ESG rating indicator matching method, device, electronic device and storage medium

Abstract

The present application relates to the field of Internet technology, and provides an indicator matching method, device, electronic device, and storage medium for ESG ratings, the method comprising: obtaining a file uploaded through an ESG rating system; identifying whether the file is a template file or a non-template file; when the file is a template file, determining the minimum ESG rating indicator corresponding to the template file; when the file is a non-template file, determining whether the non-template file is a video file; if not, performing semantic analysis on the non-template file, and determining the minimum ESG rating indicator corresponding to the non-template file based on the result of the semantic analysis; if so, inputting the video file into an indicator recognition model, and determining the minimum ESG rating indicator corresponding to the non-template file through the indicator recognition model. In the present application, the minimum ESG rating indicator of the ESG rating corresponding to the file is automatically determined based on the file uploaded by the enterprise, which can provide assistance for the enterprise to generate an ESG report when the enterprise does not have ESG professionals.

Description

Technical Field

[0001] The present invention relates to the field of Internet technology, and in particular to an ESG rating indicator matching method, device, electronic device and storage medium. Background Art

[0002] ESG (Environment, Social Responsibility, Corporate Governance) refers to the three core factors for measuring the sustainability and ethical impact of a company or corporate investment. The ESG evaluation system currently used in the market relies on the experience of industry experts and analysts, which is not only inefficient, but also costly for small and micro enterprises to hire a professional ESG team. Summary of the invention

[0003] In response to the shortcomings of the prior art, the present application provides an ESG rating indicator matching method, device, electronic device and storage medium, which are conducive to automatically determining the minimum ESG rating indicator of the ESG rating corresponding to the file uploaded by the enterprise, and can assist the enterprise in generating ESG reports when the enterprise does not have ESG professionals.

[0004] To solve the above problems, the present invention provides the following technical solutions:

[0005] In a first aspect, an embodiment of the present application provides an indicator matching method for ESG rating, the method comprising:

[0006] Access files uploaded via the ESG rating system;

[0007] Identify whether the file is a template file or a non-template file;

[0008] When the file is a template file, determining the minimum ESG rating indicator corresponding to the template file;

[0009] When the file is a non-template file, determining whether the non-template file is a video file;

[0010] If the non-template file is not a video file, performing semantic analysis on the non-template file, and determining the minimum ESG rating index corresponding to the non-template file according to the result of the semantic analysis;

[0011] If the non-template file is a video file, the video file is input into the indicator recognition model, and the minimum ESG rating indicator corresponding to the non-template file is determined by the indicator recognition model.

[0012] In some implementations, performing semantic analysis on the non-template file and determining the minimum ESG rating indicator corresponding to the non-template file according to the result of the semantic analysis includes:

[0013] Performing semantic analysis on the non-template file, outputting the semantics of the non-template file, and calculating semantic similarity between the semantics and the corresponding semantics of each of the minimum ESG rating indicators; wherein the corresponding semantics of each of the minimum ESG rating indicators is the sum of the semantics of the superior indicator to which the minimum ESG rating indicator is subordinate and its own semantics;

[0014] The minimum ESG rating indicator with the largest semantic similarity is determined as the minimum ESG rating indicator corresponding to the non-template file.

[0015] In some embodiments, inputting the video file into an indicator recognition model, and determining the minimum ESG rating indicator corresponding to the non-template file through the indicator recognition model, includes:

[0016] Identify whether each video of the non-template file includes an employee by using the indicator recognition model;

[0017] If employees are included, identifying the behavior of the employees or the activities being conducted by all employees in the video, and determining the minimum ESG rating indicator corresponding to each of the non-template files based on the behavior of the employees or the activities being conducted by all employees;

[0018] If employees are not included, the information and scenes included in the video are identified, and the minimum ESG rating index corresponding to each of the non-template files is determined based on the information and scenes.

[0019] In some implementations, the identifying the behavior of the employee or the activities being performed by all employees in the video includes:

[0020] Identify the behavior of employees or all activities being performed by employees based on the image information and audio information of the video, as well as the purpose of the said behavior or activities;

[0021] Generating indicators to classify texts according to the stated purpose and the stated actions or activities;

[0022] Determining the minimum ESG rating indicator corresponding to each non-template file based on the employee's behavior or the ongoing activities of all employees includes:

[0023] The minimum ESG rating indicator corresponding to each of the non-template files is determined according to the indicator classification text.

[0024] In some implementations, determining the minimum ESG rating indicator corresponding to each of the non-template files according to the indicator classification text includes:

[0025] Using the index classification text as a domain;

[0026] The target N-th level ESG rating indicators are used as the comment set; where N is a positive integer and when N is equal to 1, all first-level ESG rating indicators are target first-level ESG rating indicators, and the first-level ESG rating indicators include social indicators, environmental indicators and corporate governance indicators;

[0027] Obtain the membership function corresponding to each target N-level ESG rating indicator;

[0028] Calculate the membership degree of the indicator classification text to each target N-level ESG rating indicator according to the membership function corresponding to the indicator classification text and each target N-level ESG rating indicator;

[0029] Determine the subordinate ESG rating indicator of the target Nth level ESG rating indicator with the largest degree of subordination as the target N+1th level ESG rating indicator;

[0030] Let N=N+1, and return to the execution step to use the target N-th level ESG rating indicator as the comment set.

[0031] In some embodiments, the calculating the membership of the indicator classification text to each target Nth level ESG rating indicator according to the membership function corresponding to the indicator classification text and each target Nth level ESG rating indicator includes:

[0032] Determining whether the video includes a tag;

[0033] When the video includes a tag, identifying an ESG rating indicator corresponding to the tag;

[0034] Calculate the membership degree of the indicator classification text to each target N-level ESG rating indicator according to the membership function corresponding to the indicator classification text and each target N-level ESG rating indicator;

[0035] If the ESG rating indicator corresponding to the label is one of the target N-th level ESG rating indicators, the membership degree of the N-th level ESG rating indicator corresponding to the label is modified to increase the membership degree corresponding to the ESG rating indicator corresponding to the label.

[0036] In some implementations, after modifying the membership of the Nth level ESG rating indicator corresponding to the label, the method further includes:

[0037] If the membership of the Nth level ESG rating indicator corresponding to the label is still not the maximum membership after correction, an error message is fed back to the account that uploaded the file through the ESG rating system; wherein the error message includes the Nth level ESG rating indicator corresponding to the label and the Nth level ESG rating indicator corresponding to the maximum membership;

[0038] Receive feedback from the account that uploaded the file;

[0039] According to the feedback information, select one of the subordinate ESG rating indicators of the Nth level ESG rating indicator corresponding to the label and the Nth level ESG rating indicator corresponding to the maximum degree of membership as the target N+1th level ESG rating indicator.

[0040] In a second aspect, an embodiment of the present application provides an indicator matching device for ESG rating, the device comprising:

[0041] An acquisition unit, used to acquire files uploaded through the ESG rating system;

[0042] An identification unit, used for identifying whether the file is a template file or a non-template file;

[0043] A determination unit, configured to determine, when the file is a template file, a minimum ESG rating indicator corresponding to the template file;

[0044] A judging unit, used for judging whether the non-template file is a video file when the file is a non-template file;

[0045] A semantic analysis unit, configured to perform semantic analysis on the non-template file if the non-template file is not a video file, and determine a minimum ESG rating index corresponding to the non-template file according to a result of the semantic analysis;

[0046] A video recognition unit is used to input the video file into an indicator recognition model if the non-template file is a video file, and determine the minimum ESG rating indicator corresponding to the non-template file through the indicator recognition model.

[0047] In a third aspect, an embodiment of the present application provides an electronic device, including:

[0048] at least one processor; and,

[0049] a memory communicatively connected to at least one processor; wherein,

[0050] The memory stores instructions that can be executed by at least one processor, and the instructions are executed by at least one processor to enable the at least one processor to perform the indicator matching method for ESG rating of the first aspect.

[0051] In a fourth aspect, an embodiment of the present application provides a computer-readable storage medium, which stores an executable program. The executable program is executed by a processor to implement the indicator matching method for ESG rating as in the first aspect.

[0052] The present application provides an indicator matching method, device, electronic device and storage medium for ESG rating, the method comprising: obtaining a file uploaded through an ESG rating system; identifying whether the file is a template file or a non-template file; when the file is a template file, determining the minimum ESG rating indicator corresponding to the template file; when the file is a non-template file, determining whether the non-template file is a video file; if the non-template file is not a video file, performing semantic analysis on the non-template file, and determining the minimum ESG rating indicator corresponding to the non-template file according to the result of the semantic analysis; if the non-template file is a video file, inputting the video file into an indicator recognition model, and determining the minimum ESG rating indicator corresponding to the non-template file through the indicator recognition model. In the present application, by automatically determining the minimum ESG rating indicator of the ESG rating corresponding to the file according to the file uploaded by the enterprise, it can provide assistance for the enterprise to generate an ESG report when the enterprise does not have ESG professionals. BRIEF DESCRIPTION OF THE DRAWINGS

[0053] Figure 1 It is a flow chart of an indicator matching method for ESG rating provided in an embodiment of the present application.

[0054] Figure 2 yes Figure 1 A detailed flowchart of step 150 in FIG.

[0055] Figure 3 yes Figure 1 A detailed flowchart of step 160 in FIG.

[0056] Figure 4 It is a structural schematic diagram of an electronic device provided in an embodiment of the present application.

[0057] Figure 5 It is a structural block diagram of a computer-readable storage medium provided in an embodiment of the present application. DETAILED DESCRIPTION

[0058] The following will be combined with the drawings in the embodiments of the present application to clearly and completely describe the technical solutions in the embodiments of the present application. Obviously, the described embodiments are only part of the embodiments of the present application, not all of the embodiments. Based on the embodiments in the present application, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of this application.

[0059] 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 the indicated technical features. Therefore, the features defined as "first" and "second" may explicitly or implicitly include one or more of the features. In the description of the present invention, the meaning of "plurality" is two or more, unless otherwise clearly and specifically defined.

[0060] See also Figure 1 , Figure 1 Schematic diagram of a flow chart of an ESG rating indicator matching method provided in an embodiment of the present application. Figure 1 As shown, the indicator matching method 100 for ESG rating includes: steps 110 to 160.

[0061] Step 110: Obtain the file uploaded through the ESG rating system.

[0062] In some embodiments, after the ESG rating system is opened to enterprises, the ESG rating system opens file upload permissions to each employee in the enterprise, allowing employees to upload files. For example, each employee has an employee account, and can upload files by logging into the employee account. The account needs to be authenticated with real name and associated with the employee's department, supervisor, rank, etc. After uploading the file, the file uploader can be traced based on the file.

[0063] In some implementations, the file may include video files in various formats, such as company team-building videos, energy-saving and emission-reduction videos of employees riding public transportation, and various recorded company certificates.

[0064] In some implementations, the file may include non-video files such as pictures, words, and execl tables, such as company financial reports, company certificates, company employee lists, and the like.

[0065] Step 120: Identify whether the file is a template file or a non-template file.

[0066] In some embodiments, some minimum ESG rating indicators in the ESG rating will provide a template file for the user to download, and the user can download the template file and fill in the information.

[0067] In some implementations, the template file contains a template identifier, such as a watermark, a specific character string, etc. After the template file is obtained, the template identifier in the template file is used to identify which minimum ESG rating indicator the template corresponds to.

[0068] Please refer to Table 1 below, which is a partial list of indicators included in the ESG rating table provided in the embodiment of the present application.

[0069]

[0070] For example, the unit product water consumption in the four-level indicators in Table 1 has a template that can be downloaded and filled in by users.

[0071] Exemplarily, the minimum ESG rating indicator in Table 1 is a level 4 indicator.

[0072] Step 130: When the file is a template file, determine the minimum ESG rating index corresponding to the template file.

[0073] In some embodiments, the minimum ESG rating indicator corresponding to each template file is recorded and stored by identifying the template identifier in the template file. When the method of the present application is applied to a server, a correspondence table between the template identifier and the minimum ESG rating indicator can be pre-set in the server to facilitate querying which minimum ESG rating indicator corresponds to the template file.

[0074] In some embodiments, each employee's account will be associated with a department. In this application, different minimum ESG rating indicator permissions will be opened to different departments in the enterprise. For example, the statistical department corresponding to the four-level indicator unit product water consumption in Table 1 is the Product Department (PD), the statistical department for the proportion of female employees in the four-level indicator is the Human Resources Department (HR), and the statistical department for the number of suppliers that have obtained ISO 9001 certification in the four-level indicator is the Project Management Center (PMC).

[0075] By opening up permissions for different minimum ESG rating indicators to corresponding departments and allowing each department to fill in the minimum ESG rating indicators corresponding to its job responsibilities, on the one hand, it can ensure data validity and prevent companies from pushing the task of collecting ESG report data to a single department, such as the HR department. It takes a long time to resolve collection difficulties, and because it is not the responsibility of the department itself, if other departments do not cooperate, the collected data will be inaccurate. On the other hand, it can also improve the efficiency of data collection, because many companies have not been exposed to ESG ratings before, and ESG ratings require filling in a lot of content. Many times, companies feel at a loss as to where to start. The above method can split the content and let each department be responsible for a small part, completing the collection of department data first, and facilitating targeted collection of missing data later.

[0076] In addition, this application can collect files uploaded by company employees this year. Companies can improve data to update ESG reports at any time, so companies can also supplement files later. For example, if companies in the four-level indicators formulate a continuous electricity conservation system, they can re-rate themselves after improving the continuous electricity conservation system, which will facilitate corporate progress and allow companies to make targeted improvements.

[0077] Step 140: When the file is a non-template file, determine whether the non-template file is a video file.

[0078] In some implementations, whether it is a video file may be determined by the file format, for example, a file with the suffix MP3 is a video file, and a file with the suffix doc is a non-video file.

[0079] Step 150: If the non-template file is not a video file, perform semantic analysis on the non-template file, and determine the minimum ESG rating index corresponding to the non-template file according to the result of the semantic analysis.

[0080] In some embodiments, please refer to Figure 2 , Figure 2 yes Figure 1 Detailed flow chart of step 150 in FIG. Figure 2 As shown, step 150 includes the following steps.

[0081] Step 151: Perform semantic analysis on the non-template file, output the semantics of the non-template file, and calculate semantic similarity between the semantics and the corresponding semantics of each of the minimum ESG rating indicators; wherein the corresponding semantics of each of the minimum ESG rating indicators is the sum of the semantics of the superior indicator to which the minimum ESG rating indicator is subordinate and its own semantics.

[0082] In some implementations, for non-template files such as pictures, it is necessary to first extract text from the pictures.

[0083] In some embodiments, because there may be minimum ESG rating indicators with the same or similar semantics, for example, the three-level indicator in Table 1 includes the fourth-level indicator to realize information management, and the three-level indicator also includes the fourth-level indicator to realize information management. If only the semantics of the minimum ESG rating indicator are used for semantic similarity calculation, it will be impossible to determine the minimum ESG rating indicator that actually needs to be matched; therefore, the corresponding semantics of each minimum ESG rating indicator in the embodiment of the present application is the sum of the semantics of the superior indicator to which the minimum ESG rating indicator belongs and its own semantics.

[0084] Exemplarily, the superior indicators of water consumption per unit product include water use, resource use and environment.

[0085] In some embodiments, semantic similarity calculation can be performed by a semantic similarity calculation model, such as a recurrent neural network model (RNN) such as a long short-term memory network (LSTM) or a gated recurrent unit (GRU); or a transformer model such as bidirectional encoding based on transformer or bidirectional encoder representations (BERF).

[0086] Step 152: Determine the minimum ESG rating indicator with the largest semantic similarity as the minimum ESG rating indicator corresponding to the non-template file.

[0087] Through the above method, the minimum ESG rating index corresponding to the non-video file can be determined. After determining the minimum ESG rating index corresponding to the non-video file, it can provide assistance for enterprises to generate ESG reports. For example, the enterprise can calculate its own ESG rating based on the obtained ESG rating calculation table and the file corresponding to the ESG rating index, thereby assisting the enterprise in conducting ESG assessment.

[0088] Step 160: If the non-template file is a video file, the video file is input into the indicator recognition model, and the minimum ESG rating indicator corresponding to the non-template file is determined by the indicator recognition model.

[0089] In some embodiments, please refer to Figure 3 , Figure 3 yes Figure 1 Detailed flow chart of step 160 in FIG. Step 160 includes steps 161 to 163.

[0090] Step 161: Identify whether each video of the non-template file includes an employee through the indicator recognition model.

[0091] In some embodiments, step 161 includes the following steps.

[0092] (1) Video preprocessing: Perform preprocessing on the uploaded video, including format conversion, resolution adjustment, and frame rate unification.

[0093] (2) Feature extraction: Use convolutional neural networks (CNN) and object detection models (YOLO, Mask R-CNN) to extract key image features in the video and identify environmental elements (such as pollution, vegetation cover), social activities (such as employee benefits, community participation), governance structures (such as meetings, safety signs), scene elements (people, trees, cars), etc.

[0094] (3) Scene understanding: Natural language processing (NLP) and sequence models (such as LSTM) are used to analyze the voice dialogue in the video, and the overall context of the video is understood in combination with the image content.

[0095] Step 162: If employees are included, identify the behavior of the employees in the video or the activities being performed by all employees, and determine the minimum ESG rating indicator corresponding to each of the non-template files based on the behavior of the employees or the activities being performed by all employees.

[0096] For example, if employees are having a dinner together, it is determined as a team-building video, which corresponds to the minimum ESG rating indicator of employee care.

[0097] Step 163: If employees are not included, identify the information and scenes included in the video, and determine the minimum ESG rating index corresponding to each of the non-template files based on the information and scenes.

[0098] In some embodiments, the information may include certificates, company regulatory documents, various statistical tables, etc.

[0099] In some implementations, the scene includes a conference room, a presentation site, etc.

[0100] For example, if the information in the video shows the company's ISO45001 certificate, it corresponds to the minimum ESG rating indicator occupational health and safety management system certification.

[0101] In some embodiments, step 162 includes the following steps.

[0102] Step 1621: Identify the behavior of employees or the activities being performed by all employees, and the purpose of the behavior or activity based on the image information and audio information of the video.

[0103] In some embodiments, only identifying employee behavior or all employees' ongoing activities cannot determine the true content of the video. For example, the activity currently being performed by all employees is a meeting. Depending on the content of the meeting, the minimum ESG rating indicator corresponding to the video may be the company's formulation of a continuous water conservation system, the company's formulation of a continuous electricity conservation system, the construction of a safety culture, etc. In this case, it is necessary to identify the purpose of the behavior or activity based on the audio information.

[0104] Step 1622: Generate indicator classification text based on the purpose and the behavior or activity.

[0105] For example, the activity is a meeting, the purpose of the activity is to publicize the rules and regulations related to water conservation, and the indicator classification text can be the rules and regulations related to water conservation promulgated at the meeting.

[0106] In some implementations, in order to make the indicator classification text more accurate, an activity description may be added to the indicator classification text.

[0107] For example, the indicator classification text after adding the activity description may be that rules and regulations on water conservation were promulgated at a meeting with 300 participants, and a goal of saving 30% of water this year was set.

[0108] Based on step 1621 to step 1622, step 163 includes the following steps.

[0109] Step 1631: Determine the minimum ESG rating indicator corresponding to each of the non-template files based on the indicator classification text.

[0110] Through the above method, the video files containing complex information are simplified into indicator classification texts, and the minimum ESG rating indicator corresponding to each non-template file is determined by the indicator classification text. Compared with directly determining the minimum ESG rating indicator based on the video, this can reduce the difficulty of model construction and reduce the computing power required by the model.

[0111] In some embodiments, step 1631 includes the following steps.

[0112] (1) The index classification text is used as the domain.

[0113] (2) The target N-th level ESG rating indicators are used as the comment set; where N is a positive integer and when N is equal to 1, all first-level ESG rating indicators are target first-level ESG rating indicators. The first-level ESG rating indicators include social indicators, environmental indicators and corporate governance indicators.

[0114] (3) Obtain the membership function corresponding to each target N-level ESG rating indicator.

[0115] (4) Calculating the degree of membership of the indicator classification text to each target Nth level ESG rating indicator based on the indicator classification text and the membership function corresponding to each target Nth level ESG rating indicator.

[0116] (5) The subordinate ESG rating indicator of the target N-level ESG rating indicator with the maximum degree of membership is determined as the target N+1-level ESG rating indicator.

[0117] (6) Set N = N + 1 and return to the execution step to use the target N-th level ESG rating indicator as the comment set.

[0118] Specifically, the domain, comment set and membership function together constitute the basis of fuzzy recognition; among them:

[0119] The domain reflects the set of all possible input values.

[0120] The linguistic variable set reflects the words or labels that describe or evaluate the objects or states in the domain.

[0121] The membership function defines the degree of membership of each element in the domain to a fuzzy set, that is, how much each element belongs to the fuzzy set. The value range of the membership function is usually between 0 and 1, where 0 means it does not belong at all and 1 means it belongs completely. Values ​​in between indicate a certain degree of membership. For example, in the temperature control scenario, the membership function of the comment "warmth" may have a higher value within a certain temperature range, indicating that the temperature is closer to the concept of "warmth".

[0122] In some embodiments, each ESG rating indicator corresponds to a membership function.

[0123] In some implementations, an initial domain is defined in advance, for example, the initial domain includes multiple elements, such as energy conservation and emission reduction, effective water conservation, weekends off, birthday gifts, etc. Step (1) using the indicator classification text as the domain includes: identifying elements in the initial domain that are included in or semantically close to the indicator classification text, and using the elements in the initial domain that are included in or semantically close to the indicator classification text as the domain.

[0124] That is to say, the domain in this application belongs to a subset of the initial domain.

[0125] In some embodiments, each ESG rating indicator has a corresponding membership function for each element in the initial domain.

[0126] For example, suppose that the elements in the domain include "energy conservation and emission reduction". For "environment", since "energy conservation and emission reduction" is directly aimed at environmental protection, we can assume that its membership to "environment" is very high. Assuming that the membership ranges from 0 (not belonging at all) to 1 (completely belonging), we can simply define a high membership value, such as 0.9 or 1; for example, the membership function of "energy conservation and emission reduction" to "environment" is f(environment: energy conservation and emission reduction) = 1 or 0.9.

[0127] For example, for "society", although "energy conservation and emission reduction" is part of social sustainable development, its direct relevance is not as strong as that of "environment". We can think that its membership to "society" is medium to high, because society includes many aspects, and "energy conservation and emission reduction" is only a part of it; for example, the membership function of "energy conservation and emission reduction" to "society" is f (society: energy conservation and emission reduction) = 0.7, and the specific value can be adjusted according to the actual situation, but it should be lower than the membership of "energy conservation and emission reduction" to "environment".

[0128] For example, for a "company", "energy conservation and emission reduction" may appear as a strategy or goal in corporate governance, but it is not the core content of corporate governance. Therefore, its degree of affiliation to "corporate governance" should be relatively low; for example, f(corporate governance: energy conservation and emission reduction) = 0.3. The specific value can be adjusted according to the specific situation, but it should be significantly lower than the degree of affiliation of "energy conservation and emission reduction" to "environment" and "society".

[0129] Through the above method, the sum of the membership of all elements in the domain to the ESG rating indicator of the same level can be calculated, and the ESG rating indicator corresponding to the maximum membership can be determined as the target N-th level ESG rating indicator.

[0130] In some embodiments, when the video includes multiple purposes, the domain may include multiple indicator classification texts, for example, indicator classification text 1 is the rules and regulations on water conservation issued at the meeting, and indicator classification text 2 is the rules and regulations on electricity conservation issued at the meeting.

[0131] Exemplarily, the domain includes indicator classification text 1 and indicator classification text 2, and the comment set includes environment, society, and corporate governance. Then indicator classification text 1 and indicator classification text 2 respectively calculate the membership degree belonging to the environment through the membership function corresponding to the environment, indicator classification text 1 and indicator classification text 2 respectively calculate the membership degree belonging to the society through the membership function corresponding to the society, and indicator classification text 1 and indicator classification text 2 respectively calculate the membership degree belonging to corporate governance through the membership function corresponding to corporate governance.

[0132] In some embodiments, because the number of minimum ESG rating indicators is very large, which can be as many as hundreds, if all minimum ESG rating indicators are used as a comment set, the amount of calculation will be very large, and the above content of the specification also mentioned that some minimum ESGs have similar or even identical semantics. For example, the fourth-level indicators under the third-level indicator exhaust gas include the realization of information management, and the fourth-level indicators under the third-level indicator electricity consumption also include the realization of information management. Therefore, if all minimum ESG rating indicators are used as a comment set, the membership functions of some minimum ESG rating indicators must be specially processed to distinguish between similar or even identical minimum ESG rating indicators.

[0133] Therefore, in the above method, calculating the membership degree step by step starting from the upper-level indicators can greatly reduce the amount of calculation and improve the matching accuracy of the minimum ESG rating indicators.

[0134] In some embodiments, step (4) calculates the membership of the indicator classification text to each target Nth level ESG rating indicator based on the indicator classification text and the membership function corresponding to each target Nth level ESG rating indicator, including the following steps.

[0135] (4.1) Determine whether the video includes a tag.

[0136] In some implementations, when an employee uploads a file, a label filling box will pop up, allowing the employee to fill in a label in the label filling box that is easy for the indicator identification model to identify, such as energy conservation and emission reduction, team building, etc.

[0137] (4.2) When the video includes a tag, identifying an ESG rating indicator corresponding to the tag.

[0138] In some embodiments, the ESG rating indicator directly referred to by the label and all its parent indicators are the ESG rating indicators corresponding to the label.

[0139] For example, if the label of the video is physical examination, then the corresponding fourth-level indicator is physical examination for all employees. The third-level indicator occupational health, the second-level indicator health and safety, and the first-level indicator society of the four-level indicator physical examination for all employees are all ESG rating indicators corresponding to the label physical examination.

[0140] In some embodiments, when calculating the ESG rating indicator directly referred to by the label, if through semantic recognition, it is necessary to calculate whether there are ESG rating indicators whose semantic similarity is greater than the similarity threshold, and determine the ESG rating indicator with the greatest semantic similarity among the ESG rating indicators whose semantic similarity is greater than the similarity threshold as the ESG rating indicator directly referred to by the label.

[0141] Specifically, the labels given by users are sometimes not accurate enough or even irrelevant to ESG indicators, which is equivalent to noise for the indicator identification model. At this time, if no similarity threshold judgment is performed, directly determining the ESG rating indicator with the greatest semantic similarity as the ESG rating indicator directly referred to by the label is likely to be wrong. The above threshold judgment method can reduce the possibility of error.

[0142] (5.1) Calculating the degree of membership of the indicator classification text to each target Nth level ESG rating indicator based on the indicator classification text and the membership function corresponding to each target Nth level ESG rating indicator.

[0143] (5.2) If the ESG rating indicator corresponding to the label is one of the target N-level ESG rating indicators, the membership degree of the N-level ESG rating indicator corresponding to the label is modified to increase the membership degree corresponding to the ESG rating indicator corresponding to the label.

[0144] Exemplarily, the target Level 1 ESG rating indicators are society, environment, and governance. If the ESG rating indicator corresponding to the label is society, then after calculating the social affiliation corresponding to the indicator identification text, the social affiliation is corrected to increase the social affiliation. Through user sample annotation, the AI ​​model recognition can be more accurate.

[0145] Optionally, the correction may be to change the membership degree to 1.3 times, but the membership degree after correction is 1 at most.

[0146] In some embodiments, after the membership of the Nth level ESG rating indicator corresponding to the label is corrected in step (5.2), the method further includes the following steps.

[0147] (5.3) If the membership of the Nth level ESG rating indicator corresponding to the label is still not the maximum membership after correction, an error message is fed back to the account that uploaded the file through the ESG rating system; wherein the error message includes the Nth level ESG rating indicator corresponding to the label and the Nth level ESG rating indicator corresponding to the maximum membership;

[0148] (5.4) Receive feedback from the account that uploaded the file;

[0149] (5.5) Selecting, based on the feedback information, a subordinate ESG rating indicator of one of the Nth-level ESG rating indicators corresponding to the label and the Nth-level ESG rating indicator corresponding to the maximum membership as the target N+1th-level ESG rating indicator.

[0150] Exemplarily, if the user selects environmental indicators from the environmental indicators and social indicators of the first-level ESG rating indicators, the subordinate ESG rating indicators of the environmental indicators will be used as the target second-level ESG rating indicators.

[0151] Specifically, if the model recognition result is inconsistent with the result of the label given by the user, additional confirmation will be required from the user to give the user room for remedy. For example, when the model's recognition effect on a certain file is poor and it always recognizes it incorrectly, the above method allows the user to manually correct the result. In addition, the above method can also make the model recognition results more accurate and prevent the model from identifying obviously incorrect ESG ratings.

[0152] The embodiment of the present application also provides an ESG rating indicator matching device, the device comprising:

[0153] An acquisition unit, used to acquire files uploaded through the ESG rating system;

[0154] An identification unit, used for identifying whether the file is a template file or a non-template file;

[0155] A determination unit, configured to determine, when the file is a template file, a minimum ESG rating indicator corresponding to the template file;

[0156] A judging unit, used for judging whether the non-template file is a video file when the file is a non-template file;

[0157] A semantic analysis unit, configured to perform semantic analysis on the non-template file if the non-template file is not a video file, and determine a minimum ESG rating index corresponding to the non-template file according to a result of the semantic analysis;

[0158] A video recognition unit is used to input the video file into an indicator recognition model if the non-template file is a video file, and determine the minimum ESG rating indicator corresponding to the non-template file through the indicator recognition model.

[0159] In some implementations, the model further includes an execution unit, which can execute some or all of the steps in the above method embodiments.

[0160] See also Figure 4 , Figure 4 Schematic diagram of the structure of an electronic device provided in an embodiment of the present application. Figure 4 As shown, the electronic device 200 includes: one or more processors 210 and a memory 220, Figure 4 A processor 210 is taken as an example.

[0161] In some implementations, the processor 210 and the memory 220 may be connected via a bus or other means. Figure 4 The example of connecting through bus is taken in the following.

[0162] In some embodiments, the processor 210 is used to obtain a file uploaded through the ESG rating system; identify whether the file is a template file or a non-template file; when the file is a template file, determine the minimum ESG rating indicator corresponding to the template file; when the file is a non-template file, determine whether the non-template file is a video file; if the non-template file is not a video file, perform semantic analysis on the non-template file, and determine the minimum ESG rating indicator corresponding to the non-template file based on the result of the semantic analysis; if the non-template file is a video file, input the video file into the indicator recognition model, and determine the minimum ESG rating indicator corresponding to the non-template file through the indicator recognition model.

[0163] In some embodiments, the memory 220, as a non-volatile computer-readable storage medium, can be used to store non-volatile software programs, non-volatile computer executable programs and modules, such as program instructions / modules of the indicator matching method for ESG rating in the embodiment of the present application. The processor 210 executes various functional applications and data processing of the electronic device by running the non-volatile software programs, instructions and modules stored in the memory 220, that is, the indicator matching method for ESG rating in the above method embodiment is implemented.

[0164] In some embodiments, the memory 220 may include a program storage area and a data storage area, wherein the program storage area may store an operating system, an application required for at least one function; the data storage area may store data created according to the use of the electronic device, etc. In addition, the memory 220 may include a high-speed random access memory, and may also include a non-volatile memory, such as at least one disk storage device, a flash memory device, or other non-volatile solid-state storage device. In some embodiments, the memory 220 may optionally include a memory remotely arranged relative to the processor 210, and these remote memories may be connected to the controller via a network. Examples of the above-mentioned network include, but are not limited to, the Internet, an intranet, a local area network, a mobile communication network, and combinations thereof.

[0165] In some embodiments, one or more modules are stored in the memory 220, and when executed by one or more processors 210, the ESG rating indicator matching method in any of the above method embodiments is executed, for example, the above described Figure 1 Method steps 110 to 160 in.

[0166] Please refer to Figure 5 , Figure 5 The computer-readable storage medium 300 stores a program code 310, which can be called by a processor to execute the index matching method for ESG rating described in the above method embodiment.

[0167] The computer-readable storage medium 300 may be an electronic memory such as a flash memory, an EEPROM (electrically erasable programmable read-only memory), an EPROM, a hard disk or a ROM. Optionally, the computer-readable storage medium includes a non-transitory computer-readable storage medium. The computer-readable storage medium 300 has a storage space for program codes for executing any method steps in the above control method. These program codes may be read from or written to one or more computer program products. The program code may be compressed, for example, in an appropriate form.

[0168] In summary, the present application provides an indicator matching method, device, electronic device and storage medium for ESG rating, the method comprising: obtaining a file uploaded through an ESG rating system; identifying whether the file is a template file or a non-template file; when the file is a template file, determining the minimum ESG rating indicator corresponding to the template file; when the file is a non-template file, determining whether the non-template file is a video file; if the non-template file is not a video file, performing semantic analysis on the non-template file, and determining the minimum ESG rating indicator corresponding to the non-template file according to the result of the semantic analysis; if the non-template file is a video file, inputting the video file into an indicator recognition model, and determining the minimum ESG rating indicator corresponding to the non-template file through the indicator recognition model. In the present application, by automatically determining the minimum ESG rating indicator of the ESG rating corresponding to the file according to the file uploaded by the enterprise, it can provide assistance for the enterprise to generate an ESG report when the enterprise does not have ESG professionals.

[0169] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present application, rather than to limit it. Although the present application has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that they can still modify the technical solutions described in the aforementioned embodiments, or make equivalent replacements for some of the technical features therein. However, these modifications or replacements 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 application.

Claims

1. An indicator matching method for ESG rating, characterized in that: The method comprises: Access files uploaded via the ESG rating system; Identify whether the file is a template file or a non-template file; When the file is a template file, determining the minimum ESG rating indicator corresponding to the template file; the minimum ESG rating indicator is the minimum level indicator in the ESG rating; When the file is a non-template file, determining whether the non-template file is a video file; If the non-template file is not a video file, performing semantic analysis on the non-template file, and determining a minimum ESG rating index corresponding to the non-template file according to a result of the semantic analysis; If the non-template file is a video file, the video file is input into the indicator recognition model, and the minimum ESG rating indicator corresponding to the non-template file is determined by the indicator recognition model.

2. The method according to claim 1, characterized in that The performing semantic analysis on the non-template file and determining the minimum ESG rating index corresponding to the non-template file according to the result of the semantic analysis includes: Performing semantic analysis on the non-template file, outputting the semantics of the non-template file, and calculating semantic similarity between the semantics and the corresponding semantics of each of the minimum ESG rating indicators; wherein the corresponding semantics of each of the minimum ESG rating indicators is the sum of the semantics of the superior indicator to which the minimum ESG rating indicator is subordinate and its own semantics; The minimum ESG rating indicator with the largest semantic similarity is determined as the minimum ESG rating indicator corresponding to the non-template file.

3. The method according to claim 1, characterized in that The step of inputting the video file into an indicator recognition model and determining the minimum ESG rating indicator corresponding to the non-template file through the indicator recognition model includes: Identify whether each video of the non-template file includes an employee by using the indicator recognition model; If employees are included, identifying the behavior of the employees or the activities being conducted by all employees in the video, and determining the minimum ESG rating indicator corresponding to each of the non-template files based on the behavior of the employees or the activities being conducted by all employees; If employees are not included, the information and scenes included in the video are identified, and the minimum ESG rating index corresponding to each of the non-template files is determined based on the information and scenes.

4. The method according to claim 3, characterized in that The actions of the employee or all activities being performed by the employee in the identified video, including: Identify the behavior of employees or all activities being performed by employees based on the image information and audio information of the video, as well as the purpose of the said behavior or activities; Generating indicators to classify texts according to the stated purpose and the stated actions or activities; Determining the minimum ESG rating indicator corresponding to each non-template file based on the employee's behavior or the ongoing activities of all employees includes: The minimum ESG rating indicator corresponding to each of the non-template files is determined according to the indicator classification text.

5. The method according to claim 4, characterized in that Determining the minimum ESG rating indicator corresponding to each of the non-template files according to the indicator classification text includes: Using the index classification text as a domain; The target N-th level ESG rating indicators are used as the comment set; where N is a positive integer and when N is equal to 1, all first-level ESG rating indicators are target first-level ESG rating indicators, and the first-level ESG rating indicators include social indicators, environmental indicators and corporate governance indicators; Obtain the membership function corresponding to each target N-level ESG rating indicator; Calculate the membership degree of the indicator classification text to each target N-level ESG rating indicator according to the membership function corresponding to the indicator classification text and each target N-level ESG rating indicator; Determine the subordinate ESG rating indicator of the target Nth level ESG rating indicator with the largest degree of subordination as the target N+1th level ESG rating indicator; Let N=N+1, and return to the execution step to use the target N-th level ESG rating indicator as the comment set.

6. The method according to claim 5, characterized in that The calculating the membership degree of the indicator classification text to each target Nth level ESG rating indicator according to the membership function corresponding to the indicator classification text and each target Nth level ESG rating indicator includes: Determining whether the video includes a tag; When the video includes a tag, identifying an ESG rating indicator corresponding to the tag; Calculate the membership degree of the indicator classification text to each target N-level ESG rating indicator according to the membership function corresponding to the indicator classification text and each target N-level ESG rating indicator; If the ESG rating indicator corresponding to the label is one of the target N-th level ESG rating indicators, the membership degree of the N-th level ESG rating indicator corresponding to the label is modified to increase the membership degree corresponding to the ESG rating indicator corresponding to the label.

7. The method according to claim 6, characterized in that After correcting the degree of membership of the Nth level ESG rating indicator corresponding to the label, the method further includes: If the membership of the Nth level ESG rating indicator corresponding to the label is still not the maximum membership after correction, an error message is fed back to the account that uploaded the file through the ESG rating system; wherein the error message includes the Nth level ESG rating indicator corresponding to the label and the Nth level ESG rating indicator corresponding to the maximum membership; Receive feedback from the account that uploaded the file; According to the feedback information, select one of the subordinate ESG rating indicators of the Nth level ESG rating indicator corresponding to the label and the Nth level ESG rating indicator corresponding to the maximum degree of membership as the target N+1th level ESG rating indicator.

8. An indicator matching device for ESG rating, characterized in that: The device comprises: An acquisition unit, used to acquire files uploaded through the ESG rating system; An identification unit, used for identifying whether the file is a template file or a non-template file; A determination unit, configured to determine, when the file is a template file, a minimum ESG rating indicator corresponding to the template file; the minimum ESG rating indicator is a minimum level indicator in the ESG rating; A judging unit, used for judging whether the non-template file is a video file when the file is a non-template file; A semantic analysis unit, configured to perform semantic analysis on the non-template file if the non-template file is not a video file, and determine a minimum ESG rating index corresponding to the non-template file according to a result of the semantic analysis; A video recognition unit is used to input the video file into an indicator recognition model if the non-template file is a video file, and determine the minimum ESG rating indicator corresponding to the non-template file through the indicator recognition model.

9. An electronic device, characterized in that: The electronic device comprises: at least one processor; and, a memory communicatively connected to the at least one processor; wherein, The memory stores instructions that can be executed by the at least one processor, and the instructions are executed by the at least one processor so that the at least one processor can execute the indicator matching method for ESG rating as described in any one of claims 1-7.

10. A computer-readable storage medium, characterized in that: The computer-readable storage medium stores an executable program, and the executable program is executed by a processor to implement the indicator matching method for ESG rating as described in any one of claims 1-7.

Images