Document-based business processing method and device, equipment, storage medium and program product

Abstract

The invention discloses a document-based business processing method and device, equipment, a storage medium and a program product, which can be applied to the field of financial science and technology. The method comprises the steps of obtaining a to-be-analyzed ESG document of a to-be-evaluated enterprise, and performing text preprocessing on the to-be-analyzed ESG document to obtain target ESG text data; inputting the target ESG text data into a pre-trained named entity recognition model to obtain a vocabulary sequence set output by the model; performing vocabulary analysis on each entity vocabulary in the vocabulary sequence set, and determining an ESG association index in each entity vocabulary; performing index feature analysis on the ESG association index, and determining an ESG evaluation result of the target enterprise; and performing business processing on the target enterprise according to the ESG evaluation result of the target enterprise. According to the technical scheme, the accuracy and comprehensiveness of identifying and evaluating the ESG document are improved, and the accurate processing of related business of an enterprise is realized.

Description

Technical Field

[0001] The present invention relates to the field of financial technology, and in particular to a document-based business processing method, device, equipment, storage medium and program product. Background Art

[0002] With the increasing attention paid to sustainable development and corporate social responsibility around the world, ESG (Environmental, Social, and Governance) information disclosure has become an important dimension for evaluating a company's non-financial performance. However, most of the existing ESG information exists in the form of unstructured text, making it difficult to use the data efficiently.

[0003] In the existing technology, natural language processing methods are usually used to analyze ESG reports, but the existing technology is often limited to simple word segmentation and topic model analysis, that is, the report text needs to be segmented first, and then put into the LDA (Latent Dirichlet Allocation) topic model to output information such as topic sets, coherence scores, and topic word frequencies, and assign weights for rating. This solution relies on word segmentation, which may cause information loss or incorrect processing; at the same time, the LDA topic model has great limitations and cannot accurately identify and extract the core content in the report text, making it difficult to accurately evaluate the report file, which can easily cause a greater negative impact on the evaluation of the company's business and business decisions.

[0004] Therefore, how to conduct a comprehensive and accurate evaluation of a company based on its ESG data, thereby achieving precise processing of the company's business, has become a technical problem that technical personnel urgently need to solve. Summary of the invention

[0005] The present invention provides a document-based business processing method, apparatus, equipment, storage medium and program product, which can improve the accuracy and comprehensiveness of ESG document recognition and evaluation by extracting entity vocabulary from ESG text and obtaining evaluation results corresponding to feature indicators, thereby achieving precise processing of enterprise-related business.

[0006] According to one aspect of the present invention, a document-based business processing method is provided, comprising:

[0007] Obtaining the ESG document to be analyzed of the enterprise to be evaluated, and performing text preprocessing on the ESG document to be analyzed to obtain target ESG text data;

[0008] Input the target ESG text data into a pre-trained named entity recognition model to obtain a vocabulary sequence set output by the model; the vocabulary sequence set includes at least one entity vocabulary;

[0009] Performing lexical analysis on each of the entity words in the lexical sequence set to determine an ESG relevance index in each of the entity words;

[0010] Conducting an indicator characteristic analysis on the ESG-related indicators to determine the ESG evaluation results of the target enterprise;

[0011] Perform business processing on the target enterprise based on the ESG evaluation results of the target enterprise.

[0012] According to another aspect of the present invention, there is provided a document-based service processing device, comprising:

[0013] A text data acquisition module is used to acquire the ESG document to be analyzed of the enterprise to be evaluated, and perform text preprocessing on the ESG document to be analyzed to obtain target ESG text data;

[0014] A vocabulary sequence set generation module, used for inputting the target ESG text data into a pre-trained named entity recognition model to obtain a vocabulary sequence set output by the model; the vocabulary sequence set includes at least one entity vocabulary;

[0015] A vocabulary analysis module, used to perform vocabulary analysis on each of the entity words in the vocabulary sequence set to determine the ESG association index in each of the entity words;

[0016] An indicator characteristic analysis module, used to perform indicator characteristic analysis on the ESG-related indicators to determine the ESG evaluation results of the target enterprise;

[0017] The business processing module is used to perform business processing on the target enterprise according to the ESG evaluation result of the target enterprise.

[0018] According to another aspect of the present invention, there is provided an electronic device, the electronic device comprising:

[0019] at least one processor; and

[0020] a memory communicatively connected to the at least one processor; wherein,

[0021] The memory stores a computer program that can be executed by the at least one processor, and the computer program is executed by the at least one processor so that the at least one processor can execute the document-based business processing method described in any embodiment of the present invention.

[0022] According to another aspect of the present invention, a computer-readable storage medium is provided, wherein the computer-readable storage medium stores computer instructions, and the computer instructions are used to enable a processor to implement the document-based business processing method described in any embodiment of the present invention when executed.

[0023] According to another aspect of the present invention, a computer program product is provided. The computer program product includes a computer program. When the computer program is executed by a processor, the document-based business processing method described in any embodiment of the present invention is implemented.

[0024] The technical solution of the embodiment of the present invention obtains the ESG document to be analyzed of the enterprise to be evaluated, and performs text preprocessing on the ESG document to be analyzed to obtain the target ESG text data; inputs the target ESG text data into the pre-trained named entity recognition model to obtain a vocabulary sequence set output by the model; the vocabulary sequence set includes at least one entity vocabulary; performs vocabulary analysis on each entity vocabulary in the vocabulary sequence set to determine the ESG related indicators in each entity vocabulary; performs indicator feature analysis on the ESG related indicators to determine the ESG evaluation results of the target enterprise; and performs business processing on the target enterprise according to the ESG evaluation results of the target enterprise. This technical solution can use the named entity recognition model to extract entity vocabulary in the ESG document and determine the ESG related indicators in the entity vocabulary without manual word segmentation, thereby improving the comprehensiveness and efficiency of the identification and analysis of the ESG document; it can also process the business according to the ESG evaluation results of the target enterprise, thereby improving the accuracy and comprehensiveness of the identification and evaluation of the ESG document, and realizing the precise processing of the enterprise-related business.

[0025] It should be understood that the contents described in this section are not intended to identify the key or important features of the embodiments of the present invention, nor are they intended to limit the scope of the present invention. Other features of the present invention will become easily understood through the following description. BRIEF DESCRIPTION OF THE DRAWINGS

[0026] In order to more clearly illustrate the technical solutions in the embodiments of the present invention, the following briefly introduces the drawings required for use in the description of 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 work.

[0027] Figure 1 is a flowchart of a document-based business processing method provided according to Embodiment 1 of the present invention;

[0028] Figure 2 is a flowchart of a document-based business processing method provided according to Embodiment 2 of the present invention;

[0029] Figure 3 is a schematic diagram of the structure of a document-based business processing device provided according to Embodiment 3 of the present invention;

[0030] Figure 4 It is a structural schematic diagram of an electronic device for implementing the document-based business processing method according to an embodiment of the present invention. DETAILED DESCRIPTION

[0031] In order to enable those skilled in the art to better understand the scheme of the present invention, the technical scheme in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work should fall within the scope of protection of the present invention.

[0032] It should be noted that the terms "first", "second", etc. in the specification and claims of the present invention and the above-mentioned drawings are used to distinguish similar objects, and are not necessarily used to describe a specific order or sequence. It should be understood that the data used in this way can be interchanged where appropriate, so that the embodiments of the present invention described herein can be implemented in an order other than those illustrated or described herein. In addition, the terms "including" and "having" and any variations thereof are intended to cover non-exclusive inclusions, for example, a process, method, system, product or device that includes a series of steps or units is not necessarily limited to those steps or units that are clearly listed, but may include other steps or units that are not clearly listed or inherent to these processes, methods, products or devices.

[0033] Embodiment 1

[0034] Figure 1 A flowchart of a document-based business processing method is provided for the first embodiment of the present invention. This embodiment can be applied to the situation where a comprehensive and accurate evaluation of an enterprise is conducted and a decision is made based on the evaluation result whether to conduct business processing on the enterprise. The method can be executed by a document-based business processing device, which can be implemented in the form of hardware and / or software, and can be configured in an electronic device. Figure 1 As shown, the method includes:

[0035] S110 , obtaining the ESG document to be analyzed of the enterprise to be evaluated, and performing text preprocessing on the ESG document to be analyzed to obtain target ESG text data.

[0036] The enterprise to be evaluated may be a target enterprise that needs to undergo ESG analysis. The ESG documents to be analyzed may be original report files collected in advance to record the strategies and actions of the enterprise to be evaluated in terms of environment, society and governance. Specifically, they may include the enterprise's strategies and actions in terms of reducing carbon emissions, resource use and sustainability, and may also include report contents such as how the enterprise handles its relationships with employees, supply chains, customers and society. The ESG documents to be analyzed are usually recorded in PDF (Portable Document Format) or Word (Word document) and other unstructured text information that cannot be analyzed for content recognition.

[0037] Among them, the text preprocessing of the ESG document to be analyzed can specifically be to convert the original report file of the ESG document to be analyzed into text data that can be analyzed and processed. The preprocessing process may include noise removal, word segmentation, stop word removal, stem extraction and word form restoration, and non-alphabetic character processing methods.

[0038] Among them, the target ESG text data can be formatted text data that can be used for content recognition analysis after preprocessing the original report file. Specifically, it can be TXT (Text) format, or XML (eXtensible Markup Language) format and other data formats. The specific target ESG text data can be set and generated by technical personnel according to actual needs.

[0039] Specifically, when obtaining the ESG document of the target enterprise, the unstructured document format that cannot be subjected to content recognition analysis can be preprocessed to obtain text data content that can be subjected to content recognition analysis. For example, if the collected ESG document to be analyzed is a PDF file, the PDF file can be converted into an XML format file through preprocessing.

[0040] It should be noted that the relevant information (including but not limited to user device information, user personal information, etc.) and data (including but not limited to data for display, data for analysis, etc.) involved in this disclosure are all information and data authorized by the user or fully authorized by all parties, and the collection, use and processing of relevant data comply with relevant laws, regulations and standards in relevant regions.

[0041] Optionally, the ESG document to be analyzed is subjected to text preprocessing to obtain target ESG text data, including: formatting the ESG document to be analyzed to obtain structured ESG text data; performing text cleaning and denoising on the structured text data to obtain cleaned and denoised ESG text data; performing text normalization on the cleaned and denoised ESG text data to obtain normalized ESG text data; and adjusting the text structure of the normalized ESG text data to obtain target ESG text data.

[0042] Among them, the structured ESG text data can be standardized text data obtained by converting the ESG document to be analyzed through a set format conversion method. The specific formatting conversion method can be a regular expression or text segmentation, etc., which can be pre-set by the technician based on experience.

[0043] Among them, text cleaning and denoising processing can be to process the noise in the standardized text data and unify the format of the standardized text data, which can specifically include text processing methods such as character encoding unification, removal of special characters and non-printing characters, removal of punctuation marks, removal of redundant spaces and removal of web page tags. Among them, character encoding unification can be to encode all standardized text data with unified characters to avoid garbled text data due to the use of different encoding methods. For example, UTF-8 (Unicode Transformation Format-8-bit, Unicode conversion format-8 bits) and other encoding methods can be used to unify the text data. Removing special characters and non-printing characters can be to delete content characters such as control characters and line breaks that do not belong to the text. Removing punctuation marks can be to delete or retain the punctuation marks of the text data according to the actual content. For example, commas or periods between values ​​and units can be retained, while punctuation marks in other positions can be deleted. Removing redundant spaces and deleting web page tags may be performed by merging multiple consecutive spaces in the text data into one, and deleting web page tags in the text data, such as HTML (Hyper Text Markup Language, standard markup language) tags.

[0044] Among them, text normalization processing can be to divide the cleaned and denoised ESG text data into words, which can specifically include word segmentation, stop word removal, vocabulary normalization, and retention of professional terms, etc. Among them, word segmentation can be to split a continuous text string into individual words or sub-words. Stop word removal can be to delete the words in the text data that do not carry key information, such as words like "of", "and", etc. can be deleted. Vocabulary normalization can be to restore words to their basic forms through a stemming algorithm or a lemmatization technique. Exemplarily, words can be restored to their stem forms through a stemming algorithm, such as "running" can be restored to "run"; words can also be restored to their basic grammatical forms through a lemmatization technique, such as "better" can be restored to "good", etc. Retaining professional terms can be to retain the specific terms in the ESG evaluation and analysis process, and the specific terms can specifically include descriptions such as carbon emissions, energy use efficiency, sustainable development, and the number of employee training sessions.

[0045] Among them, text structure adjustment can be to divide and adjust the paragraphs and sentences of the text data according to the logical structure of the text data. Specifically, the normalized ESG text data can be divided into paragraphs or chapters according to the logical structure of the text data, and further can be subdivided into sentences.

[0046] This technical solution can perform text preprocessing on the ESG document to be analyzed. Specifically, through processing methods such as formatting conversion, text cleaning and denoising, text normalization, and text structure adjustment, the target ESG text data can be obtained, making the processing of ESG documents more standardized, and thus improving the accuracy and comprehensiveness of the recognition and evaluation of ESG documents.

[0047] S120: Input the target ESG text data into a pre-trained named entity recognition model to obtain a set of vocabulary sequences output by the model; the set of vocabulary sequences includes at least one entity vocabulary.

[0048] Among them, named entity recognition can refer to identifying and extracting specific entity nouns in the target ESG text data, which can specifically include named entities such as company names, policy names, and legal terms.

[0049] Among them, entity vocabulary can refer to vocabulary with clear meanings, which can specifically include quantitative indicators such as carbon emissions, energy use efficiency, the number of employee training sessions, and the number of social responsibility projects, as well as vocabulary for qualitative descriptions in aspects such as corporate governance, risk control, and community participation.

[0050] Specifically, a pre-trained named entity recognition model can be used to identify and extract quantitative indicators such as carbon emissions, energy efficiency, employee training times, number of social responsibility projects, and qualitative descriptions of corporate governance, risk control, community participation, etc. contained in the target ESG text data, and obtain a set of vocabulary sequences output by the named entity recognition model. For example, if the input text is "In 2023, XX Company announced that it will hold a conference on sustainable development in XX region and announced a reduction of 200 tons of carbon emissions throughout the year", the named entity vocabulary sequence that can be output by the named entity recognition model is "2023", "XX Company", "XX Region", "Sustainable Development", "Carbon Emissions" and "200 Tons". Among them, the named entity recognition model can be pre-trained by a technician.

[0051] Furthermore, the training process of the named entity recognition model may include steps such as data collection, sample selection and data cleaning, sample labeling, model training, and model testing and deployment.

[0052] Among them, data collection can be the collection of ESG document information, which can be obtained through public ESG report annual reports, industry databases and literature, or by crawling on public websites using crawler technology. Specifically, many companies and institutions will publicly release their ESG reports, which usually contain detailed information about the company's environmental responsibility, social responsibility, governance structure, etc. The report can be obtained directly from the company's official website, government website, international organizations, etc.; if possible, relevant documents can also be obtained from professional databases or industry reports, especially ESG reports involving specific industries. Finally, if it is not possible to directly obtain a large number of ESG reports, crawlers can also be used to crawl relevant documents or web page content from public websites to achieve data collection. Specific data collection can be collected by technical personnel based on actual conditions.

[0053] Among them, sample selection and data cleaning can be pre-processing of the acquired ESG document information, which can include data desensitization, removal of format information, removal of irrelevant paragraphs, and unified coding and formatting. The original ESG report can be converted into a structured, standardized text format that is easy to be processed by the deep learning model.

[0054] Among them, sample annotation can be to add named entity labels to ESG document data when using ESG document information to train deep learning models, and generate training sample sets and verification sample sets for deep learning models. Specifically, various types of key information can be annotated in the cleaned text data. For example, carbon emission regulations and sustainable development goals can be marked as environmental protection policies / regulations; operating income and profit margins can be marked as operating data; and educational funding can be marked as social responsibility projects. Specific sample annotation can be performed by technical personnel based on experience.

[0055] Among them, model training can be to convert the labeled data into the format required for model input, such as pairing each labeled entity with its category, and converting the data into a JSON format acceptable to the deep learning model. Use the training sample set to train the model, input the training sample set into the deep learning model, and obtain the prediction results of the deep learning model, wherein the deep learning model can be a long short-term memory network model, etc. Among them, the gradient descent method can be used to optimize the model parameters, and the parameters of the model are continuously updated so that the model can gradually recognize the named entities in the ESG text until the convergence conditions of the model training are met. Furthermore, the validation sample set can be used to tune the deep learning model. By monitoring the performance of the deep learning model on the validation sample set, the parameters of the deep learning model, such as the learning rate, optimizer, number of training rounds, etc., can be adjusted to further improve the recognition performance of the deep learning model for named entities in ESG documents.

[0056] The testing and deployment of the model can be to use the collected ESG reports as the test set to conduct a final evaluation of the trained model to understand the performance of the model on real data. When the model achieves the expected effect on the test set, the model can be deployed in the actual environment to perform real-time or batch ESG report named entity recognition.

[0057] S130: Perform lexical analysis on each entity word in the lexical sequence set to determine the ESG association index in each entity word.

[0058] Specifically, determining the ESG-related index in each entity vocabulary may be to determine a named entity recognition model, and the named entity vocabulary identified and extracted in the target ESG text data has characteristic indicators associated with ESG.

[0059] For example, continuing with the above example, the named entity vocabulary sequence output by the named entity recognition model is "2023", "XX Company", "XX Region", "Sustainable Development", "Carbon Emissions" and "200 Tons", then the ESG-related indicator features in each named entity vocabulary can be determined as "sustainable development", "carbon emissions" and "200 tons".

[0060] Optionally, lexical analysis is performed on each entity word in the lexical sequence set to determine the ESG association index of each entity word, including: performing a word vector conversion operation on each entity word in the lexical sequence set to obtain an entity word vector corresponding to each entity word; inputting the entity word vector corresponding to each entity word in the lexical sequence set into a pre-trained text feature extraction model to obtain context feature information corresponding to each entity word output by the text feature extraction model; and determining the ESG association index based on the context feature information corresponding to each entity word.

[0061] The entity word vector can be a word vector generated by mapping entity words to a vector space of fixed dimension. Specifically, the word vector can be generated based on word embedding method or knowledge graph. The specific generation method can be set by technicians based on experience.

[0062] Specifically, the obtained vocabulary sequence set can be converted into word vectors, and the entity word vector sequence corresponding to the vocabulary sequence can be obtained. The entity word vector corresponding to each entity word is used as the input of the text feature extraction model, and then the context information corresponding to each entity word is obtained and used as the output of the text feature extraction model, wherein the text feature extraction model can be a bidirectional long short-term memory network or a long short-term memory network, which can be specifically set by technical personnel based on experience. Furthermore, the entity word can also be determined as an ESG-related indicator in combination with the context feature information corresponding to each entity word.

[0063] This technical solution can obtain the corresponding context feature information according to the entity word vectors corresponding to each entity vocabulary. Furthermore, the ESG association index can be determined according to the context feature information, and the accuracy and comprehensiveness of the identification and evaluation of ESG files can be improved according to the dependency relationship between the contexts in the text data.

[0064] Optionally, ESG-related indicators are determined based on contextual feature information corresponding to each entity vocabulary, including: inputting each entity vocabulary and its corresponding contextual feature information into a pre-trained entity category label prediction model to obtain entity category labels corresponding to each entity vocabulary output by the entity category label prediction model; and determining ESG-related indicators based on the entity category labels of each entity vocabulary.

[0065] Among them, the entity category can be the entity type corresponding to each entity vocabulary. For example, the entity type corresponding to "2023" can be "date", and the entity type corresponding to "sustainable development" can be "social responsibility project", etc. The specific entity category can be pre-set by technical personnel based on experience.

[0066] Specifically, the context feature information of each entity vocabulary can be input into the label prediction model, and the output result of the label prediction model can be the annotation of the predicted label for each input vocabulary, wherein the predicted label used for annotation is the named entity category corresponding to each entity vocabulary. Specifically, the label represents the entity category of each entity vocabulary in the entity vocabulary sequence, which can be a label such as "company name", "date" or "place". Furthermore, the label prediction model can output a label sequence with the same length as the input sequence, wherein each predicted label corresponds to an input entity vocabulary, and identifies the entity type of the vocabulary in a specific context, and determines the indicator vocabulary associated with ESG. Exemplarily, continuing the above example, a label sequence with the same sequence length can be output according to "2023", "XX company", "XX region", "sustainable development", "carbon emissions" and "200 tons", and the label sequence can be "date", "company", "place", "social responsibility project", "environmental project" and "quantity", and then according to the label sequence and the entity type of the corresponding vocabulary in the context, the entity vocabulary "sustainable development", "carbon emissions" and "200 tons" can be determined as ESG related indicators.

[0067] This technical solution can determine ESG-related indicators based on the entity category labels corresponding to each entity vocabulary, thereby improving the efficiency of ESG document recognition and the accuracy and comprehensiveness of entity naming vocabulary recognition and extraction.

[0068] S140. Analyze the indicator characteristics of ESG-related indicators to determine the ESG evaluation results of the target enterprise.

[0069] Among them, the ESG evaluation result can be a comprehensive assessment of the company's environment, social responsibility and governance. For example, if the company has a good performance in reducing carbon emissions, uses renewable energy, has a low carbon footprint, and has good water resource management, it can score 80 points in terms of environment; the company provides employees with good benefits and a safe working environment, and has active investment in community activities and social responsibility, but needs to improve customer data protection, and the social score can be 75 points; the board of directors has a sound structure, strong independence, and follows high standards of business ethics. However, there is some room for improvement in transparency and remuneration mechanisms, and the governance score can be 70 points. In the end, the ESG evaluation result of the target company can be 75 points.

[0070] S150. Perform business processing on the target enterprise based on the ESG assessment results of the target enterprise.

[0071] Among them, the business of the target enterprise can be processed according to the comprehensive evaluation of the efforts made by the target enterprise in terms of environment, social responsibility and governance. Specifically, the corresponding ESG scoring level can be set to divide the target enterprise, and then the business processing flow can be determined according to the corresponding scoring level of the target enterprise. Exemplary, continuing the above example, if the target enterprise requests a loan business, and the ESG scoring level corresponding to the loan business is: 80-100 points, the loan business can be executed; 60-80 points, further investigation of the target enterprise is required, and whether the loan business can be executed is determined based on the results of the further investigation; 0-60 points, the loan business can be rejected. As above, if the ESG evaluation result of the target enterprise is 75 points, it is necessary to further investigate the target enterprise, and determine whether to execute the loan business initiated by the target enterprise based on the results of the investigation.

[0072] The technical solution of the embodiment of the present invention obtains the ESG document to be analyzed of the enterprise to be evaluated, and performs text preprocessing on the ESG document to be analyzed to obtain the target ESG text data; inputs the target ESG text data into the pre-trained named entity recognition model to obtain a vocabulary sequence set output by the model; the vocabulary sequence set includes at least one entity vocabulary; performs vocabulary analysis on each entity vocabulary in the vocabulary sequence set to determine the ESG related indicators in each entity vocabulary; performs indicator feature analysis on the ESG related indicators to determine the ESG evaluation results of the target enterprise; and performs business processing on the target enterprise according to the ESG evaluation results of the target enterprise. This technical solution can use the named entity recognition model to extract entity vocabulary in the ESG document and determine the ESG related indicators in the entity vocabulary without manual word segmentation, thereby improving the comprehensiveness and efficiency of the identification and analysis of the ESG document; it can also process the business according to the ESG evaluation results of the target enterprise, thereby improving the accuracy and comprehensiveness of the identification and evaluation of the ESG document, and realizing the precise processing of the enterprise-related business.

[0073] Embodiment 2

[0074] Figure 2 This is a flow chart of a document-based business processing method provided in the second embodiment of the present invention. Based on the above embodiments, this embodiment further optimizes the above document-based business processing method.

[0075] Furthermore, the step of "analyzing the words of each entity in the word sequence set and determining the ESG-related indicators in each entity word" is refined into "determining the indicator data of the ESG-related indicators based on the ESG documents to be analyzed; determining the initial weights of the ESG-related indicators; and determining the ESG evaluation results of the target enterprise based on the indicator data and initial weights of the ESG-related indicators." to improve the way of identifying and evaluating ESG documents. Figure 2 As shown, the method includes:

[0076] S210 , obtaining the ESG document to be analyzed of the enterprise to be evaluated, and performing text preprocessing on the ESG document to be analyzed to obtain target ESG text data.

[0077] S220. Input the target ESG text data into a pre-trained named entity recognition model to obtain a vocabulary sequence set output by the model; the vocabulary sequence set includes at least one entity vocabulary.

[0078] S230: Perform lexical analysis on each entity word in the lexical sequence set to determine the ESG association index in each entity word.

[0079] S240. Determine indicator data of ESG-related indicators based on the ESG document to be analyzed.

[0080] Among them, the indicator data for determining ESG-related indicators can specifically be to determine the indicator characteristics related to the environment, society and governance in the ESG documents to be analyzed, such as descriptions of carbon emissions, energy efficiency, number of employee trainings, number of social responsibility projects and community participation.

[0081] S250. Determine the initial weights of ESG-related indicators.

[0082] Specifically, the initial weights of the evaluation proportions of environment, society and governance in the ESG documents to be analyzed can be obtained, where the initial weights can be pre-set by technical personnel based on experience. For example, environmental strategies can account for 30%, social strategies can account for 30% and corporate governance can account for 40%.

[0083] S260. Determine the ESG evaluation results of the target enterprise based on the indicator data and initial weights of the ESG-related indicators.

[0084] Specifically, the ESG evaluation results of the target enterprise can be calculated based on the indicator characteristics related to the environment, society and governance in the ESG documents to be analyzed of the target enterprise and the corresponding weights. For example, if the enterprise scores 80 points in terms of environment, 70 points in terms of society and 90 points in terms of corporate governance, the ESG evaluation result of the enterprise can be 81 points.

[0085] Optionally, the ESG evaluation results of the target enterprise are determined based on the indicator data and initial weights of the ESG-related indicators, including: inputting the ESG-related indicators and their corresponding indicator data into a pre-trained indicator status analysis model to obtain status scores corresponding to the ESG-related indicators output by the indicator status analysis model; determining the target weights of the ESG-related indicators based on the initial weights and status scores of the ESG-related indicators; and determining the ESG evaluation results of the target enterprise based on the target weights and status scores of the ESG-related indicators.

[0086] Specifically, the ESG related indicators and their corresponding indicator data can be marked to indicate the classification information corresponding to each related indicator, and input into the pre-trained indicator state analysis model. Furthermore, the pre-trained indicator state analysis model can be used as the basic structure, and a multi-layer perceptron classifier layer can be added on top to generate the probability distribution of each indicator state category. At the same time, a loss function suitable for the sentiment classification task is selected at the model training stage, such as the cross entropy loss function, and a suitable optimizer, such as the learning rate strategy, is configured to update the model parameters. During the training process, the training set can be iterated for multiple rounds, and the performance indicators on the validation set, such as accuracy, F1 score, etc., can be monitored to ensure that the model does not overfit.

[0087] Furthermore, after the training is completed, the indicator text fragments extracted from the test set are sent to the fine-tuned indicator state analysis model after the same preprocessing process. The indicator state analysis model can output the category score results of each ESG-related indicator and its corresponding indicator data, and then determine the target weight of each ESG-related indicator according to the category score results of each ESG-related indicator and its corresponding indicator data. For example, continuing the above example, the initial weight of the environmental policy can be 30%, the initial weight of the social strategy can be 30%, and the initial weight of the corporate governance can be 40%; and after the indicator state analysis model, it is determined that the environmental strategy is more important, the corporate governance is second, and the social strategy is the lowest, then the weight of the environmental policy can be adjusted by 50%, the initial weight of the corporate governance can be adjusted to 30%, and the weight of the social strategy can be adjusted to 20%, and then the ESG evaluation results of the target enterprise can be determined according to the adjusted weight ratios.

[0088] This technical solution can determine the target weights corresponding to the ESG-related indicators according to the status scores corresponding to the ESG-related indicators, and then determine the ESG evaluation results of the target enterprise. Through the dynamic weight adjustment strategy, the evaluation impact of each indicator is quantified and calculated, which more realistically reflects the impact of each indicator on the ESG performance of the enterprise, and further improves the comprehensiveness of the identification and evaluation of ESG documents.

[0089] S270. Perform business processing on the target enterprise based on the ESG assessment results of the target enterprise.

[0090] The technical solution of the embodiment of the present invention obtains the ESG document to be analyzed of the enterprise to be evaluated, and performs text preprocessing on the ESG document to be analyzed to obtain the target ESG text data; inputs the target ESG text data into the pre-trained named entity recognition model to obtain a vocabulary sequence set output by the model; the vocabulary sequence set includes at least one entity vocabulary; performs vocabulary analysis on each entity vocabulary in the vocabulary sequence set to determine the ESG related indicators in each entity vocabulary; performs indicator feature analysis on the ESG related indicators to determine the ESG evaluation results of the target enterprise; and performs business processing on the target enterprise according to the ESG evaluation results of the target enterprise. This technical solution can use the named entity recognition model to extract entity vocabulary in the ESG document and determine the ESG related indicators in the entity vocabulary without manual word segmentation, thereby improving the comprehensiveness and efficiency of the identification and analysis of the ESG document; further, it can also improve the accuracy and comprehensiveness of the identification and evaluation of the ESG document according to the indicator feature analysis of the ESG related indicators and the evaluation results obtained, and realize the precise processing of the enterprise-related business.

[0091] Embodiment 3

[0092] Figure 3 A schematic diagram of the structure of a document-based business processing device provided in Embodiment 3 of the present invention. A document-based business processing device provided in an embodiment of the present invention can be used to conduct a comprehensive and accurate evaluation of an enterprise and decide whether to conduct business processing on the enterprise based on the evaluation result. The document-based business processing device can be implemented in the form of hardware and / or software, such as Figure 3 As shown, it specifically includes: a text data acquisition module 310, a vocabulary sequence set generation module 320, a vocabulary analysis module 330, an indicator feature analysis module 340 and a business processing module 350. Among them,

[0093] The text data acquisition module 310 is used to acquire the ESG document to be analyzed of the enterprise to be evaluated, and perform text preprocessing on the ESG document to be analyzed to obtain target ESG text data.

[0094] The vocabulary sequence set generation module 320 is used to input the target ESG text data into a pre-trained named entity recognition model to obtain a vocabulary sequence set output by the model; the vocabulary sequence set includes at least one entity vocabulary.

[0095] The vocabulary analysis module 330 is used to perform vocabulary analysis on each of the entity words in the vocabulary sequence set to determine the ESG association index in each of the entity words.

[0096] The indicator characteristic analysis module 340 is used to perform indicator characteristic analysis on the ESG-related indicators to determine the ESG evaluation results of the target enterprise.

[0097] The business processing module 350 is used to perform business processing on the target enterprise according to the ESG evaluation result of the target enterprise.

[0098] This technical solution can use the named entity recognition model to extract entity vocabulary in ESG documents and determine the ESG-related indicators in the entity vocabulary without manual word segmentation, thereby improving the comprehensiveness and efficiency of ESG document recognition and analysis; further, it can also improve the accuracy and comprehensiveness of ESG document recognition and evaluation based on indicator feature analysis of ESG-related indicators and the obtained evaluation results, and realize accurate processing of enterprise-related businesses.

[0099] Optionally, the vocabulary analysis module 330 includes:

[0100] The entity word vector acquisition unit is used to perform a word vector conversion operation on each of the entity words in the word sequence set to obtain the entity word vector corresponding to each of the entity words.

[0101] The feature information acquisition unit is used to input the entity word vector corresponding to each entity word in the vocabulary sequence set into a pre-trained text feature extraction model to obtain context feature information corresponding to each entity word output by the text feature extraction model.

[0102] The association index determination unit is used to determine the ESG association index according to the context feature information corresponding to each of the entity words.

[0103] Optionally, the associated indicator determination unit is specifically used to:

[0104] Inputting each entity word and its corresponding context feature information into a pre-trained entity category label prediction model to obtain entity category labels corresponding to each entity word output by the entity category label prediction model;

[0105] According to the entity category label of each entity vocabulary, an ESG related indicator is determined.

[0106] Optionally, the indicator feature analysis module 340 includes:

[0107] An indicator data determination unit is used to determine the indicator data of the ESG-related indicator based on the ESG document to be analyzed.

[0108] The initial weight determination unit is used to determine the initial weight of the ESG related indicator.

[0109] The evaluation result determination unit is used to determine the ESG evaluation result of the target enterprise based on the indicator data and initial weights of the ESG-related indicators.

[0110] Optionally, the evaluation result determination unit is specifically used to:

[0111] Input the ESG-related indicators and their corresponding indicator data into a pre-trained indicator state analysis model to obtain the state scores corresponding to the ESG-related indicators output by the indicator state analysis model;

[0112] Determining the target weight of the ESG-related indicator based on the initial weight and status score of the ESG-related indicator;

[0113] The ESG evaluation result of the target enterprise is determined based on the target weight and status score of the ESG-related indicators.

[0114] Optionally, the text data acquisition module 310 is specifically used for:

[0115] Formatting and converting the ESG document to be analyzed to obtain structured ESG text data;

[0116] Performing text cleaning and denoising processing on the structured text data to obtain cleaned and denoised ESG text data;

[0117] Performing text normalization processing on the cleaned and denoised ESG text data to obtain normalized ESG text data;

[0118] The text structure of the standardized ESG text data is adjusted to obtain target ESG text data.

[0119] The document-based business processing device provided in the embodiment of the present invention can execute the document-based business processing method provided in any embodiment of the present invention, and has the corresponding functional modules and beneficial effects of the execution method.

[0120] Embodiment 4

[0121] Figure 4A schematic diagram of the structure of an electronic device 40 that can be used to implement an embodiment of the present invention is shown. The electronic device is intended to represent various forms of digital computers, such as laptop computers, desktop computers, workstations, personal digital assistants, servers, blade servers, mainframe computers, and other suitable computers. The electronic device can also represent various forms of mobile devices, such as personal digital processing, cellular phones, smart phones, wearable devices (such as helmets, glasses, watches, etc.) and other similar computing devices. The components shown herein, their connections and relationships, and their functions are merely examples and are not intended to limit the implementation of the present invention described and / or required herein.

[0122] like Figure 4 As shown, the electronic device 40 includes at least one processor 41, and a memory connected to the at least one processor 41, such as a read-only memory (ROM) 42, a random access memory (RAM) 43, etc., wherein the memory stores a computer program that can be executed by at least one processor, and the processor 41 can perform various appropriate actions and processes according to the computer program stored in the read-only memory (ROM) 42 or the computer program loaded from the storage unit 48 to the random access memory (RAM) 43. In RAM43, various programs and data required for the operation of the electronic device 40 can also be stored. The processor 41, ROM42 and RAM43 are connected to each other through a bus 44. An input / output (I / O) interface 45 is also connected to the bus 44.

[0123] A number of components in the electronic device 40 are connected to the I / O interface 45, including: an input unit 46, such as a keyboard, a mouse, etc.; an output unit 47, such as various types of displays, speakers, etc.; a storage unit 48, such as a disk, an optical disk, etc.; and a communication unit 49, such as a network card, a modem, a wireless communication transceiver, etc. The communication unit 49 allows the electronic device 40 to exchange information / data with other devices through a computer network such as the Internet and / or various telecommunication networks.

[0124] The processor 41 may be a variety of general and / or special processing components with processing and computing capabilities. Some examples of the processor 41 include, but are not limited to, a central processing unit (CPU), a graphics processing unit (GPU), various special artificial intelligence (AI) computing chips, various processors running machine learning model algorithms, a digital signal processor (DSP), and any appropriate processor, controller, microcontroller, etc. The processor 41 executes the various methods and processes described above, such as a document-based business processing method.

[0125] In some embodiments, the document-based business processing method can be implemented as a computer program, which is tangibly contained in a computer-readable storage medium, such as a storage unit 48. In some embodiments, part or all of the computer program can be loaded and / or installed on the electronic device 40 via the ROM 42 and / or the communication unit 49. When the computer program is loaded into the RAM 43 and executed by the processor 41, one or more steps of the document-based business processing method described above can be performed. Alternatively, in other embodiments, the processor 41 can be configured to execute the document-based business processing method in any other appropriate manner (for example, by means of firmware).

[0126] Various implementations of the systems and techniques described above herein can be implemented in digital electronic circuit systems, integrated circuit systems, field programmable gate arrays (FPGAs), application specific integrated circuits (ASICs), application specific standard products (ASSPs), systems on chips (SOCs), load programmable logic devices (CPLDs), computer hardware, firmware, software, and / or combinations thereof. These various implementations can include: being implemented in one or more computer programs that can be executed and / or interpreted on a programmable system including at least one programmable processor, which can be a special purpose or general purpose programmable processor that can receive data and instructions from a storage system, at least one input device, and at least one output device, and transmit data and instructions to the storage system, the at least one input device, and the at least one output device.

[0127] Computer programs for implementing the methods of the present invention may be written in any combination of one or more programming languages. These computer programs may be provided to a processor of a general-purpose computer, a special-purpose computer, or other programmable data processing device, so that when the computer program is executed by the processor, the functions / operations specified in the flow chart and / or block diagram are implemented. The computer program may be executed entirely on the machine, partially on the machine, partially on the machine and partially on a remote machine as a stand-alone software package, or entirely on a remote machine or server.

[0128] In the context of the present invention, a computer-readable storage medium may be a tangible medium that may contain or store a computer program for use by or in combination with an instruction execution system, device or equipment. A computer-readable storage medium may include, but is not limited to, electronic, magnetic, optical, electromagnetic, infrared, or semiconductor systems, devices or equipment, or any suitable combination of the foregoing. Alternatively, a computer-readable storage medium may be a machine-readable signal medium. A more specific example of a machine-readable storage medium may include an electrical connection based on one or more lines, a portable computer disk, a hard disk, a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), an optical fiber, a portable compact disk read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination of the foregoing.

[0129] To provide interaction with a user, the systems and techniques described herein may be implemented on an electronic device having: a display device (e.g., a CRT (cathode ray tube) or LCD (liquid crystal display) monitor) for displaying information to the user; and a keyboard and a pointing device (e.g., a mouse or trackball) through which the user can provide input to the electronic device. Other types of devices may also be used to provide interaction with the user; for example, the feedback provided to the user may be any form of sensory feedback (e.g., visual feedback, auditory feedback, or tactile feedback); and input from the user may be received in any form (including acoustic input, voice input, or tactile input).

[0130] The systems and techniques described herein may be implemented in a computing system that includes backend components (e.g., as a data server), or a computing system that includes middleware components (e.g., an application server), or a computing system that includes frontend components (e.g., a user computer with a graphical user interface or a web browser through which a user can interact with implementations of the systems and techniques described herein), or a computing system that includes any combination of such backend components, middleware components, or frontend components. The components of the system may be interconnected by any form or medium of digital data communication (e.g., a communication network). Examples of communication networks include: a local area network (LAN), a wide area network (WAN), a blockchain network, and the Internet.

[0131] A computing system may include a client and a server. The client and the server are generally remote from each other and usually interact through a communication network. The client and server relationship is generated by computer programs running on the corresponding computers and having a client-server relationship with each other. The server may be a cloud server, also known as a cloud computing server or cloud host, which is a host product in the cloud computing service system to solve the defects of difficult management and weak business scalability in traditional physical hosts and VPS services.

[0132] It should be understood that the various forms of processes shown above can be used to reorder, add or delete steps. For example, the steps described in the present invention can be executed in parallel, sequentially or in different orders, as long as the desired results of the technical solution of the present invention can be achieved, and this document does not limit this.

[0133] The above specific implementations do not constitute a limitation on the protection scope of the present invention. It should be understood by those skilled in the art that various modifications, combinations, sub-combinations and substitutions can be made according to design requirements and other factors. Any modification, equivalent substitution and improvement made within the spirit and principle of the present invention should be included in the protection scope of the present invention.

Claims

1. A document-based business processing method, characterized in that: include: Obtaining the ESG document to be analyzed of the enterprise to be evaluated, and performing text preprocessing on the ESG document to be analyzed to obtain target ESG text data; Input the target ESG text data into a pre-trained named entity recognition model to obtain a vocabulary sequence set output by the model; the vocabulary sequence set includes at least one entity vocabulary; Performing lexical analysis on each of the entity words in the lexical sequence set to determine an ESG relevance index in each of the entity words; Conducting an indicator characteristic analysis on the ESG-related indicators to determine the ESG evaluation results of the target enterprise; Perform business processing on the target enterprise based on the ESG evaluation results of the target enterprise.

2. The method according to claim 1, characterized in that The performing lexical analysis on each of the entity words in the lexical sequence set to determine the ESG association index in each of the entity words includes: Performing a word vector conversion operation on each of the entity words in the word sequence set to obtain entity word vectors corresponding to each of the entity words; Inputting the entity word vector corresponding to each of the entity words in the vocabulary sequence set into a pre-trained text feature extraction model to obtain context feature information corresponding to each of the entity words output by the text feature extraction model; The ESG relevance index is determined according to the context feature information corresponding to each of the entity words.

3. The method according to claim 2, characterized in that Determining the ESG relevance index according to the context feature information corresponding to each of the entity words includes: Inputting each entity word and its corresponding context feature information into a pre-trained entity category label prediction model to obtain entity category labels corresponding to each entity word output by the entity category label prediction model; According to the entity category label of each entity vocabulary, an ESG related indicator is determined.

4. The method according to claim 1, characterized in that: The performing of indicator characteristic analysis on the ESG-related indicators to determine the ESG evaluation results of the target enterprise includes: Determining indicator data of the ESG-related indicator based on the ESG document to be analyzed; Determining the initial weights of the ESG-related indicators; The ESG evaluation results of the target enterprise are determined based on the indicator data and initial weights of the ESG-related indicators.

5. The method according to claim 4, characterized in that Determining the ESG evaluation results of the target enterprise based on the indicator data and initial weights of the ESG-related indicators includes: Input the ESG-related indicators and their corresponding indicator data into a pre-trained indicator state analysis model to obtain the state scores corresponding to the ESG-related indicators output by the indicator state analysis model; Determining the target weight of the ESG-related indicator based on the initial weight and status score of the ESG-related indicator; The ESG evaluation result of the target enterprise is determined based on the target weight and status score of the ESG-related indicators.

6. The method according to claim 1, characterized in that The step of performing text preprocessing on the ESG document to be analyzed to obtain target ESG text data includes: Formatting and converting the ESG document to be analyzed to obtain structured ESG text data; Performing text cleaning and denoising processing on the structured text data to obtain cleaned and denoised ESG text data; Performing text normalization processing on the cleaned and denoised ESG text data to obtain normalized ESG text data; The text structure of the standardized ESG text data is adjusted to obtain target ESG text data.

7. A document-based business processing device, characterized in that: include: A text data acquisition module is used to acquire the ESG document to be analyzed of the enterprise to be evaluated, and perform text preprocessing on the ESG document to be analyzed to obtain target ESG text data; A vocabulary sequence set generation module, used for inputting the target ESG text data into a pre-trained named entity recognition model to obtain a vocabulary sequence set output by the model; the vocabulary sequence set includes at least one entity vocabulary; A vocabulary analysis module, used to perform vocabulary analysis on each of the entity words in the vocabulary sequence set to determine the ESG association index in each of the entity words; An indicator characteristic analysis module, used to perform indicator characteristic analysis on the ESG-related indicators to determine the ESG evaluation results of the target enterprise; The business processing module is used to perform business processing on the target enterprise according to the ESG evaluation result of the target enterprise.

8. 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 a computer program executed by the at least one processor, and the computer program is executed by the at least one processor so that the at least one processor can execute the document-based business processing method according to any one of claims 1 to 6.

9. A computer-readable storage medium, characterized in that: The computer-readable storage medium stores computer instructions, and the computer instructions are used to enable a processor to implement the document-based business processing method according to any one of claims 1 to 6 when executed.

10. A computer program product, characterized in that The computer program product comprises a computer program, and when the computer program is executed by a processor, the computer program implements the document-based business processing method according to any one of claims 1 to 6.

Images