate strategies that comply with legal regulations, principles and social norms; and provide decision support.

[0082] Reference Figure 3 and Figure 4 In one implementation, preferably, in order to adapt to the data optimization of the above-mentioned multi-layer and multi-objective attributes, this embodiment is based on a multi-objective and multi-layer probability fusion LSTM model optimized layer by layer, and the specific steps of constructing a social responsibility evaluation model for platform-type enterprises are as follows:

[0083] Step S201, using a shallow LSTM model for training, and retaining the classification layer and objective function in the original LSTM model after the training converges;

[0084] Step S202, adding a new level of LSTM model and objective function to retrain the model, and fine-tuning the original parameters of the social responsibility evaluation model of platform-type enterprises;

[0085] Step S203, transforming the Softmax functions used by different classification layers to obtain the predicted probability score of social responsibility assessment at each layer;

[0086] Step S204, weighted fusion of the multi-layer probability scores, calculation of the comprehensive prediction score of the social responsibility assessment and output after determining it as result data.

[0087] The present embodiment needs to be further explained that the multi-objective and multi-layer probability fusion LSTM model includes an input gate for determining the information that needs to be updated for the corporate social responsibility feature, an output gate for determining the corporate social responsibility feature information in the storage state for output, and a forget gate;

[0088] The above LSTM model with a new layer and the objective function are retrained. The objective function is expressed as follows:

[0089]

[0090] Where f(·) is the system function, x is the sample data of the training set, y is the sample data of the test set, θ1 is the parameter set of the shallow LSTM model, and θ2 is the parameter set of the LSTM model network at the new level. is the loss function, R is the real number domain; it is to find a set of parameters θ1 of the shallow LSTM model and a set of parameters θ2 in the new level LSTM model network in the real number domain R, so that the loss function Minimum;

[0091] The loss function Divided into the loss function of the shallow LSTM model And the loss function of the new layer LSTM model network Then the loss function It is expressed as follows:

[0092]

[0093] Among them, n is the total number of layers used by the new level LSTM model, is the loss function of the LSTM model of the i-th level in the LSTM model network of the new level.

[0094] Based on the above, it can be seen that the scientific benefit return mechanism of this embodiment can stimulate the potential of corporate social responsibility investment. This is a multi-objective optimization problem. From the perspective of artificial intelligence, the model of this embodiment can not only describe the goals and scenarios in social responsibility investment, but also accurately express and efficiently analyze and process the relationship between goals and goals, and between goals and scenarios, which is conducive to subsequent evaluation and governance, and can generate strategies that comply with legal regulations, principles and social norms.

[0095] In one implementation, this embodiment further states that the social responsibility evaluation model of the platform-type enterprise and the multimodal data set data are called through the big model, wherein the big model includes, for example, the Spark big model, the Wenxin Yiyan big model, the Zidong Taichu big model, the ERNIE big model and the GTP-4 big model, etc.; the input multimodal data can be classified, trained, identified and associated, and an expert group for social responsibility evaluation of the platform-type enterprise and an expert database for storing the expert group review data are constructed, wherein the expert group includes a manual expert group and a virtual expert group, specifically:

[0096] The input multimodal data is classified, trained, identified and associated. After determining the review data of multiple expert groups in the same macro model or the review data of different expert groups under different macro models, a tree table structure of the review data of multiple expert groups in the same macro model or a tree table structure of the review data of expert groups crossing different macro models is formed, and the autonomous learning model file is deployed, that is, the review data of multiple expert groups in the same macro model or the review data of different expert groups under different macro models are classified, trained, identified and associated, and the corresponding expert group review content library table is generated after determining the category of the associated corporate social responsibility characteristics;

[0097] According to the review data associated with the expert database, that is, according to the current social responsibility evaluation model of platform-type enterprises and the multimodal data set, respond to the expert group review content library table, associate the expert group review data tree table structure and the corresponding expert group review data in the expert group review content library table, and determine the large model evaluation associated data with the expert group review data tree table structure and the corresponding expert group review data;

[0098] Based on the big model assessment related data, and combined with fuzzy comprehensive evaluation and empirical analysis, the result data of the big model expert group's social responsibility assessment is generated.

[0099] Reference Figure 5 As shown, in one implementation, this embodiment combines fuzzy comprehensive evaluation and empirical analysis to generate the result data of social responsibility evaluation of the large model expert group, and specifically uses hesitant fuzzy sets to evaluate the result data of social responsibility evaluation of the large model expert group, as follows:

[0100] Step 301: Hesitant fuzzy comprehensive evaluation specifies that the target object of evaluation is platform-type enterprises, and based on the social responsibility characteristics of platform-type enterprises, cluster analysis is performed to identify factors and factor sets that affect platform-type enterprises;

[0101] Step 302: After determining the factor set of the platform enterprise, determine the corresponding evaluation level according to the characteristics of the ESG evaluation indicators and the characteristics of the corporate social responsibility characteristics;

[0102] Step 303: Based on different evaluation levels and corresponding factors, a hesitant fuzzy evaluation is performed based on the evaluation level to obtain a comprehensive hesitant fuzzy evaluation matrix of corporate social responsibility of platform-type enterprises; specifically, let X = {x1, x2, …, x m} is the set of platform enterprises to be evaluated, U = {u1,u2,…,u n} is a set of evaluation indicators for platform enterprises. According to the definitions of hesitant fuzzy sets and intuitionistic fuzzy sets, a comprehensive hesitant fuzzy matrix is ​​obtained through large-scale expert evaluation.

[0103] Step 304: weighted fusion of the multi-layer probability scores, calculate the comprehensive weight vector data result of social responsibility assessment and output it, as follows:

[0104]

[0105] Among them, X is the comprehensive prediction score of social responsibility assessment; TI e For the e The confidence data of the LSTM model at each level, i.e., the probability score; W e For the e The weight of the LSTM model at each level is determined by weighted conversion of probability scores; x e The data of the training set samples of the current e-th level LSTM model;

[0106] Step 305: Fuzzy synthesis is performed on the comprehensive hesitant fuzzy evaluation matrix and the weight vector, specifically, the membership set is calculated to obtain a generalized comparison table of the hesitant fuzzy set, and the row sum, column sum, and row sum and column sum difference of the generalized comparison table are calculated after comparison using the characteristic function, and the comprehensive evaluation results of the platform-type enterprise are calculated and empirically analyzed, and the result data is generated and output;

[0107] The characteristic function is defined as follows:

[0108] Among them, x and y are the row value and column value of the generalized comparison table respectively;

[0109] The calculation of the comprehensive evaluation results of platform-based enterprises is shown in the following formula:

[0110]

[0111] Among them, S i (X1), S i (X2) represents the membership score and non-membership score of the ith platform enterprise, respectively; The comprehensive evaluation score of platform-based enterprises is Sorting is performed to obtain the result data. In addition, the characteristic of the hesitant fuzzy comprehensive evaluation model is that its evaluation results are not absolutely affirmative or negative, but are expressed as a hesitant fuzzy set. This gives the model an advantage in dealing with uncertainty problems and can better reflect the real situation of the platform-type enterprises being evaluated regarding the ESG review objectives and review targets. At the same time, the model also has the characteristics of clear results and strong systematicity, and can comprehensively consider the weights and membership of various indicators to obtain a comprehensive evaluation result.

[0112] Based on the above, this example uses a combination of qualitative and quantitative research methods to conduct an empirical evaluation of the social responsibility of Internet platform companies, such as e-commerce platforms. Through case analysis of multiple e-commerce platforms, we combine the operating characteristics of e-commerce platforms to identify their key areas of social responsibility, such as protecting consumer rights, promoting employment and training, participating in social welfare, protecting the environment, etc. On this basis, we have constructed a multi-dimensional social responsibility evaluation system that takes into account both the economic performance of e-commerce platforms and their contributions to society and the environment.

[0113] Reference Figure 5 In one implementation, preferably, different types of corporate social responsibility governance virtual experiments, governance scenario simulations and governance results evaluations are carried out through blockchain. Specifically, through the intuition of deep learning and the reward and punishment mechanism of reinforcement learning, combined with the valuation network of the Monte Carlo algorithm, the optimal decision is obtained and automatically or semi-automatically executed after large-scale calculations; at the same time, after the blockchain is combined with supervisory agencies, industry digital visas and public welfare funds, a collaborative optimization governance strategy is generated, and a collaborative optimization governance plan for social responsibility of platform-type enterprises is carried out according to the collaborative optimization governance strategy.

[0114] Specifically, in this embodiment, the blockchain adopts a consortium chain framework, that is, based on peer-to-peer networking, distributed consensus collaboration, and contribution-based incentive mechanisms, a social responsibility governance model for platform-based enterprises is constructed, including block analysis, smart contract design, and the construction of consortium chains and smart contracts;

[0115] Block analysis includes: chain configuration, block header key configuration, initial account asset configuration, loading configuration and installation block;

[0116] Among them, the chain configuration is the configuration information of the Config chain and the consensus protocol, the block header focuses on the gasLimit configuration, which refers to the upper limit of the currency and the difficulty, the initial account asset configuration is that in the test environment, several accounts can be configured for Alloc, and the accounts that will be produced later can be stored in the account in advance. The alliance chain focuses on transaction testing; the loading configuration is that different network states can be provided through command parameter configuration when starting geth. When geth starts, it will create the corresponding network environment based on the parameters and network configuration. If these parameters are not filled in, the default main network configuration of geth will be used. The installation block will start by reading the hash address of the block from the database. If it cannot be found, a block will be created. The first run also requires a new block configuration before it can be executed. If you need to compare whether the block hash has been stored, the block hash is consistent. If there is an error, an error message will be returned, and then the chain configuration information will be checked. Get the chain configuration from the database. If it cannot be found, there is no need to check and use the chain configuration information. If the query is successful, you need to verify whether the running chain configuration information is normal. Only when it is correct will the data be updated and replaced.

[0117] Smart contract design includes: alliance chain functions, account-related alliance chain functions, blockchain-related alliance chain functions, and transaction-related alliance chain functions;

[0118] The construction of alliance chain and smart contract includes: node content, block configuration, configuration organization startup, alliance chain intercommunication and smart contract function design details; the smart contract function design details specifically include:

[0119] Account-related API design, 1. The process of creating a new wallet account is: randomly generate a set of mnemonics, generate a seed based on the mnemonics, generate ECKeyPair based on the seed, create a wallet based on ECKeyPair and password, and then check whether the generated wallet is correct. 2. When querying an account, the wallet data will be imported into Rocks DB at the moment of loading the port. We only need to find the account in DB and then show it to the user. 3. The principle of querying all accounts is the same as that of querying a single account.

[0120] Blockchain-related API design, 1. Mining blocks: first get the last block and then add a transaction to the block, remove the transaction pool, and then store the block using asynchronous events, waiting for other nodes to confirm the block. 2. Browse the header block: extract data directly from the database. 3. Add a new node: submit data to the client and database. 4. Query the node list: query all node data in the database.

[0121] Transaction-related API design: 1. To send a transaction, the account balance will be verified to see if it meets the transaction conditions. If it meets the conditions, it will be submitted to the block, and the payment and receiving addresses will be verified. The transaction object will be constructed and the sign signature will be generated. Then the sign private key signature will be verified. If it is correct, it will be added to the transaction pool (the transaction pool will also determine whether the transaction exists), waiting for packaging. After the transaction practice is triggered, it will initiate a full network broadcast and wait for confirmation. 2. To query the transaction status, use the submitted hash query to query the transaction pool first and then query the block information in the database.

[0122] The smart contract function design details include smart contract startup, wallet creation request, startup request and transaction request. When the smart contract project is started, it will check the configuration port data information written in the properties configuration file, and then obtain the alliance chain data through the port information. When it runs for the first time, it will create an address and a default wallet address, and load the data into the database;

[0123] Create a wallet request. After receiving the API request, the smart contract will call AccountController to create a wallet account. When the class is created, two classes will be loaded: database control class (DBAccess) and account class (personal). The newAccount method in the class is used to create a new wallet user and return Json data (JsonVo class). The following figure details the creation process, the main part of which is the production of KeyPair, which is the public key and private key of Ethereum;

[0124] Start the request. After receiving the API request, the smart contract will call the BlokcController class. When the class is created, two classes will be loaded: the database control class (DBAccess) and the account class (personal). The mining method in the class is used to execute the request user and return Json data (JsonVo class).

[0125] After receiving the API request, the smart contract will call the TransactionController class. When the class is created, four classes will be loaded: database control class (DBAccess), account class (personal), AppConfig (configuration class), and TransactionPool (transaction class). The sentTransaction method in the class is used to execute the transaction request, and the transaction (transaction object) is returned to encapsulate the data into Json data (JsonVo class). The figure below details the request implementation process).

[0126] Based on the above, we can know that blockchain data is time-stamped, verified and recorded by consensus nodes, and cannot be tampered with or forged, making it widely used in CSR data notarization and auditing scenarios. For example, blockchain can permanently and securely store commendations, awards, certificates, registration forms, licenses, certifications, and records issued by government agencies, and can easily prove the existence and authenticity of a certain data at any time, and conduct real-time audits at low cost and high efficiency.

[0127] Embodiment 2

[0128] Reference Figure 6 , is an embodiment of the present invention, which provides a system for corporate social responsibility supervision on an Internet platform based on artificial intelligence based on the method of the above embodiment, including:

[0129] The data acquisition and processing module 10 acquires the relevant data of the platform-type enterprises, and based on the corporate social responsibility characteristics, sequentially performs feature extraction, cluster analysis and pattern recognition on the relevant data of the platform-type enterprises, collects the feature data to be evaluated of the platform-type enterprises, performs data preprocessing, and generates a corresponding multimodal data set;

[0130] Data analysis module 20, used to build a social responsibility evaluation model for platform-type enterprises based on a multi-objective and multi-layer probability fusion LSTM network optimized layer by layer, dividing the training set and the test set for parameter optimization;

[0131] The data evaluation module 30 uses the big model to call the social responsibility evaluation model of the platform-type enterprise and the multimodal data of the multimodal data set, classifies, trains, identifies and associates the input multimodal data, and constructs the review objectives of the social responsibility evaluation of the platform-type enterprise. According to the review objectives, an expert database and a corresponding expert group are generated. After determining the evaluation index system of the social responsibility evaluation of the platform-type enterprise, the fuzzy comprehensive evaluation and empirical analysis of the multi-objective and multi-layer probability fusion LSTM model are combined to generate the result data of the social responsibility evaluation of the big model expert group, and the result data is exported and the result data is transmitted and written into the blockchain for visualization;

[0132] The distributed blockchain storage governance module 40 is used to store the processed evaluation data of platform-type enterprises, and to build a social responsibility supervision and governance mechanism for platform-type enterprises based on the blockchain, that is, to build a social responsibility governance model for platform-type enterprises based on peer-to-peer networking, distributed consensus collaboration and contribution-based incentive mechanisms, in which different types of corporate social responsibility governance virtual experiments, governance scenario simulations and governance results evaluations are carried out through the blockchain.

[0133] In summary, this embodiment provides an evaluation system for Internet platform enterprises with modern governance. Combining information from multiple ports such as Internet users, social supervision agencies, and platform-type enterprises, it can construct a platform-type enterprise social responsibility evaluation model, enrich and expand the perspective of corporate social responsibility CSR evaluation, and enrich the theory of corporate social responsibility CSR governance modernization; it builds an Internet platform-type enterprise social responsibility model, establishes a platform-type enterprise social responsibility evaluation system based on the long short-term memory network LSTM, and supervises and governs by updating and replacing data through blockchain, which not only improves the monitoring and evaluation effect of platform-type enterprise social responsibility, but also realizes the collaborative governance of multiple subjects. In addition, the present invention shares and promotes the evaluation system and blockchain governance architecture of Internet enterprise social responsibility CSR, contributes wisdom to the creation of a new pattern of social responsibility investment, and has practical application value for the development of Internet enterprises and society.

[0134] In the above embodiments, it can be implemented in whole or in part by software, hardware, firmware or any combination thereof. When implemented using software, it can be implemented in whole or in part in the form of a computer program product. The computer program product includes one or more computer instructions. When the computer program instructions are loaded and executed on a computer, the process or function according to the present application is generated in whole or in part. The computer can be a general-purpose computer, a special-purpose computer, a computer network, or other programmable devices. The computer instructions can be stored in a computer-readable storage medium, or transmitted from one computer-readable storage medium to another computer-readable storage medium.

[0135] In the description of this specification, the description with reference to the terms "one embodiment", "some embodiments", "example", "specific example", or "some examples" means that the specific features, structures, materials or characteristics described in conjunction with the embodiment or example are included in at least one embodiment or example of the present application. Moreover, the specific features, structures, materials or characteristics described may be combined in any one or more embodiments or examples in a suitable manner. In addition, those skilled in the art may combine and combine different embodiments or examples described in this specification and the features of different embodiments or examples, unless they are contradictory.

[0136] In addition, the terms "first" and "second" are used for descriptive purposes only and should not be understood as indicating or implying relative importance or implicitly indicating the number of technical features indicated. Therefore, a feature defined as "first" or "second" may explicitly or implicitly include at least one of the features. In the description of this application, the meaning of "plurality" is two or more, unless otherwise clearly and specifically defined.

[0137] Any process or method description in the flow chart or otherwise described herein can be understood to represent a module, fragment or portion of a code including one or more executable instructions for implementing the steps of a specific logical function or process. And the scope of the preferred embodiment of the present application includes other implementations, in which the functions may not be performed in the order shown or discussed, including in a substantially simultaneous manner or in a reverse order according to the functions involved.

[0138] The logic and / or steps represented in the flowchart or otherwise described herein, for example, can be considered as an ordered list of executable instructions for implementing logical functions, which can be embodied in any computer-readable medium for use by an instruction execution system, apparatus or device (such as a computer-based system, a system including a processor or other system that can fetch instructions from an instruction execution system, apparatus or device and execute instructions), or used in combination with these instruction execution systems, apparatuses or devices.

[0139] It should be understood that the various parts of the present application can be implemented with hardware, software, firmware or a combination thereof. In the above embodiments, multiple steps or methods can be implemented with software or firmware stored in a memory and executed by a suitable instruction execution system. All or part of the steps of the above embodiment method can be completed by instructing the relevant hardware through a program, which can be stored in a computer-readable storage medium, and when the program is executed, it includes one of the steps of the method embodiment or a combination thereof.

[0140] In addition, each functional unit in each embodiment of the present application can be integrated into a processing module, or each unit can exist physically separately, or two or more units can be integrated into one module. The above-mentioned integrated module can be implemented in the form of hardware or in the form of a software functional module. If the above-mentioned integrated module is implemented in the form of a software functional module and sold or used as an independent product, it can also be stored in a computer-readable storage medium. The storage medium can be a read-only memory, a disk or an optical disk, etc.

[0141] The above are only specific implementations of the present application, but the protection scope of the present application is not limited thereto. Any technician familiar with the technical field can easily think of various changes or substitutions within the technical scope disclosed in the present application, which should be included in the protection scope of the present application. Therefore, the protection scope of the present application should be based on the protection scope of the claims.

Claims

1. An AI-based social responsibility assessment and governance method for Internet companies. It is characterized in that include: Obtain relevant data of platform-type enterprises, and based on the characteristics of corporate social responsibility, perform feature extraction, cluster analysis and pattern recognition on the relevant data of the platform-type enterprises in turn, aggregate them into feature data to be evaluated for the platform-type enterprises, perform data preprocessing, and generate corresponding multimodal data sets; Based on the multi-objective and multi-layer probability fusion LSTM network optimized layer by layer, the social responsibility evaluation model of the platform-based enterprise is constructed after dividing the training set and the test set for parameter optimization; The social responsibility evaluation model of the platform-type enterprise and the multimodal data of the multimodal data set are called by the big model to classify, train, identify and associate the input multimodal data, and construct the review objectives of the social responsibility evaluation of the platform-type enterprise. According to the review objectives, an expert database and a corresponding expert group are generated. After determining the evaluation index system of the social responsibility evaluation of the platform-type enterprise, the result data of the social responsibility evaluation of the big model expert group is generated by combining the fuzzy comprehensive evaluation and empirical analysis of the multi-objective and multi-layer probability fusion LSTM model, and the result data is exported and the result data is transmitted and written into the blockchain for visualization; Build a platform-type enterprise social responsibility supervision and governance mechanism based on blockchain, that is, build a social responsibility governance model for platform-type enterprises based on peer-to-peer networking, distributed consensus collaboration and contribution-based incentive mechanism. Based on the combination of smart contracts and the digital RMB system, different types of corporate social responsibility governance virtual experiments, governance scenario simulations and governance results evaluations are carried out through blockchain.

2. The method for evaluating and managing the social responsibility of Internet companies based on artificial intelligence as claimed in claim 1, It is characterized in that Based on the multi-objective and multi-layer probability fusion LSTM model optimized layer by layer, the specific steps of constructing the social responsibility evaluation model of the platform-based enterprise are as follows: Use a shallow LSTM model for training until the training converges, and retain the classification layer and objective function in the original LSTM model; Add a new layer of LSTM model and objective function to retrain the model, and fine-tune the original parameters of the social responsibility evaluation model of the platform-type enterprise; The Softmax functions used in different classification layers are transformed to obtain the predicted probability scores of social responsibility assessment at each layer; The probability scores of multiple layers are weighted and fused to calculate the comprehensive prediction score of social responsibility assessment and output it as result data.

3. The method for evaluating and managing the social responsibility of Internet companies based on artificial intelligence as claimed in claim 1, It is characterized in that The review objectives include ESG review objectives; the corporate social responsibility characteristics include ESG evaluation indicator characteristics; the result data include ESG evaluation result data, and the ESG evaluation result data is used for blockchain, ESG investment and smart contracts; The ESG evaluation indicator characteristics are divided into primary attributes and secondary attributes. The primary attributes of the ESG evaluation indicator characteristics include environmental issues, social concerns and corporate governance. The environmental issues are climate change and sustainability category characteristics. The corporate governance is diversity and consumer protection category characteristics. The corporate governance attributes are management structure, executive compensation and employee compensation category characteristics. The secondary attributes of the ESG evaluation indicator characteristics include social public services, environmental protection, scientific and technological innovation, cyberspace security and expanding open cooperation. The corporate social responsibility characteristics also include corporate social responsibility characteristics influencing characteristics, and the corporate social responsibility characteristics influencing characteristics include economic benefit attributes and non-economic benefit attributes; The economic benefit attributes include economic returns, cost reduction, subsidies and tax incentives; the non-economic benefit attributes include corporate image, goodwill, corporate culture, administrative penalties and operating status.

4. The method for evaluating and managing the social responsibility of Internet companies based on artificial intelligence as claimed in claim 1, It is characterized in that Classify, train, identify and associate the input multimodal data, build an expert group for platform-based corporate social responsibility assessment and an expert database for storing the expert group's review data, specifically: The input multimodal data is classified, trained, identified and associated, and after determining the expert group review data in the same large model, a tree table structure of the large model expert group review data is formed, and the autonomous learning model file is deployed, that is, the expert group review data in the large model is classified, trained, identified and associated, and the corresponding expert group review content library table is generated after determining the associated corporate social responsibility feature category; According to the review data associated with the expert database, that is, according to the current social responsibility evaluation model of the platform-type enterprise and the multimodal data set, respond to the expert group review content library table, associate the expert group review data tree table structure and the corresponding expert group review data in the expert group review content library table, and determine the large model evaluation associated data with the expert group review data tree table structure and the corresponding expert group review data; Based on the large model assessment associated data, combined with fuzzy comprehensive evaluation and empirical analysis, the result data of the large model expert group social responsibility assessment is generated; The expert group includes a manual expert group and a virtual expert group.

5. The method for evaluating and managing the social responsibility of Internet companies based on artificial intelligence as claimed in claim 2, It is characterized in that The multi-objective and multi-layer probability fusion LSTM model includes an input gate for determining the corporate social responsibility feature information that needs to be updated, an output gate for determining the corporate social responsibility feature information in the storage state for output, and a forget gate; The LSTM model and objective function with a new layer are added to retrain the model. The objective function is expressed as follows: Among them, f(·) is the system function, x is the sample data of the training set, y is the sample data of the test set, θ 1 is the parameter set of the shallow LSTM model, θ 2 is the parameter set in the new layer LSTM model network, is the loss function, R is the real number domain; it is to find a set of parameters θ of the shallow LSTM model in the real number domain R 1 And the parameter set θ in the new layer LSTM model network 2 , so that the loss function Minimum; The loss function Divided into the loss function of the shallow LSTM model And the loss function of the new layer LSTM model network Then the loss function It is expressed as follows: Among them, E is the total number of layers used by the new level LSTM model, is the loss function of the LSTM model at the eth level in the LSTM model network at the new level; The probability scores of multiple layers are weighted and fused, and the comprehensive weight vector data results of social responsibility assessment are calculated and output.

6. The method for evaluating and managing the social responsibility of Internet companies based on artificial intelligence as claimed in claim 1, It is characterized in that The blockchain adopts the alliance chain framework, that is, based on peer-to-peer networking, distributed consensus collaboration and contribution-based incentive mechanisms, to build a social responsibility governance model for platform-based enterprises, including block analysis, smart contract design, and alliance chain and smart contract construction; The block analysis includes: chain configuration, block header key configuration, initial account asset configuration, loading configuration and installation block; The smart contract design includes: alliance chain functions, account-related alliance chain functions, blockchain-related alliance chain functions, and transaction-related alliance chain functions; The alliance chain and smart contract construction includes: node content, block configuration, configuration organization startup, alliance chain intercommunication and smart contract function design details. The smart contract function design details include smart contract startup, wallet creation request, startup request and transaction request.

7. The method for evaluating and managing the social responsibility of Internet companies based on artificial intelligence as claimed in claim 4, It is characterized in that Combining fuzzy comprehensive evaluation and empirical analysis to generate the result data of social responsibility assessment of the large model expert group, the hesitant fuzzy set is used to evaluate the result data of social responsibility assessment of the large model expert group, as follows: The hesitant fuzzy comprehensive evaluation clearly defines the target object of the evaluation as platform-type enterprises. According to the social responsibility characteristics of platform-type enterprises, a cluster analysis is conducted on the factors and factor sets that affect platform-type enterprises. After determining the factor set of platform enterprises, determine the corresponding evaluation level based on the characteristics of ESG evaluation indicators and the influencing characteristics of corporate social responsibility characteristics; Based on the different evaluation levels and corresponding factors, a hesitant fuzzy evaluation is performed based on the evaluation level to obtain a comprehensive hesitant fuzzy evaluation matrix of corporate social responsibility of platform-type enterprises; specifically, let X = {x 1 ,x 2 ,…,x m } is the set of platform enterprises to be evaluated, U = {u 1 ,u 2 ,…,u n } is a set of evaluation indicators for platform enterprises. According to the definitions of hesitant fuzzy sets and intuitionistic fuzzy sets, a comprehensive hesitant fuzzy matrix is ​​obtained through large-scale expert evaluation. The comprehensive hesitant fuzzy evaluation matrix and the weight vector are fuzzy synthesized, and the membership set is specifically calculated to obtain a generalized comparison table of the hesitant fuzzy set. After comparison with the generalized comparison table using the characteristic function, the row sum, column sum, and row sum and column sum difference are calculated, and the comprehensive evaluation results of the platform-type enterprises are calculated and empirically analyzed, and then the result data is generated and output; The characteristic function is defined as follows: Among them, x and y are the row value and column value of the generalized comparison table respectively; The calculation of the comprehensive evaluation results of platform-based enterprises is shown in the following formula: Among them, S i (X 1 ), S i (X 2 ) represent the membership score and non-membership score of the ith platform enterprise respectively; The comprehensive evaluation score of platform-based enterprises is Sort and get the result data.

8. The method for evaluating and managing the social responsibility of Internet companies based on artificial intelligence as claimed in claim 1, It is characterized in that Conduct different types of virtual experiments on corporate social responsibility governance, governance scenario simulations, and governance results evaluations through blockchain. Specifically, through the intuition of deep learning and the reward and punishment mechanism of reinforcement learning, combined with the valuation network of the Monte Carlo algorithm, the optimal decision is obtained and automatically or semi-automatically executed after large-scale calculations; At the same time, after blockchain is combined with supervisory agencies, industry digital visas and public welfare funds, a collaborative optimization governance strategy is generated, and a collaborative optimization governance plan for the social responsibility of platform-based enterprises is implemented based on the collaborative optimization governance strategy.

9. The method for evaluating and managing the social responsibility of Internet companies based on artificial intelligence as claimed in claim 1, It is characterized in that Obtain relevant data of platform-based enterprises. Specifically, collect relevant data of various aspects of platform-based enterprises through a combination of distributed web crawlers, distributed search engines, questionnaire surveys, and empirical analysis; The distributed web crawler and the distributed search engine also include performing sentiment analysis on the acquired relevant data, and obtaining neutral, positive and negative data through non-negative probability and negative probability.

10. The system of the Internet enterprise social responsibility assessment and governance method based on artificial intelligence as claimed in claim 1, It is characterized in that include: A data acquisition and processing module (10) acquires relevant data of platform-type enterprises, and based on the characteristics of corporate social responsibility, sequentially performs feature extraction, cluster analysis and pattern recognition on the relevant data of the platform-type enterprises, collects the feature data to be evaluated of the platform-type enterprises, performs data preprocessing, and generates a corresponding multimodal data set; A data analysis module (20) is used to construct a social responsibility evaluation model for the platform-based enterprise after dividing the training set and the test set for parameter optimization based on a multi-objective and multi-layer probability fusion LSTM network optimized layer by layer; The data evaluation module (30) uses the big model to call the social responsibility evaluation model of the platform-type enterprise and the multimodal data of the multimodal data set, classifies, trains, identifies and associates the input multimodal data, and constructs the evaluation objectives of the social responsibility evaluation of the platform-type enterprise. According to the evaluation objectives, an expert database and a corresponding expert group are generated. After determining the evaluation index system of the social responsibility evaluation of the platform-type enterprise, the fuzzy comprehensive evaluation and empirical analysis of the multi-objective and multi-layer probability fusion LSTM model are combined to generate the result data of the social responsibility evaluation of the big model expert group, and the result data is exported and the result data is transmitted and written into the blockchain for visual presentation; The distributed blockchain storage governance module (40) is used to store the processed evaluation data of platform-type enterprises and to build a social responsibility supervision and governance mechanism for platform-type enterprises based on the blockchain, that is, to build a social responsibility governance model for platform-type enterprises based on peer-to-peer networking, distributed consensus collaboration and contribution-based incentive mechanisms. Based on the combination of smart contracts and the digital RMB system, different types of corporate social responsibility governance virtual experiments, governance scenario simulations and governance results evaluations are carried out through the blockchain.

Images