Business data processing method and device, storage medium and electronic equipment

Abstract

The invention discloses a business data processing method and device, a storage medium and electronic equipment. The method comprises the steps of determining a processing category of business evaluation data based on a data attribute of the business evaluation data; under the condition that the processing category of the service evaluation data is a first category, executing a first identification operation on the service evaluation data to obtain a first processing result, the first processing result reflecting a service risk of a service associated with the service evaluation data; and under the condition that the processing category of the business evaluation data is a second category, executing a second identification operation on the business evaluation data to obtain a second processing result, the second processing result reflecting the business risk of the business associated with the business evaluation data. The technical problem that the risk prediction accuracy of the business data is low is solved.

Description

Technical Field

[0001] The present application relates to the field of computers, and more specifically, to a method and device for processing business data, a storage medium, and an electronic device. Background Art

[0002] When predicting risks for business data in related technologies, traditional machine learning models or neural network models are mainly relied upon. However, when faced with massive, complex, and changeable business evaluation data, some traditional models have fast processing speeds but limited depth of understanding, or some models have strong comprehension capabilities but slow processing speeds and high resource consumption. In actual applications, it is impossible to balance analysis efficiency and prediction accuracy, resulting in a technical problem of low accuracy in risk prediction of business data.

[0003] To address the above-mentioned problems, no effective solutions have been proposed so far. Summary of the Invention

[0004] The embodiments of the present application provide a method and device for processing business data, a storage medium, and an electronic device to at least solve the technical problem of low accuracy in risk prediction of business data.

[0005] According to one aspect of an embodiment of the present application, a method for processing business data is provided, comprising: determining a processing category of the business evaluation data based on data attributes of the business evaluation data, wherein the data attributes of the business evaluation data include at least one of data volume and data source address; when the processing category of the business evaluation data is a first category, performing a first identification operation on the business evaluation data to obtain a first processing result, wherein the first identification operation is used to determine the emotional tendency type corresponding to the business evaluation data and the entity name in the business evaluation data, and the first processing result includes at least one of the risk level, emotional tendency type, entity name and cause analysis of the business evaluation data, and the first processing result reflects the business risk of the business associated with the business evaluation data; when the processing category of the business evaluation data is a second category, performing a second identification operation on the business evaluation data to obtain a second processing result, wherein the first identification operation is used to determine the context association of the business evaluation data, the emotional tendency type corresponding to the business evaluation data and the entity name in the business evaluation data, and the second processing result includes at least one of the risk level, emotional tendency type, entity name and cause analysis of the business evaluation data, and the second processing result reflects the business risk of the business associated with the business evaluation data.

[0006] According to another aspect of an embodiment of the present application, a business data processing device is also provided, including: a determination module, configured to determine a processing category of the business evaluation data based on data attributes of the business evaluation data, wherein the data attributes of the business evaluation data include at least one of data volume and data source address; a first identification module, configured to perform a first identification operation on the business evaluation data when the processing category of the business evaluation data is the first category, to obtain a first processing result, wherein the first identification operation is used to determine the emotional tendency type corresponding to the business evaluation data and the entity name in the business evaluation data, the first processing result includes at least one of the risk level, emotional tendency type, entity name and cause analysis of the business evaluation data, and the first processing result reflects the business risk of the business associated with the business evaluation data; a second identification module, configured to perform a second identification operation on the business evaluation data when the processing category of the business evaluation data is the second category, to obtain a second processing result, wherein the first identification operation is used to determine the context association of the business evaluation data, the emotional tendency type corresponding to the business evaluation data and the entity name in the business evaluation data, the second processing result includes at least one of the risk level, emotional tendency type, entity name and cause analysis of the business evaluation data, and the second processing result reflects the business risk of the business associated with the business evaluation data.

[0007] Optionally, the device is used to perform a first recognition operation on the business evaluation data to obtain a first processing result when the processing category of the business evaluation data is the first category in the following manner: input the business evaluation data into a first recognition model, wherein the first recognition model includes a keyword matching module and a sentiment analysis module; use the keyword matching module to determine the semantic similarity between the business evaluation data and a preset target keyword library, and determine the entity name based on the semantic similarity; use the sentiment analysis module to determine the sentiment tendency type of the business evaluation data; and generate the first processing result based on the sentiment tendency type corresponding to the business evaluation data and the entity name in the business evaluation data.

[0008] Optionally, the device is used to generate the first processing result according to the emotional tendency type corresponding to the business evaluation data and the entity name in the business evaluation data in the following manner: when the emotional tendency type corresponding to the business evaluation data is a positive type, determining that the risk level of the business evaluation data is the first level, and generating a positive cause analysis associated with the entity name in the business evaluation data, wherein the cause analysis includes the positive cause analysis; when the emotional tendency type corresponding to the business evaluation data is a neutral type, determining that the risk level of the business evaluation data is the second level, and generating a neutral cause analysis associated with the entity name in the business evaluation data, wherein the cause analysis includes the neutral cause analysis, and the business risk of the second level is higher than the business risk of the first level; when the emotional tendency type corresponding to the business evaluation data is a negative type, determining that the risk level of the business evaluation data is the third level, and generating a negative cause analysis associated with the entity name, wherein the cause analysis includes the negative cause analysis, and the business risk of the third level is higher than the business risk of the second level.

[0009] Optionally, the device is used to perform a second recognition operation on the business evaluation data to obtain a second processing result when the processing category of the business evaluation data is the second category in the following manner: input the business evaluation data into a second recognition model, wherein the second recognition model includes an event label recognition module, a sentiment tendency recognition module, and an entity recognition module; use the event label recognition module to process the business evaluation data to obtain the context association of the business evaluation data; use the entity recognition module to process the business evaluation data and the context association of the business evaluation data to obtain the sentiment tendency type corresponding to the business evaluation data; use the sentiment tendency recognition module to process the business evaluation data and the context association of the business evaluation data to obtain the entity name in the business evaluation data; generate the second processing result according to the sentiment tendency type corresponding to the business evaluation data and the entity name in the business evaluation data.

[0010] Optionally, the device is used to generate the second processing result according to the context association of the business evaluation data, the emotional tendency type corresponding to the business evaluation data and the entity name in the business evaluation data in the following manner: when the emotional tendency type corresponding to the business evaluation data is a positive type, determine that the risk level of the business evaluation data is the fourth level, and generate a positive cause analysis associated with the entity name in the business evaluation data, wherein the cause analysis includes the positive cause analysis; when the emotional tendency type corresponding to the business evaluation data is a neutral type, determine that the risk level of the business evaluation data is the fifth level, and generate a neutral cause analysis associated with the entity name in the business evaluation data, wherein the cause analysis includes the neutral cause analysis, and the business risk of the fifth level is higher than the business risk of the fourth level; when the emotional tendency type corresponding to the business evaluation data is a negative type, determine that the risk level of the business evaluation data is the sixth level, and generate a negative cause analysis associated with the entity name, wherein the cause analysis includes the negative cause analysis, and the business risk of the sixth level is higher than the business risk of the fifth level.

[0011] Optionally, the device is used to determine the processing category of the business evaluation data based on the data attributes of the business evaluation data in the following manner: when the data volume of the business evaluation data is less than a preset data volume threshold, determine that the processing category of the business evaluation data is the first category; when the data volume of the business evaluation data is greater than or equal to the data volume threshold, determine that the processing category of the business evaluation data is the second category.

[0012] Optionally, the device is used to determine the processing category of the business evaluation data based on the data attributes of the business evaluation data in the following manner: when the data source address of the business evaluation data belongs to the preset target data source address, determine that the processing category of the business evaluation data is the first category; when the data source address of the business evaluation data does not belong to the preset target data source address, determine that the processing category of the business evaluation data is the second category.

[0013] According to another aspect of the embodiments of the present application, a computer-readable storage medium is provided, in which a computer program is stored. The computer program is configured to execute the above-mentioned method for processing business data when running.

[0014] According to another aspect of the embodiments of the present application, a computer program product or computer program is provided, the computer program product or computer program including computer instructions stored in a computer-readable storage medium. A processor of a computer device reads the computer instructions from the computer-readable storage medium and executes the computer instructions, causing the computer device to perform the above-described method for processing business data.

[0015] According to another aspect of an embodiment of the present application, an electronic device is provided, including a memory and a processor, wherein the memory stores a computer program, and the processor is configured to execute the business data processing method through the computer program.

[0016] In an embodiment of the present application, a processing category of the business evaluation data is determined based on the data attributes of the business evaluation data, wherein the data attributes of the business evaluation data include at least one of the data volume and the data source address; when the processing category of the business evaluation data is the first category, a first identification operation is performed on the business evaluation data to obtain a first processing result, wherein the first identification operation is used to determine the emotional tendency type corresponding to the business evaluation data and the entity name in the business evaluation data, and the first processing result includes at least one of the risk level, emotional tendency type, entity name and cause analysis of the business evaluation data, and the first processing result reflects the business risk of the business associated with the business evaluation data; when the processing category of the business evaluation data is the second category, a second identification operation is performed on the business evaluation data to obtain a second processing result, wherein, The first identification operation is used to determine the contextual association of the business evaluation data, the sentiment tendency type corresponding to the business evaluation data, and the entity name in the business evaluation data. The second processing result includes at least one of the risk level, sentiment tendency type, entity name and cause analysis of the business evaluation data. The second processing result reflects the business risk of the business associated with the business evaluation data. By intelligently allocating data of different complexities to the corresponding optimized first identification model to perform the first identification operation and the second identification model to perform the second identification operation, the purpose of efficiently and accurately identifying and evaluating business risks is achieved, thereby achieving the technical effect of significantly improving the accuracy of risk prediction while maintaining a high processing speed, and thus solving the technical problem of low risk prediction accuracy in traditional business data risk prediction methods due to a single model and low processing efficiency. BRIEF DESCRIPTION OF THE DRAWINGS

[0017] The drawings described herein are used to provide a further understanding of the present application and constitute a part of the present application. The illustrative embodiments of the present application and their descriptions are used to explain the present application and do not constitute an improper limitation on the present application. In the drawings:

[0018] Figure 1is a schematic diagram of an application environment of an optional business data processing method according to an embodiment of the present application;

[0019] Figure 2 This is a flowchart of an optional method for processing business data according to an embodiment of the present application;

[0020] Figure 3 is a schematic diagram of an optional method for processing business data according to an embodiment of the present application;

[0021] Figure 4 is a schematic diagram of an optional method for processing business data according to an embodiment of the present application;

[0022] Figure 5 It is a structural diagram of an optional business data processing device according to an embodiment of the present application. DETAILED DESCRIPTION

[0023] In order to enable those skilled in the art to better understand the present invention, the following will clearly and completely describe the technical solutions in the embodiments of the present invention 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 in the present invention, all other embodiments obtained by ordinary technicians in this field without making creative efforts should fall within the scope of protection of this application.

[0024] It should be noted that the terms "first", "second", etc. in the specification and claims of the present application and the above-mentioned drawings are used to distinguish similar objects and are not necessarily used to describe a specific order or sequential order. It should be understood that the data used in this way can be interchangeable where appropriate, so that the embodiments of the present application described herein can be implemented in a sequence other than those illustrated or described herein. In addition, the terms "including" and "having" and any of their variations are intended to cover non-exclusive inclusions, for example, a process, method, system, product or device comprising a series of steps or units is not necessarily limited to those steps or units clearly listed, but may include other steps or units that are not clearly listed or inherent to these processes, methods, products or devices.

[0025] The present application will be described below with reference to the following embodiments:

[0026] According to one aspect of an embodiment of the present application, a method for processing business data is provided. Optionally, in this embodiment, the method for processing business data can be applied to Figure 1 In the hardware environment composed of the server 101 and the terminal device 103 shown in FIG. Figure 1As shown, the server 101 is connected to the terminal device 103 via a network and can be used to provide services for the terminal device or the application 107 installed on the terminal device. The application can be a video application, instant messaging application, browser application, educational application, game application, etc. A database 105 may be set up on the server or independently of the server to provide data storage services for the server 101, for example, a game data storage server. The above-mentioned network may include but is not limited to: a wired network, a wireless network, wherein the wired network includes: a local area network, a metropolitan area network and a wide area network, and the wireless network includes: Bluetooth, WIFI and other networks that realize wireless communication. The terminal device 103 may be a terminal configured with an application, and may include but is not limited to at least one of the following: a mobile phone (such as an Android phone, an iOS phone, etc.), a laptop computer, a tablet computer, a PDA, a MID (Mobile Internet Devices), a PAD, a desktop computer, a smart TV, an intelligent voice interaction device, a smart home appliance, a vehicle-mounted terminal, an aircraft, a virtual reality (VR) terminal, an augmented reality (AR) terminal, a mixed reality (MR) terminal and other computer devices. The above-mentioned server may be a single server, a server cluster consisting of multiple servers, or a cloud server.

[0027] Combine Figure 1 As shown, the above-mentioned method for processing business data can be executed by an electronic device, which can be a terminal device or a server. The above-mentioned method for processing business data can be implemented separately by the terminal device or the server, or jointly by the terminal device and the server.

[0028] The above is only an example and is not specifically limited in this embodiment.

[0029] Alternatively, as an optional implementation, Figure 2 As shown, the method for processing the above business data includes:

[0030] S202, determining a processing category of the business evaluation data based on data attributes of the business evaluation data, wherein the data attributes of the business evaluation data include at least one of data volume and data source address;

[0031] Optionally, in this embodiment of the present application, the business evaluation data refers to text, audio, or video data related to corporate environmental, social, and governance (ESG) risks collected from online public opinion, and the data attributes refer to quantitative characteristics of the business evaluation data, including but not limited to data volume and data source address. The data volume reflects the scale of the business evaluation data, while the data source address indicates the source of the data, such as from a website.

[0032] It should be noted that in actual applications, the data attributes of business evaluation data may also include but are not limited to the timeliness, language, format and even complexity of the data. The selection and combination of data attributes can be adjusted according to the specific application scenario, and this application does not limit this.

[0033] For example, taking a bank conducting public opinion analysis as an example, the bank wants to evaluate the impact of City A's environmental protection plan on corporate loan business. First, the business evaluation data is uniformly converted into text data, including but not limited to news reports, corporate statements and other content. Then, the text data is preprocessed, including but not limited to removing stop words, correcting spelling errors, performing stem extraction and word segmentation, and normalizing the text, such as unifying the date format, converting digital representation, etc. Further, the data volume of the text data is obtained by counting the number of text characters, words or sentences. When the data volume of the text data is less than the preset data volume threshold, it means that the business evaluation data is a short news headline or a short comment on social media, and the bank determines the data category of the business evaluation data as the first category; or, when the data volume of the text data is greater than or equal to the preset data volume threshold, the data category of the business evaluation data is determined to be the second category.

[0034] For another example, let's take a bank conducting public opinion analysis as an example. The bank wants to evaluate the impact of City A's environmental protection plan on corporate loan business. First, the data source address of the business evaluation data is obtained. For example, information related to the environmental protection plan is captured from various media channels, including but not limited to news websites, social media, industry reports, etc. Then, the obtained business evaluation data is preprocessed to ensure the cleanliness and format consistency of the data. The preprocessing operations include but are not limited to text cleaning (removing irrelevant characters and stop words), word segmentation, part-of-speech tagging, etc. Further, it is determined whether the data source address of the business evaluation data belongs to an address in a preset data source set. In the case that the data source address of the business evaluation data belongs to an address in a preset data source set, it means that the business evaluation data is a short news headline or a short comment on social media. The bank determines the data category of the business evaluation data as the first category; or, in the case that the data volume of the text data is greater than or equal to the preset data volume threshold, the data category of the business evaluation data is determined to be the second category.

[0035] S204: If the processing category of the business evaluation data is the first category, perform a first identification operation on the business evaluation data to obtain a first processing result, wherein the first identification operation is used to determine the sentiment tendency type corresponding to the business evaluation data and the entity name in the business evaluation data, and the first processing result includes at least one of the risk level of the business evaluation data, the sentiment tendency type, the entity name, and the cause analysis, and the first processing result reflects the business risk of the business associated with the business evaluation data;

[0036] Optionally, in an embodiment of the present application, the above-mentioned first category corresponds to business evaluation data belonging to a relatively simple business scenario, the content of the business evaluation data is direct, and the emotional tendencies and entity names are easy to identify. The above-mentioned first identification operation refers to a basic-level natural language processing analysis of the business evaluation data, which may include but is not limited to using a lightweight natural language processing model (the above-mentioned first identification model) to determine the emotional tendency type and identify the entity name; emotional tendency type determination is intended to parse the positive, negative or neutral attitude expressed in the text; entity name recognition is to locate specific objects mentioned in the text, such as the names of companies and individuals.

[0037] It should be noted that the first processing result may include information on multiple dimensions such as risk level, sentiment tendency type, entity name, and cause analysis of the business evaluation data. The specific dimensions to be output can be flexibly determined based on the characteristics and application scenarios of the business evaluation data.

[0038] For example, the risk level indicates the severity of the risk reflected in the data; the sentiment type reflects the public's positive, negative, or neutral attitude towards a specific entity; the entity name is used to clarify the object associated with the risk; and the cause analysis provides detailed background information behind the risk. This application does not limit this, and you can choose the combination of output results based on actual needs to meet the risk assessment requirements of different business scenarios.

[0039] S206. When the processing category of the business evaluation data is the second category, a second identification operation is performed on the business evaluation data to obtain a second processing result, wherein the first identification operation is used to determine the contextual association of the business evaluation data, the emotional tendency type corresponding to the business evaluation data, and the entity name in the business evaluation data. The second processing result includes at least one of the risk level, emotional tendency type, entity name, and cause analysis of the business evaluation data. The second processing result reflects the business risk of the business associated with the business evaluation data.

[0040] Optionally, in an embodiment of the present application, the above-mentioned second category refers to business evaluation data corresponding to complex business scenarios. The above-mentioned second identification operation not only needs to determine the sentiment tendency type and entity name corresponding to the data, but also needs to have an in-depth understanding of the contextual association, for example, what kind of ESG risk is the entity mentioned in the data related to, what is the cause of the risk, and how this information is related to the overall business environment.

[0041] It should be noted that identifying contextual associations may involve a variety of strategies, including but not limited to using deep learning models to perform semantic analysis on long texts, enhancing entity recognition accuracy through external knowledge bases, or employing semantic similarity algorithms to understand the logical relationships between texts. The specific strategy to adopt should be determined based on the characteristics of the business evaluation data.

[0042] For example, for business evaluation data obtained from long industry analysis reports, deep learning models can be used to capture the nuances of complex contexts; for business evaluation data obtained from news reports containing multiple entities, external knowledge bases can be combined to accurately identify and distinguish these entities. This application does not limit this.

[0043] Illustratively, an embodiment of the present application provides a classification and identification method for processing business evaluation data, wherein the processing category of the business evaluation data is determined based on its data attributes, such as data volume and data source address. For data classified as the first category, a first identification operation is used to analyze the sentiment tendency type and identify the entity name, and the output first processing result includes key information such as risk level, sentiment tendency type, entity name and cause analysis, which directly reflects business-related risks. When processing the second category of data, the second identification operation not only confirms the sentiment tendency and entity, but also deeply analyzes the contextual association of the data. The output second processing result also covers risk level, sentiment tendency type, entity name and cause analysis, etc. Because it has a deeper understanding of the context, it is more in-depth and accurate in revealing business risks.

[0044] It should also be noted that the above-mentioned business evaluation data can be obtained periodically. The business evaluation data may include but is not limited to data obtained from multiple data websites. Therefore, the above-mentioned operation can be performed on each piece of the above-mentioned business evaluation data. For example, the above-mentioned business evaluation data consists of data A, data B and data C. After analysis, the data category of the above-mentioned data A is the first category, and the data categories of data B and data C are determined to be the second category.

[0045] Furthermore, a first identification operation can be performed on data A, and a second identification operation can be performed on data B and data C respectively to obtain a first processing result and a second processing result, and the business risk of the corresponding business can be comprehensively determined based on the above first processing result and second processing results.

[0046] In an exemplary embodiment, taking the application scenario of a bank conducting an ESG risk assessment on a company as an example, when the business evaluation data comes from an official announcement of the Environmental Protection Bureau and the data volume is moderate, the system classifies the data into the first category, and then uses a lightweight natural language processing model (the above-mentioned first recognition model) for rapid identification, determining that the sentiment tendency in the announcement is negative and the entity name is Company A. At the same time, it is analyzed that the data risk level is level 7. The reason is that Company A was recently found to be producing without complying with environmental protection regulations, which is directly related to the company's environmental risks. In another scenario, if the data comes from a detailed ESG research report involving a lengthy analysis, the system identifies it as the second category of business evaluation data. In this case, a deep learning model (the second recognition model mentioned above) is used for complex recognition. In addition to determining sentiment and entity names, the contextual associations of the environmental, social, and governance risks mentioned in the report are also deeply analyzed. The generated second processing result not only includes risk levels, sentiment, and entity information, but also provides a detailed causal analysis, revealing that Company B has significant social risks in its supply chain management. Its risk level is rated at 8 because the report points out that Company B's suppliers are involved in serious labor rights violations, which directly affects Company B's social responsibility evaluation.

[0047] Through the embodiment of the present application, the processing category of the business evaluation data is determined based on the data attributes of the business evaluation data, wherein the data attributes of the business evaluation data include at least one of the data volume and the data source address; when the processing category of the business evaluation data is the first category, a first identification operation is performed on the business evaluation data to obtain a first processing result, wherein the first identification operation is used to determine the emotional tendency type corresponding to the business evaluation data and the entity name in the business evaluation data, and the first processing result includes at least one of the risk level, emotional tendency type, entity name and cause analysis of the business evaluation data, and the first processing result reflects the business risk of the business associated with the business evaluation data; when the processing category of the business evaluation data is the second category, a second identification operation is performed on the business evaluation data to obtain a second processing result, wherein, The first identification operation is used to determine the contextual association of the business evaluation data, the sentiment tendency type corresponding to the business evaluation data, and the entity name in the business evaluation data. The second processing result includes at least one of the risk level, sentiment tendency type, entity name and cause analysis of the business evaluation data. The second processing result reflects the business risk of the business associated with the business evaluation data. By intelligently allocating data of different complexities to the corresponding optimized first identification model to perform the first identification operation and the second identification model to perform the second identification operation, the purpose of efficiently and accurately identifying and evaluating business risks is achieved, thereby achieving the technical effect of significantly improving the accuracy of risk prediction while maintaining a high processing speed, and thus solving the technical problem of low risk prediction accuracy in traditional business data risk prediction methods due to a single model and low processing efficiency.

[0048] As an optional solution, when the processing category of the above-mentioned business evaluation data is the first category, the above-mentioned first recognition operation is performed on the above-mentioned business evaluation data to obtain a first processing result, including: inputting the above-mentioned business evaluation data into a first recognition model, wherein the above-mentioned first recognition model includes a keyword matching module and a sentiment analysis module; using the above-mentioned keyword matching module to determine the semantic similarity between the above-mentioned business evaluation data and a preset target keyword library, and determining the above-mentioned entity name based on the above-mentioned semantic similarity; using the above-mentioned sentiment analysis module to determine the sentiment tendency type of the above-mentioned business evaluation data; and generating the above-mentioned first processing result based on the sentiment tendency type corresponding to the above-mentioned business evaluation data and the entity name in the above-mentioned business evaluation data.

[0049] Optionally, in an embodiment of the present application, the first recognition model refers to a natural language processing model that integrates keyword matching and sentiment analysis functions, including but not limited to a rule-based model, a lightweight deep learning model, or a hybrid model. The keyword matching module is used to quickly find entities closely related to business evaluation data, and the sentiment analysis module is used to determine the emotional tone contained in the data. The two work together to efficiently extract key business risk information from massive amounts of data.

[0050] It should be noted that when processing business evaluation data, the first recognition model can be fine-tuned according to the specific content and source of the data. For example, the first recognition model can customize the target keyword library for a specific industry or entity to improve the accuracy of entity recognition; the sentiment analysis module can also integrate multiple sentiment vocabularies to ensure comprehensive coverage of sentiment tendency types.

[0051] For example, the first recognition model can be a lightweight rule-based text processing engine that uses keyword matching, sentiment dictionaries, and basic natural language processing techniques to quickly evaluate the sentiment of text, identify entities, and event labels. It is suitable for processing simple or formatted text data, such as short comments on social media or news headlines. It can filter and classify large amounts of information in a short period of time, reducing the pressure of subsequent processing.

[0052] Exemplarily, the technical solution of the embodiment of the present application involves inputting business evaluation data into a first recognition model, calculating the semantic similarity between the data and a preset target keyword library through a keyword matching module to determine the entity name; at the same time, using a sentiment analysis module to analyze the sentiment tendency type of the data, and then combining the information of the two to generate a first processing result.

[0053] In an exemplary embodiment, taking the application scenario of a bank conducting a risk assessment on the ESG performance of an enterprise as an example, a news report mentioning a pollution incident of an enterprise is classified as the first category of data, and then the news text is input into the first recognition model. The keyword matching module immediately identifies the keywords related to environmental risks in the text, such as "pollutant emissions exceeding the standard", and determines that the corporate entity involved in the incident is "ABC Company". The sentiment analysis module analyzes that the sentiment tendency of the text is negative. Combining these two parts of the analysis, the first processing result is generated, indicating that "ABC Company" has environmental risks, the sentiment tendency is negative, and the risk level is level 6, because of the pollution incident mentioned in the report. This process reflects the system's rapid response and accurate evaluation of business evaluation data in simple scenarios.

[0054] Through the embodiments of the present application, keyword matching and sentiment tendency analysis based on the first recognition model are adopted to achieve effective identification of entity names and sentiment tendency types in business evaluation data, thereby achieving the purpose of improving the efficiency and accuracy of ESG risk warnings for financial institutions.

[0055] As an optional solution, the above-mentioned first processing result is generated based on the sentiment tendency type corresponding to the above-mentioned business evaluation data and the entity name in the above-mentioned business evaluation data, including: when the sentiment tendency type corresponding to the above-mentioned business evaluation data is a positive type, determining that the risk level of the above-mentioned business evaluation data is the first level, and generating a positive cause analysis associated with the entity name in the above-mentioned business evaluation data, wherein the above-mentioned cause analysis includes the above-mentioned positive cause analysis; when the sentiment tendency type corresponding to the above-mentioned business evaluation data is a neutral type, determining that the risk level of the above-mentioned business evaluation data is the second level, and generating a neutral cause analysis associated with the entity name in the above-mentioned business evaluation data, wherein the above-mentioned cause analysis includes the above-mentioned neutral cause analysis, and the business risk of the second level is higher than the business risk of the first level; when the sentiment tendency type corresponding to the above-mentioned business evaluation data is a negative type, determining that the risk level of the above-mentioned business evaluation data is the third level, and generating a negative cause analysis associated with the entity name, wherein the above-mentioned cause analysis includes the above-mentioned negative cause analysis, and the business risk of the third level is higher than the business risk of the second level.

[0056] Optionally, in an embodiment of the present application, the above-mentioned emotional tendency types of positive type, neutral type and negative type refer to the emotional colors expressed in the data tending to be positive, neutral or negative, respectively. The entity name refers to the specific enterprise or organization name mentioned in the data. The first processing result is a collection of a series of evaluation indicators and analysis results, including but not limited to risk level, emotional tendency type, entity name and cause analysis. The first level, second level and third level represent three different levels of risk, low, medium and high, respectively, among which the business risk of the third level is the highest and the business risk of the first level is the lowest.

[0057] It should be noted that the determination of the sentiment tendency type of business evaluation data is affected by many factors, including but not limited to the accuracy of the natural language processing technology used, the context of the data, and the ambiguity of words in the text.

[0058] In addition, the risk level setting can be adjusted according to different risk management systems. For example, the first level can be set as no risk, the second level can be set as potential risk, and the third level can be set as high risk. This application does not limit this. What is important is to divide the risk levels according to the emotional tendency type to generate a targeted first processing result.

[0059] Illustratively, in an embodiment of the present application, a first processing result is generated based on the sentiment tendency type and entity name of the business evaluation data, specifically including: when the sentiment tendency type is positive, the business risk associated with the evaluation data is low, the risk level is set to the first level, and a positive cause analysis matching the entity name is generated, such as the company's good ESG performance is recognized or positively reported. When the sentiment tendency type is neutral, the business risk is assessed to be medium, the risk level is set to the second level, and a neutral cause analysis is generated, which may involve a general description of the company's ESG performance or a public discussion of neutral views. When the sentiment tendency type is negative, the business risk is assessed to be high, the risk level is set to the third level, and a negative cause analysis is generated, such as a specific description of the ESG risks involved in the company being accused of misconduct or negative reports.

[0060] In one exemplary embodiment, a bank conducts public opinion analysis on corporate ESG risks. For example, consider a piece of business evaluation data, assuming it's a news report about Company A's significant progress in environmental governance. The sentiment analysis module identifies the text as positive, and the keyword matching module confirms that Company A is an entity name. Therefore, the system automatically assigns a risk level of Level 1 and generates a positive cause analysis related to Company A, highlighting its proactive environmental governance measures and thus reflecting its low ESG risk.

[0061] Through the embodiment of the present application, the first processing result is generated by combining the sentiment tendency type with the entity name, which achieves a deep understanding and accurate assessment of the business evaluation data and achieves the purpose of effectively identifying and quantifying ESG risks.

[0062] As an optional solution, when the processing category of the above-mentioned business evaluation data is the second category, a second recognition operation is performed on the above-mentioned business evaluation data to obtain a second processing result, including: inputting the above-mentioned business evaluation data into a second recognition model, wherein the above-mentioned second recognition model includes an event label recognition module, a sentiment tendency recognition module, and an entity recognition module; using the above-mentioned event label recognition module to process the above-mentioned business evaluation data to obtain the contextual association of the above-mentioned business evaluation data; using the above-mentioned entity recognition module to process the above-mentioned business evaluation data and the contextual association of the above-mentioned business evaluation data to obtain the sentiment tendency type corresponding to the above-mentioned business evaluation data; using the above-mentioned sentiment tendency recognition module to process the above-mentioned business evaluation data and the contextual association of the above-mentioned business evaluation data to obtain the entity name in the above-mentioned business evaluation data; generating the above-mentioned second processing result according to the sentiment tendency type corresponding to the above-mentioned business evaluation data and the entity name in the above-mentioned business evaluation data.

[0063] Optionally, in an embodiment of the present application, the second recognition model refers to an advanced natural language processing architecture model designed to process complex and information-intensive business evaluation data, including but not limited to long analytical reports, in-depth interview records, or cross-disciplinary review articles. The event tag recognition module is used to identify and classify key events or topics in the business evaluation data, the sentiment tendency recognition module is used to deeply analyze the sentiment tendencies expressed in the data, and the entity recognition module focuses on locating and identifying specific entities in the data, such as enterprises, organizations, or individuals.

[0064] It should be noted that, when processing the second category of business evaluation data, each component of the second recognition model can access additional domain knowledge bases or corpora to enhance its analysis capabilities.

[0065] For example, the event label recognition module relies on an ESG risk label library containing industry-specific terms and issues, while the sentiment tendency recognition module requires a special sentiment vocabulary for the financial industry.

[0066] In addition, the entity recognition module can be combined with an external database or knowledge graph to more accurately identify and distinguish entities mentioned in the text, which is not limited in this application.

[0067] It is understandable that the first and second identification models each perform efficient and accurate risk identification and analysis for business evaluation data with different characteristics. The first identification model is primarily applicable to situations where data attributes are relatively simple and intuitive, such as short text or evaluation information with clear emotional overtones. Through rapid processing and preliminary classification, it can quickly screen out key risk indicators and emotional tendencies, laying the foundation for subsequent analysis. The second identification model focuses on processing complex, information-intensive business evaluation data, such as long articles and cross-disciplinary comprehensive assessment reports. Through advanced functions such as deep semantic understanding, event label recognition, and entity sentiment analysis, it can meticulously analyze ESG risk details in the data, providing deeper risk insights and cause analysis.

[0068] For example, the second recognition model can be a complex neural network model based on deep learning, such as a large-scale pre-trained model using the Transformer architecture. This type of model possesses strong semantic understanding and contextual reasoning capabilities, capable of processing long texts, complex contexts, and cross-domain information. The rich language model obtained through training can more accurately identify subtle emotional nuances, complex entity relationships, and potential ESG risk clues in the text. This model is suitable for processing detailed corporate reports, industry analysis articles, or in-depth interview transcripts, and can provide in-depth analysis results and refined risk assessments.

[0069] When using the second recognition model, a variety of prompt engineering technologies are integrated to improve the generation quality of the second recognition model (such as GPT-4), and the task is broken down into three sub-tasks: event label recognition, sentiment tendency recognition, and company entity recognition. This provides intermediate reasoning steps for the second recognition model, enhancing the thinking logic and result accuracy of the second recognition model.

[0070] In addition, it is possible to build an ESG public opinion analysis knowledge base and design a continuous update and expansion mechanism. New samples can be regularly extracted from user feedback, and the ESG risk tag library can be dynamically updated to maintain sensitivity to emerging ESG issues and enhance the professionalism of the second recognition model in ESG risk analysis. Multi-link recall technology (such as semantic similarity + BM25 dual-link recall) can be used to recall text in the knowledge base that is highly similar to the analysis content. Rearrangement algorithms can be used to reorder the recalled texts, and parameters can be used to specify the amount of reference knowledge provided to the second recognition model, helping the second recognition model to provide more accurate responses related to the context and reduce the phenomenon of generative model hallucination.

[0071] For example, business evaluation data is fed into the second recognition model. It is first processed by the event tag recognition module, extracting contextual associations from the data, such as identifying specific ESG events or issues. Next, the entity recognition module accurately locates entity names within the data based on event tags and context. Simultaneously, the sentiment recognition module deeply analyzes the sentiment of the text, combining event tags with entity information to generate sentiment types. Finally, the sentiment types, entity names, and event tags are integrated to generate the second processing result, providing financial institutions with in-depth ESG risk insights.

[0072] In an exemplary embodiment, taking the in-depth report analysis of corporate ESG compliance by financial institutions as an example, the second recognition model is used to process a detailed ESG audit report. The report involves complex environmental, social and governance issues, as well as multi-faceted analysis. The event label recognition module first identifies key ESG events in the report, such as "supply chain labor rights dispute" and "environmental pollution compensation agreement"; then, the sentiment tendency recognition module analyzes the wording and tone of the report, and determines that the report as a whole expresses a neutral to slightly negative sentiment towards the company; then, the entity recognition module accurately identifies the company name and related parties, such as suppliers or regulators, from the report; integrating this information, the second processing result reflects that the company faces medium to high ESG risks, specifically potential problems in supply chain management and social responsibility, as well as some environmental risks, with a slightly negative sentiment tendency, and the entity name is the company XYZ Group. The cause analysis includes a detailed description of the above event labels and sentiment tendencies.

[0073] Through the embodiments of the present application, the event label recognition, sentiment tendency recognition and entity recognition of the second recognition model are adopted to achieve comprehensive analysis and accurate evaluation of complex business evaluation data, thereby achieving the goal of improving the ESG risk prevention and control capabilities of financial institutions.

[0074] As an optional solution, the above-mentioned second processing result is generated based on the contextual association of the above-mentioned business evaluation data, the emotional tendency type corresponding to the above-mentioned business evaluation data and the entity name in the above-mentioned business evaluation data, including: when the emotional tendency type corresponding to the above-mentioned business evaluation data is a positive type, determining that the risk level of the above-mentioned business evaluation data is the fourth level, and generating a positive cause analysis associated with the entity name in the above-mentioned business evaluation data, wherein the above-mentioned cause analysis includes the above-mentioned positive cause analysis; when the emotional tendency type corresponding to the above-mentioned business evaluation data is a neutral type, determining that the risk level of the above-mentioned business evaluation data is the fifth level, and generating a neutral cause analysis associated with the entity name in the above-mentioned business evaluation data, wherein the above-mentioned cause analysis includes the above-mentioned neutral cause analysis, and the business risk of the fifth level is higher than the business risk of the fourth level; when the emotional tendency type corresponding to the above-mentioned business evaluation data is a negative type, determining that the risk level of the above-mentioned business evaluation data is the sixth level, and generating a negative cause analysis associated with the entity name, wherein the above-mentioned cause analysis includes the above-mentioned negative cause analysis, and the business risk of the sixth level is higher than the business risk of the fifth level.

[0075] Optionally, in an embodiment of the present application, the contextual association of business evaluation data refers to the background, motivation and scope of influence of events in the text, including but not limited to descriptions of corporate behavior, plan changes or market trends. The sentiment tendency type covers positive, neutral and negative emotions expressed in the data. The positive type indicates that the data tends to be positively evaluated, the neutral type indicates that the data is neutral or has no obvious bias, and the negative type means that there are obvious negative evaluations in the data. The entity name refers to the name of the specific enterprise, person or organization mentioned in the business evaluation data.

[0076] It should be noted that the analysis process of business evaluation data includes but is not limited to part-of-speech tagging, named entity recognition, sentiment analysis and event labeling, so as to ensure accurate identification of context associations, sentiment tendency types and entity names.

[0077] In addition, the risk level is set according to the enterprise ESG risk assessment standard, with levels 4, 5, and 6 representing low risk, medium risk, and high risk, respectively. Level 6 has the highest business risk, and level 4 has the lowest business risk. This application does not limit this.

[0078] For example, if the sentiment type is positive, the system determines that the business risk associated with the business evaluation data is low, sets the risk level to level four, and generates a positive cause analysis that matches the entity name, such as the company's active efforts to improve environmental performance or fulfill its social responsibilities. If the sentiment type is neutral, the system assesses the business risk as medium, sets the risk level to level five, and generates a neutral cause analysis, which may include a general description of the company's ESG performance. If the sentiment type is negative, the system determines that the business risk is high, sets the risk level to level six, and generates a negative cause analysis, such as the company's involvement in major environmental violations or social issues.

[0079] In an exemplary embodiment, taking the application scenario of a bank conducting public opinion monitoring on corporate ESG risks as an example, the bank obtained an in-depth media report on Company A's sustainable development. The report describes in detail the cooperative relationship between Company A and its suppliers, involving labor rights disputes (contextual association situation), the overall tone is neutral to negative (emotional tendency type), and Company A (entity name) is explicitly mentioned. According to the embodiment of the present application, the system first identifies the contextual association situation and the emotional tendency type, and then combines entity recognition to determine that the emotional tendency type is a negative type, and then determines that the risk level is the sixth level. Next, a negative cause analysis related to Company A is generated to clarify the company's potential risks in supply chain management and social responsibility, and the negative impact these risks may have on ESG performance.

[0080] Through the embodiments of the present application, a comprehensive analysis method based on the contextual association of business evaluation data, sentiment tendency type and entity name is adopted to achieve accurate quantification of ESG risks in complex business evaluation data, thereby achieving the goal of improving the ESG risk management level of financial institutions.

[0081] As an optional solution, the above-mentioned determination of the processing category of the above-mentioned business evaluation data based on the data attributes of the business evaluation data includes: when the data volume of the above-mentioned business evaluation data is less than the preset data volume threshold, determining that the processing category of the above-mentioned business evaluation data is the above-mentioned first category; when the data volume of the above-mentioned business evaluation data is greater than or equal to the above-mentioned data volume threshold, determining that the processing category of the above-mentioned business evaluation data is the above-mentioned second category.

[0082] Optionally, in the embodiments of the present application, the data volume threshold is a pre-set value used to distinguish the size of the data to determine which processing model to use for analysis. The first category and the second category represent two different processing methods or models. The first category generally corresponds to processing simpler or smaller business evaluation data, while the second category corresponds to processing more complex or larger business evaluation data.

[0083] Exemplarily, the processing category of the business evaluation data is determined based on its data attributes, primarily its volume. When the volume of the business evaluation data is less than a preset data volume threshold, the system selects the first processing method, which uses a small model with fast response and low computational resource consumption for analysis. Conversely, if the data volume is greater than or equal to the data volume threshold, the system selects the second processing method, activating a large model for in-depth analysis to handle more complex scenarios and data structures.

[0084] It should be noted that the setting of the data volume threshold can be flexibly adjusted according to factors such as the application scenario, computing resource limitations, and analysis efficiency requirements.

[0085] For example, for an environment with abundant computing resources, the threshold can be appropriately increased to expand the scope of use of large models; while for an environment with limited computing resources, the threshold should be appropriately lowered and more reliance should be placed on small model processing.

[0086] In one exemplary embodiment, a commercial bank evaluates corporate financing projects. The bank receives a short news report about a company receiving subsidies for a green energy project. The system first detects that the word count in the report falls below a preset data volume threshold and automatically categorizes it as category one, triggering processing by the first recognition model. The first recognition model quickly analyzes the report's mention of Company A, identifies its receipt of green energy subsidies, and identifies positive sentiment and a low risk rating. The reasoning includes the public's support for Company A's environmentally friendly initiatives.

[0087] As an optional solution, the above-mentioned determination of the processing category of the above-mentioned business evaluation data based on the data attributes of the business evaluation data includes: when the data source address of the above-mentioned business evaluation data belongs to the preset target data source address, determining that the processing category of the above-mentioned business evaluation data is the above-mentioned first category; when the data source address of the above-mentioned business evaluation data does not belong to the preset target data source address, determining that the processing category of the above-mentioned business evaluation data is the above-mentioned second category.

[0088] Optionally, in this embodiment of the present application, the data source address refers to the original publication location or platform of the business evaluation data, including but not limited to reliable sources such as official websites, certified industry journals, well-known financial media, as well as informal or less reliable sources such as social media, blogs, and forums. The preset target data source address set can be determined, including but not limited to, by the following methods:

[0089] 1. By weighting the historical records, professional level and industry recognition of the data sources, we screen out data source addresses that are considered to be highly authoritative, such as official websites, well-known financial media, and reports from leading industry analysis agencies.

[0090] 2. Conduct data quality checks on data sources regularly or irregularly, including the accuracy, completeness, timeliness, etc. of the information, and include data sources that pass the quality standards in the target data source address set.

[0091] 3. Establish a user feedback system to collect users’ evaluations and suggestions on various data sources, and determine the target data source address set based on user satisfaction and the practicality of the data source.

[0092] 4. Use machine learning technology to analyze the historical performance of data sources and user interaction data, and recommend data source addresses with excellent performance and user preference to update the target data source address set.

[0093] It should be noted that even if the data comes from a preset target data source, if the data content is highly irrelevant to the current business environment, the system can also classify it as the second category for deeper analysis. This application does not limit this.

[0094] For example, after receiving a piece of business evaluation data, the system first checks whether its data source address belongs to a preset set of target data source addresses. If the data originates from a known and authoritative financial news website, the system automatically marks it as Category 1, indicating high credibility and suitable for rapid identification and processing. Conversely, if the data originates from social media or personal blogs, such sources may contain more subjective or unverified information. The system will mark it as Category 2 and prepare to conduct a more complex and detailed analysis process to ensure the reliability of the evaluation results.

[0095] In one exemplary embodiment, a bank conducts an ESG risk assessment of a potential partner. The bank collects business evaluation data on Company A from multiple sources, including certified industry reports (pre-set target data source addresses) and social media comments (non-pre-set target data source addresses). The bank's analysis system first checks the data source addresses and finds that the industry report data source belongs to the pre-set target data source address. Therefore, this data is classified as the first category and quickly performs keyword matching and sentiment analysis. The social media comments are then labeled as second-category data and a more sophisticated text analysis model is then used, taking into account context and potential semantic associations, to ensure a deep understanding of public opinion and assess its complex impact on Company A's ESG risks.

[0096] Through the embodiments of the present application, a technical solution of dynamically adjusting processing categories based on data volume is adopted to achieve efficient and intelligent analysis of business evaluation data, thereby achieving the purpose of optimizing the risk assessment process, that is, saving computing resources and time costs to the greatest extent while ensuring the quality of analysis and the accuracy of risk identification.

[0097] In an exemplary embodiment, considering that existing technical solutions often fall into the dilemma of insufficient computing resources or low analysis accuracy when facing hundreds of millions of public opinion data, and cannot achieve the accuracy of the results while ensuring the efficiency of public opinion analysis, the embodiment of the present application can decompose it into a series of smaller and easier to manage subtasks, thereby restricting the quality and interpretability of the analysis results. Through an innovative fusion model architecture, combined with the fast processing capability of the small model (the above-mentioned first recognition model) and the precise labeling capability of the large model (the above-mentioned second recognition model), it can effectively identify the ESG risk labels, emotional tendencies, and corporate entities involved in public opinion and give specific reasons. The specific process is as follows: Figure 3 As shown, including but not limited to:

[0098] S1, ESG risk label library construction, based on ESG core issues, builds a standardized ESG risk label system, covering risk categories in dimensions such as environmental impact, social responsibility fulfillment, and corporate governance effectiveness involved in the company's operations, and accurately defines the connotation and boundaries of each risk category.

[0099] S2, data collection and preprocessing, collects public opinion data from multiple channels and sources such as websites and other media (the above-mentioned business evaluation data), performs unified format conversion and field standardization, and removes irrelevant characters, stop words and noise data.

[0100] S3: Multi-tiered models collaborate to analyze ESG risk public opinion. Based on the complexity of the public opinion content, this system combines the efficient processing capabilities of small models with the precise annotation paradigm of large models to form a hierarchical processing mechanism. Small models prioritize inference speed and memory usage, providing rapid screening and preliminary classification, suitable for processing simple public opinion. Large models, on the other hand, enhance semantic understanding and contextual reasoning capabilities, enabling in-depth analysis of complex public opinion, combining knowledge bases and reasoning capabilities to generate high-quality results.

[0101] S3-1, Lightweight Natural Language Processing Model (the first identification model mentioned above): This model uses a lightweight natural language processing model to quickly screen and classify simple public opinion data (i.e., the business evaluation data in the first category mentioned above). Based on keyword matching and semantic similarity, the data is divided into several ESG-related thematic categories (such as emissions management, production safety, and business ethics) and the relevant corporate entities are identified. A pre-trained sentiment analysis module is used to classify the text as positive, negative, or neutral.

[0102] S3-2, the deep semantic understanding model (the second recognition model mentioned above), integrates multiple prompt engineering technologies to improve the generation quality of large language models (such as GPT-4) for complex public opinions that have passed the initial screening (that is, the second category of business evaluation data mentioned above).

[0103] When using the second recognition model, multiple prompt engineering technologies are integrated to improve the generation quality of the second recognition model (such as GPT-4), and the task is broken down into three sub-tasks: event label recognition, sentiment tendency recognition, and company entity recognition. This provides intermediate reasoning steps for the second recognition model, enhances the thinking logic and result accuracy of the second recognition model, and can also generate more professional answers by simulating experts in the field of ESG public opinion analysis using the deep semantic understanding model.

[0104] In addition, it is possible to build an ESG public opinion analysis knowledge base and design a continuous update and expansion mechanism. New samples can be regularly extracted from user feedback, and the ESG risk tag library can be dynamically updated to maintain sensitivity to emerging ESG issues and enhance the professionalism of the second recognition model in ESG risk analysis. Multi-link recall technology (such as semantic similarity + BM25 dual-link recall) can be used to recall text in the knowledge base that is highly similar to the analysis content. Rearrangement algorithms can be used to reorder the recalled texts, and parameters can be used to specify the amount of reference knowledge provided to the second recognition model, helping the second recognition model to provide more accurate responses related to the context and reduce the phenomenon of generative model hallucination.

[0105] S4. The core output fields of ESG risk public opinion analysis results include but are not limited to: Figure 4 As shown:

[0106] Risk labels: include ESG classification labels (such as "Environment-Emission Management-Excessive Pollution Emissions"), risk levels (1-10, the larger the number, the higher the risk); sentiment tendency: positive, negative, neutral; company entity: the full name of the company or organization involved; cause analysis: gives the reasons for the label analysis results. For example, if the public opinion mentions that the company was notified by the regulator for exceeding pollution emission standards, it directly points to the problem of excessive pollution emission in emission management.

[0107] Through the embodiment of the present application, a multi-level model collaboration mechanism of "small model combined with large model" is constructed to carry out ESG risk public opinion analysis, taking into account both analysis efficiency and accuracy. A professional ESG public opinion knowledge base is also constructed, and a multi-dimensional prompting strategy (including expert role-playing, thought chain reasoning, and retrieval enhancement generation) is introduced to assist the large model in intelligently judging emotional tendencies, extracting risk entities, and forming ESG public opinion risk indicators. By scientifically allocating computing resources, task stratification is achieved. The first recognition model is responsible for fast processing and basic feature extraction, and the second recognition model focuses on deep semantic understanding and complex reasoning, so as to reasonably allocate computing resources and maximize the effectiveness of public opinion analysis. By creating a professional ESG public opinion analysis knowledge base and designing a continuous update and expansion mechanism, the professionalism of the large model for ESG risk analysis is enhanced. A multi-dimensional prompting strategy (including expert role-playing, thought chain reasoning, and retrieval enhancement generation) is introduced to assist the large model in intelligently judging emotional tendencies, extracting risk entities, and forming ESG public opinion risk indicators. Through directional constraints and knowledge injection, the hallucination phenomenon of the generated model is effectively reduced, and the reliability and accuracy of the analysis results are improved.

[0108] It is understandable that in the specific implementation of this application, related data such as user information is involved. When the above embodiments of this application are applied to specific products or technologies, user permission or consent is required, and the collection, use and processing of relevant data must comply with relevant laws, regulations and standards of relevant countries and regions.

[0109] It should be noted that for the aforementioned method embodiments, for the sake of simplicity, they are all expressed as a series of action combinations, but those skilled in the art should be aware that this application is not limited by the order of the actions described, because according to this application, certain steps can be performed in other orders or simultaneously. Secondly, those skilled in the art should also be aware that the embodiments described in the specification are all preferred embodiments, and the actions and modules involved are not necessarily required by this application.

[0110] According to another aspect of the embodiment of the present application, a business data processing device for implementing the above-mentioned business data processing method is also provided. Figure 5 As shown, the device includes:

[0111] S502, a determination module, configured to determine a processing category of the business evaluation data based on data attributes of the business evaluation data, wherein the data attributes of the business evaluation data include at least one of data volume and data source address;

[0112] S504, a first identification module, configured to, when the processing category of the business evaluation data is the first category, perform a first identification operation on the business evaluation data to obtain a first processing result, wherein the first identification operation is used to determine the sentiment tendency type corresponding to the business evaluation data and the entity name in the business evaluation data, the first processing result including at least one of the risk level, sentiment tendency type, entity name, and cause analysis of the business evaluation data, and the first processing result reflects the business risk of the business associated with the business evaluation data;

[0113] S506, a second identification module, is used to perform a second identification operation on the business evaluation data when the processing category of the business evaluation data is the second category, to obtain a second processing result, wherein the first identification operation is used to determine the contextual association of the business evaluation data, the emotional tendency type corresponding to the business evaluation data, and the entity name in the business evaluation data, and the second processing result includes at least one of the risk level, emotional tendency type, entity name and cause analysis of the business evaluation data, and the second processing result reflects the business risk of the business associated with the business evaluation data.

[0114] As an optional solution, the above-mentioned device is used to perform a first recognition operation on the business evaluation data to obtain a first processing result in the following manner when the processing category of the business evaluation data is the first category: the business evaluation data is input into a first recognition model, wherein the first recognition model includes a keyword matching module and a sentiment analysis module; the keyword matching module is used to determine the semantic similarity between the business evaluation data and a preset target keyword library, and the entity name is determined based on the semantic similarity; the sentiment analysis module is used to determine the sentiment tendency type of the business evaluation data; and the first processing result is generated based on the sentiment tendency type corresponding to the business evaluation data and the entity name in the business evaluation data.

[0115] As an optional solution, the above-mentioned device is used to generate a first processing result according to the emotional tendency type corresponding to the business evaluation data and the entity name in the business evaluation data in the following manner: when the emotional tendency type corresponding to the business evaluation data is a positive type, the risk level of the business evaluation data is determined to be the first level, and a positive cause analysis associated with the entity name in the business evaluation data is generated, wherein the cause analysis includes a positive cause analysis; when the emotional tendency type corresponding to the business evaluation data is a neutral type, the risk level of the business evaluation data is determined to be the second level, and a neutral cause analysis associated with the entity name in the business evaluation data is generated, wherein the cause analysis includes a neutral cause analysis, and the business risk of the second level is higher than the business risk of the first level; when the emotional tendency type corresponding to the business evaluation data is a negative type, the risk level of the business evaluation data is determined to be the third level, and a negative cause analysis associated with the entity name is generated, wherein the cause analysis includes a negative cause analysis, and the business risk of the third level is higher than the business risk of the second level.

[0116] As an optional solution, the above-mentioned device is used to perform a second recognition operation on the business evaluation data to obtain a second processing result when the processing category of the business evaluation data is the second category in the following manner: input the business evaluation data into a second recognition model, wherein the second recognition model includes an event label recognition module, an emotional tendency recognition module, and an entity recognition module; use the event label recognition module to process the business evaluation data to obtain the contextual association of the business evaluation data; use the entity recognition module to process the business evaluation data and the contextual association of the business evaluation data to obtain the emotional tendency type corresponding to the business evaluation data; use the emotional tendency recognition module to process the business evaluation data and the contextual association of the business evaluation data to obtain the entity name in the business evaluation data; generate the second processing result according to the emotional tendency type corresponding to the business evaluation data and the entity name in the business evaluation data.

[0117] As an optional solution, the above-mentioned device is used to generate a second processing result based on the contextual association of the business evaluation data, the emotional tendency type corresponding to the business evaluation data and the entity name in the business evaluation data in the following manner: when the emotional tendency type corresponding to the business evaluation data is a positive type, the risk level of the business evaluation data is determined to be the fourth level, and a positive cause analysis associated with the entity name in the business evaluation data is generated, wherein the cause analysis includes a positive cause analysis; when the emotional tendency type corresponding to the business evaluation data is a neutral type, the risk level of the business evaluation data is determined to be the fifth level, and a neutral cause analysis associated with the entity name in the business evaluation data is generated, wherein the cause analysis includes a neutral cause analysis, and the business risk of the fifth level is higher than the business risk of the fourth level; when the emotional tendency type corresponding to the business evaluation data is a negative type, the risk level of the business evaluation data is determined to be the sixth level, and a negative cause analysis associated with the entity name is generated, wherein the cause analysis includes a negative cause analysis, and the business risk of the sixth level is higher than the business risk of the fifth level.

[0118] As an optional solution, the above-mentioned device is used to determine the processing category of the business evaluation data based on the data attributes of the business evaluation data in the following manner: when the data volume of the business evaluation data is less than a preset data volume threshold, the processing category of the business evaluation data is determined to be the first category; when the data volume of the business evaluation data is greater than or equal to the data volume threshold, the processing category of the business evaluation data is determined to be the second category.

[0119] As an optional solution, the above-mentioned device is used to determine the processing category of the business evaluation data based on the data attributes of the business evaluation data in the following manner: when the data source address of the business evaluation data belongs to the preset target data source address, the processing category of the business evaluation data is determined to be the first category; when the data source address of the business evaluation data does not belong to the preset target data source address, the processing category of the business evaluation data is determined to be the second category.

[0120] In the embodiments of the present application, the term "module" or "unit" refers to a computer program or a part of a computer program that has a predetermined function and works together with other related parts to achieve a predetermined goal, and can be implemented in whole or in part by using software, hardware (such as processing circuits or memories) or a combination thereof. Similarly, a processor (or multiple processors or memories) can be used to implement one or more modules or units. In addition, each module or unit can be part of an overall module or unit that includes the function of the module or unit.

[0121] Regarding the apparatus in the above embodiment, the specific manner in which each module performs operations has been described in detail in the embodiment of the method, and will not be elaborated here.

[0122] According to one aspect of the present application, a computer program product is provided, which includes a computer program.

[0123] The serial numbers of the above-mentioned embodiments of the present application are for description only and do not represent the advantages or disadvantages of the embodiments.

[0124] In particular, according to an embodiment of the present application, the processes described in the various method flow charts can be implemented as computer software programs. For example, an embodiment of the present application includes a computer program product comprising a computer program carried on a computer-readable medium, the computer program containing program code for executing the methods shown in the flow charts. In such an embodiment, the computer program can be downloaded and installed from a network via a communication portion, and / or installed from a removable medium. When the computer program is executed by a central processing unit, the various functions defined in the system of the present application are performed.

[0125] In such an embodiment, the computer program can be downloaded and installed from a network via the communication portion, and / or installed from a removable medium. When the computer program is executed by the central processing unit, various functions provided by the embodiments of the present application are performed.

[0126] In other embodiments, the terminal device or server may be a node in a distributed system, wherein the distributed system may be a blockchain system, and the blockchain system may be a distributed system formed by connecting multiple nodes via network communication. The nodes may form a peer-to-peer network, and any computing device, such as a server, terminal, or other electronic device, may become a node in the blockchain system by joining the peer-to-peer network.

[0127] According to one aspect of the present application, a computer-readable storage medium is provided, and a processor of an electronic device reads the computer instructions from the computer-readable storage medium, and the processor executes the computer instructions, so that the electronic device executes the business data processing method provided in various optional implementation methods of the above-mentioned business data processing.

[0128] Optionally, in this embodiment, the above-mentioned computer-readable storage medium can be configured to store data for executing the methods in various embodiments of the present application.

[0129] Optionally, in this embodiment, a person of ordinary skill in the art may understand that all or part of the steps in the various methods of the above embodiments may be completed by instructing the hardware related to the terminal device through a program, and the program may be stored in a computer-readable storage medium, which may include: a flash drive, a read-only memory (ROM), a random access memory (RAM), a magnetic disk or an optical disk, etc.

[0130] The serial numbers of the above-mentioned embodiments of the present application are for description only and do not represent the advantages or disadvantages of the embodiments.

[0131] If the integrated units in the above embodiments are implemented in the form of software functional units and sold or used as independent products, they can be stored in the above-mentioned computer-readable storage medium. Based on this understanding, the technical solution of the present application, or the part that contributes to the prior art, or all or part of the technical solution can be embodied in the form of a software product. The computer software product is stored in a storage medium and includes a number of instructions for causing one or more electronic devices to execute all or part of the steps of the method described in each embodiment of the present application.

[0132] In the above embodiments of the present application, the description of each embodiment has its own focus. For parts that are not described in detail in a certain embodiment, please refer to the relevant description of other embodiments.

[0133] In the several embodiments provided in this application, it should be understood that the disclosed applications can be implemented in other ways. Among them, the device embodiments described above are only schematic. For example, the division of the units is only a logical function division. In actual implementation, there may be other division methods, such as multiple units or components can be combined or integrated into another system, or some features can be ignored or not executed. Another point is that the mutual coupling or direct coupling or communication connection shown or discussed can be through some interfaces, indirect coupling or communication connection of units or modules, which can be electrical or other forms.

[0134] The units described as separate components may or may not be physically separate, and the components shown as units may or may not be physical units, that is, they may be located in one place or distributed across multiple network units. Some or all of these units may be selected to achieve the purpose of this embodiment according to actual needs.

[0135] In addition, the functional units in the various embodiments of the present application may be integrated into a single processing unit, or each unit may exist physically separately, or two or more units may be integrated into a single unit. The aforementioned integrated units may be implemented in the form of hardware or software functional units.

[0136] The above is only a preferred embodiment of the present application. It should be pointed out that for ordinary technicians in this technical field, several improvements and modifications can be made without departing from the principles of the present application. These improvements and modifications should also be regarded as the scope of protection of the present application.

Claims

1. A method for processing business data, characterized in that: include: Determining a processing category of the business evaluation data based on data attributes of the business evaluation data, wherein the data attributes of the business evaluation data include at least one of data volume and data source address; When the processing category of the business evaluation data is the first category, performing a first identification operation on the business evaluation data to obtain a first processing result, wherein the first identification operation is used to determine the sentiment tendency type corresponding to the business evaluation data and the entity name in the business evaluation data, and the first processing result includes at least one of the risk level, sentiment tendency type, entity name, and cause analysis of the business evaluation data, and the first processing result reflects the business risk of the business associated with the business evaluation data; When the processing category of the business evaluation data is the second category, a second identification operation is performed on the business evaluation data to obtain a second processing result, wherein the first identification operation is used to determine the contextual association of the business evaluation data, the emotional tendency type corresponding to the business evaluation data and the entity name in the business evaluation data, and the second processing result includes at least one of the risk level, emotional tendency type, entity name and cause analysis of the business evaluation data, and the second processing result reflects the business risk of the business associated with the business evaluation data.

2. The method according to claim 1, characterized in that When the processing category of the business evaluation data is the first category, performing a first identification operation on the business evaluation data to obtain a first processing result includes: Inputting the business evaluation data into a first recognition model, wherein the first recognition model includes a keyword matching module and a sentiment analysis module; Determine the semantic similarity between the business evaluation data and a preset target keyword library using the keyword matching module, and determine the entity name based on the semantic similarity; Determine the sentiment tendency type of the business evaluation data using the sentiment analysis module; The first processing result is generated according to the emotional tendency type corresponding to the business evaluation data and the entity name in the business evaluation data.

3. The method according to claim 2, characterized in that Generating the first processing result according to the sentiment tendency type corresponding to the business evaluation data and the entity name in the business evaluation data includes: If the sentiment tendency type corresponding to the business evaluation data is a positive type, determining that the risk level of the business evaluation data is a first level, and generating a positive cause analysis associated with the entity name in the business evaluation data, wherein the cause analysis includes the positive cause analysis; If the sentiment tendency type corresponding to the business evaluation data is a neutral type, determining that the risk level of the business evaluation data is a second level, and generating a neutral cause analysis associated with the entity name in the business evaluation data, wherein the cause analysis includes the neutral cause analysis, and the business risk of the second level is higher than the business risk of the first level; When the sentiment tendency type corresponding to the business evaluation data is a negative type, the risk level of the business evaluation data is determined to be the third level, and a negative cause analysis associated with the entity name is generated, wherein the cause analysis includes the negative cause analysis, and the business risk of the third level is higher than the business risk of the second level.

4. The method according to claim 1, wherein When the processing category of the business evaluation data is the second category, performing a second identification operation on the business evaluation data to obtain a second processing result includes: Inputting the business evaluation data into a second recognition model, wherein the second recognition model includes an event label recognition module, a sentiment tendency recognition module, and an entity recognition module; Processing the business evaluation data using the event tag identification module to obtain contextual association of the business evaluation data; Using the entity recognition module to process the business evaluation data and the contextual association of the business evaluation data, to obtain the sentiment tendency type corresponding to the business evaluation data; Using the sentiment tendency recognition module to process the business evaluation data and the contextual association of the business evaluation data to obtain entity names in the business evaluation data; The second processing result is generated according to the sentiment tendency type corresponding to the business evaluation data and the entity name in the business evaluation data.

5. The method according to claim 1, characterized in that Generating the second processing result according to the context association of the business evaluation data, the sentiment tendency type corresponding to the business evaluation data, and the entity name in the business evaluation data includes: If the sentiment tendency type corresponding to the business evaluation data is a positive type, determining that the risk level of the business evaluation data is a fourth level, and generating a positive cause analysis associated with the entity name in the business evaluation data, wherein the cause analysis includes the positive cause analysis; If the sentiment tendency type corresponding to the business evaluation data is a neutral type, determining that the risk level of the business evaluation data is a fifth level, and generating a neutral cause analysis associated with the entity name in the business evaluation data, wherein the cause analysis includes the neutral cause analysis, and the business risk of the fifth level is higher than the business risk of the fourth level; When the sentiment tendency type corresponding to the business evaluation data is a negative type, the risk level of the business evaluation data is determined to be the sixth level, and a negative cause analysis associated with the entity name is generated, wherein the cause analysis includes the negative cause analysis, and the business risk of the sixth level is higher than the business risk of the fifth level.

6. The method according to claim 1, characterized in that The determining of the processing category of the business evaluation data based on the data attributes of the business evaluation data includes: When the amount of the service evaluation data is less than a preset data amount threshold, determining that the processing category of the service evaluation data is the first category; When the data volume of the service evaluation data is greater than or equal to the data volume threshold, it is determined that the processing category of the service evaluation data is the second category.

7. The method according to claim 1, characterized in that The determining of the processing category of the business evaluation data based on the data attributes of the business evaluation data includes: In a case where the data source address of the business evaluation data belongs to a preset target data source address, determining that the processing category of the business evaluation data is the first category; In a case where the data source address of the service evaluation data does not belong to the preset target data source address, it is determined that the processing category of the service evaluation data is the second category.

8. A business data processing device, characterized in that: include: a determination module, configured to determine a processing category of the business evaluation data based on data attributes of the business evaluation data, wherein the data attributes of the business evaluation data include at least one of data volume and data source address; a first identification module configured to, when the processing category of the business evaluation data is the first category, perform a first identification operation on the business evaluation data to obtain a first processing result, wherein the first identification operation is used to determine the sentiment tendency type corresponding to the business evaluation data and the entity name in the business evaluation data, the first processing result including at least one of the risk level, sentiment tendency type, entity name, and cause analysis of the business evaluation data, and the first processing result reflects the business risk of the business associated with the business evaluation data; The second identification module is used to perform a second identification operation on the business evaluation data to obtain a second processing result when the processing category of the business evaluation data is the second category, wherein the first identification operation is used to determine the context association of the business evaluation data, the emotional tendency type corresponding to the business evaluation data and the entity name in the business evaluation data, and the second processing result includes at least one of the risk level, emotional tendency type, entity name and cause analysis of the business evaluation data, and the second processing result reflects the business risk of the business associated with the business evaluation data.

9. A computer-readable storage medium, characterized in that: The computer-readable storage medium includes a stored computer program, wherein the computer program can be executed by an electronic device to perform the method according to any one of claims 1 to 7.

10. A computer program product comprising a computer program, characterized in that When the computer program is executed by a processor, the steps of the method according to any one of claims 1 to 7 are implemented.

11. An electronic device comprising a memory and a processor, characterized in that: A computer program is stored in the memory, and the processor is configured to execute the method according to any one of claims 1 to 7 through the computer program.

Images