Financial index intelligent question and answer method, device and system, medium and product

Abstract

The invention discloses a financial index intelligent question-answering method, device and system, a medium and a product, and relates to the technical field of finance, and the method comprises the steps: carrying out the entity recognition of a financial index question of a user, and generating entity information; performing question judgment on the financial index question according to the entity information to obtain a judgment result; when the judgment result meets the query condition, intention recognition is conducted on the entity information, and a query statement is generated; performing data query according to the query statement to obtain target data; and based on a target operator function, analyzing the target data to obtain an analysis result, and generating an answer to the financial index question according to the analysis result through a large model. The reliability of question answering can be improved.

Description

Technical Field

[0001] The present application relates to the field of financial technology, and in particular to methods, devices, systems, media, and products for intelligent question-answering of financial indicators. Background Art

[0002] DataGPT is a technology that uses natural language processing (NLP) to understand user intent and convert it into structured SQL queries. It parses user input and generates executable database queries, enabling efficient data interaction. However, when faced with complex questions or incomplete database data, it can easily lead to "hallucination" (i.e., incorrect SQL queries are generated), resulting in unreliable answers.

[0003] The above content is only used to assist in understanding the technical solution of this application and does not constitute an admission that the above content is prior art. Summary of the Invention

[0004] The main purpose of this application is to provide a method, device, system, medium and product for intelligent question answering of financial indicators, aiming to solve the technical problem of unreliable answers when faced with complex questions or incomplete database data.

[0005] To achieve the above objectives, this application proposes a financial indicator intelligent question-answering method, which includes:

[0006] Perform entity recognition on the user's financial indicator questions and generate entity information;

[0007] Performing problem judgment on the financial indicator problem based on the entity information to obtain a judgment result;

[0008] When the judgment result meets the query condition, the entity information is subjected to intent recognition to generate a query statement;

[0009] Perform data query according to the query statement to obtain target data; analyze the target data based on the target operator function to obtain analysis results, and generate answers to the financial indicator questions based on the analysis results through a large model.

[0010] In one embodiment, the financial indicator intelligent question-answering system is configured with a target database; wherein the step of performing entity recognition on the user's financial indicator question and generating entity information includes:

[0011] Reading a target data table in the target database, wherein the target data table is organized according to an access table specification, and the access table specification includes one of a narrow table format and a wide table format;

[0012] Entity recognition is performed on the financial indicator problem according to field configuration information of the target data table to generate entity information, wherein the field configuration information includes alias configuration information.

[0013] In one embodiment, the step of determining the financial indicator problem based on the entity information to obtain a determination result includes:

[0014] Determining external authority information of the user;

[0015] Based on the external authority information and the entity information, determining whether the user has query authority and whether the financial indicator problem is a data analysis problem or a future time problem;

[0016] When it is determined that the user has query authority, the financial indicator problem is determined to be a data analysis problem, and the financial indicator problem is determined not to be a future time problem, it is determined that the judgment result meets the query condition.

[0017] In one embodiment, after determining whether the user has query authority and determining whether the financial indicator question is a data analysis question or a future time question based on the external authority information and the entity information, the method further includes:

[0018] When it is determined that the user does not have query authority or when it is determined that the financial indicator question is a future time question, an error message is returned to the user through the large model; and / or,

[0019] When it is determined that the financial indicator problem is not a data analysis problem, a prompt message is returned to the user through the big model.

[0020] In one embodiment, the step of performing intent recognition on the entity information and generating a query statement includes:

[0021] Enumerate the entity information based on at least one of the organization key, the indicator key, the customer type key, the product type key, the industry type key, and the customer group type key to obtain enumeration information;

[0022] Identifying real combination problems in the financial indicator problems based on the enumeration information, and creating a key-value pair set for each real combination problem;

[0023] Put all key-value pairs into the user_intent list and generate a query statement.

[0024] In one embodiment, the steps of performing a data query according to the query statement to obtain target data; analyzing the target data based on a target operator function to obtain an analysis result, and generating an answer to the financial indicator question based on the analysis result using a large model include:

[0025] Determine a target operator function according to the financial indicator problem, wherein the target operator function includes at least one of a feature extraction function, a data aggregation function, a missing value supplementation function, a statistical analysis function, a time series analysis function, a financial business calculation function, and a multi-node analysis function;

[0026] Calling the target operator function to analyze the target data to obtain an analysis result;

[0027] An answer to the financial indicator question is generated based on the target data and the analysis results through a large model.

[0028] In addition, to achieve the above-mentioned purpose, the present application also proposes a financial indicator intelligent question-answering device, which includes:

[0029] The generation module is used to perform entity recognition on the user's financial indicator questions and generate entity information;

[0030] A determination module, configured to determine the financial indicator problem based on the entity information and obtain a determination result;

[0031] The generating module is further configured to perform intent recognition on the entity information and generate a query statement when the judgment result satisfies the query condition;

[0032] A query module, configured to perform data query according to the query statement to obtain target data;

[0033] The generation module is further configured to analyze the target data based on the target operator function to obtain analysis results, and generate answers to the financial indicator questions based on the analysis results through a large model.

[0034] In addition, to achieve the above-mentioned purpose, the present application also proposes a financial indicator intelligent question and answer system, wherein the device includes: a memory, a processor, and a computer program stored on the memory and executable on the processor, wherein the computer program is configured to implement the steps of the financial indicator intelligent question and answer method as described above.

[0035] In addition, to achieve the above-mentioned purpose, the present application also proposes a storage medium, which is a computer-readable storage medium, and a computer program is stored on the storage medium. When the computer program is executed by the processor, the steps of the financial indicator intelligent question and answer method as described above are implemented.

[0036] In addition, to achieve the above-mentioned purpose, the present application also provides a computer program product, which includes a computer program, and when the computer program is executed by a processor, it implements the steps of the financial indicator intelligent question and answer method as described above.

[0037] One or more technical solutions proposed in this application have at least the following technical effects:

[0038] The financial indicator intelligent question-answering method, device, system, medium and product proposed in this application generate entity information by performing entity recognition on the user's financial indicator question; judge the financial indicator question based on the entity information to obtain a judgment result; when the judgment result meets the query condition, perform intent recognition on the entity information to generate a query statement; perform data query based on the query statement to obtain target data; analyze the target data based on the target operator function to obtain an analysis result, and generate an answer to the financial indicator question based on the analysis result through a large model, solving the technical problem of unreliable answers when facing complex questions or incomplete database data. Compared with the existing technology, this invention The application needs to break down the financial indicator problem into multiple entities, and judge whether the financial indicator problem meets the query conditions based on these entities. When the financial indicator problem meets the query conditions, the entity information is identified for intent, and then a query statement is generated. Then, the target data for the financial indicator problem is determined based on the query statement, and the target operator function is called to analyze the target data to obtain the analysis results. Finally, the analysis results are sorted out through a large model and the answer is output. This method can avoid the "unknown number" problem that may arise when directly applying large models for problem analysis, thereby enhancing the accuracy and credibility of the analysis. At the same time, the natural language generation capabilities of the large model are utilized to make the final answer easier for users to understand and meet user needs. BRIEF DESCRIPTION OF THE DRAWINGS

[0039] The accompanying drawings, which are incorporated in and constitute a part of this specification, illustrate embodiments consistent with the present application and, together with the description, serve to explain the principles of the present application.

[0040] In order to more clearly illustrate the embodiments of the present application or the technical solutions in the prior art, the following briefly introduces the drawings required for use in the embodiments or the description of the prior art. Obviously, for ordinary technicians in this field, other drawings can be obtained based on these drawings without any creative work.

[0041] Figure 1 A flowchart of the first embodiment of the intelligent question-answering method for financial indicators provided in this application;

[0042] Figure 2A schematic diagram of the specific process of the intelligent question-answering method for financial indicators provided in Example 1 of this application;

[0043] Figure 3 A schematic diagram of a scenario for the intelligent question-answering method for financial indicators provided in Example 1 of this application;

[0044] Figure 4 A flowchart of the second embodiment of the intelligent question-answering method for financial indicators provided in this application;

[0045] Figure 5 This is a schematic diagram of the module structure of the intelligent question-answering device for financial indicators according to an embodiment of the present application;

[0046] Figure 6 This is a schematic diagram of the device structure of the hardware operating environment involved in the financial indicator intelligent question-answering method in the embodiment of this application.

[0047] The purpose, features and advantages of this application will be further explained in conjunction with the embodiments and with reference to the accompanying drawings. DETAILED DESCRIPTION

[0048] It should be understood that the specific embodiments described herein are merely used to explain the technical solutions of the present application and are not intended to limit the present application.

[0049] In order to better understand the technical solution of the present application, a detailed description will be given below in conjunction with the accompanying drawings and specific implementation methods.

[0050] The main solution of the embodiment of the present application is: performing entity recognition on the user's financial indicator question to generate entity information; performing problem judgment on the financial indicator question based on the entity information to obtain a judgment result; when the judgment result meets the query condition, performing intent recognition on the entity information to generate a query statement; performing data query based on the query statement to obtain target data; analyzing the target data based on the target operator function to obtain an analysis result, and generating an answer to the financial indicator question based on the analysis result through a large model.

[0051] In this embodiment, for ease of description, the following description is made using the intelligent question-answering system for identifying financial indicators as the execution entity.

[0052] As can be seen from the above embodiments, the present application generates entity information by performing entity recognition on the user's financial indicator question; performs problem judgment on the financial indicator question based on the entity information to obtain a judgment result; when the judgment result meets the query condition, performs intent recognition on the entity information to generate a query statement; performs data query based on the query statement to obtain target data; analyzes the target data based on the target operator function to obtain an analysis result, and generates an answer to the financial indicator question based on the analysis result through a large model, which solves the technical problem of unreliable answers when facing complex problems or incomplete database data. Compared with the existing technology, the present application needs to convert the financial indicator into a real-time financial indicator. The target problem is decomposed into multiple entities, and based on these entities, it is judged whether the financial indicator problem meets the query conditions. When the financial indicator problem meets the query conditions, the entity information is identified for intent, and then a query statement is generated. Then, the target data for the financial indicator problem is determined based on the query statement, and the target operator function is called to analyze the target data to obtain the analysis result. Finally, the analysis results are sorted out through the large model and the answer is output. This method can avoid the "unknown number" problem that may occur when directly applying large models for problem analysis, thereby enhancing the accuracy and credibility of the analysis. At the same time, the natural language generation capability of the large model is utilized to make the final answer easier for users to understand and meet user needs.

[0053] It should be noted that the execution entity of this embodiment can be a computing service device with data processing, network communication, and program execution functions, such as a tablet computer, personal computer, mobile phone, etc., or an electronic device capable of implementing the above functions, such as a financial indicator intelligent question-and-answer system. The following uses the financial indicator intelligent question-and-answer system as an example to illustrate this embodiment and the following embodiments.

[0054] Based on this, the embodiment of the present application provides a financial indicator intelligent question answering method, referring to Figure 1 , Figure 1 This is a flow chart of the first embodiment of the intelligent question-answering method for financial indicators of this application.

[0055] In this embodiment, the financial indicator intelligent question-answering method includes steps S10 to S40:

[0056] Step S10: Perform entity recognition on the user's financial indicator question to generate entity information;

[0057] In practice, named entity recognition (NER) technology can be used to identify entities in financial indicator queries, breaking the query down into multiple entities rather than relying on large models for identification. This effectively avoids the "illusions" that can occur with large models and ensures the accuracy of entity information. The core of entity recognition is to clarify the specific data content of the query (i.e., the question of what data is being retrieved). For example, in the question "What is the deposit of the Beijing branch?", the identified entities are "Beijing branch" and "deposits," clearly indicating that the query is for the "deposits" data of the "Beijing branch."

[0058] In a feasible implementation, the financial indicator intelligent question-answering system is configured with a target database; wherein, the step of performing entity recognition on the user's financial indicator question and generating entity information includes: reading a target data table in the target database, wherein the target data table is organized according to an access table specification, and the access table specification includes one of a narrow table format and a wide table format; performing entity recognition on the financial indicator question according to field configuration information of the target data table and generating entity information, wherein the field configuration information includes alias configuration information.

[0059] It should be noted that during the implementation of the Financial Indicators Intelligent Question and Answer System, a target database was configured to ensure accurate processing of questions related to banking indicators. This target database contains the various data tables required by the system, organized according to specific access table specifications to facilitate efficient data reading and parsing. Access table specifications primarily come in two formats: narrow and wide. Narrow tables typically contain multiple records, each representing a single indicator value at a different point in time or under different conditions; wide tables, on the other hand, contain multiple indicator values in a single record, potentially covering different time or analytical dimensions. By pre-establishing these two access table specifications, the system can flexibly adapt to different data structures, thereby improving the accuracy and efficiency of data processing.

[0060] It is understandable that when performing entity recognition on users' financial indicator questions, the financial indicator intelligent question-answering system will read the target data tables in the target database that are organized according to the access table specifications, and then parse the financial indicator questions based on the field configuration information of these data tables to identify the key entities in the financial indicator questions, such as specific dates, institutions, indicators and dimensions, etc. These identified entities will be converted into entity information as the basis for answering user questions.

[0061] In its specific implementation, the intelligent question-and-answer system for financial indicators also supports the configuration of aliases. Banks have many agreed-upon names. For example, "Beijing Branch" is often abbreviated as "North Branch." These entity information is the key to resolving "illusions." When a user asks, "How much is the deposit in the North Branch?", configuring an alias allows the system to understand that the user actually wants to ask, "How much is the deposit in the Beijing branch?"

[0062] Step S20, judging the financial indicator problem based on the entity information to obtain a judgment result;

[0063] It can be understood that judging the problem of bank indicators based on the entity information can determine in advance whether the financial indicator problem meets the query conditions. When the query conditions are not met, there is no need to generate corresponding query conditions, and error information or corresponding prompt information can be directly output through the large model.

[0064] In a feasible implementation, the step of performing problem judgment on the financial indicator problem based on the entity information and obtaining a judgment result includes: determining the external permission information of the user; based on the external permission information and the entity information, judging whether the user has query permission and judging whether the financial indicator problem is a data analysis problem and a future time problem; when it is determined that the user has query permission, the financial indicator problem is a data analysis problem, and the financial indicator problem is not a future time problem, determining that the judgment result meets the query conditions.

[0065] It should be noted that external permission information refers to the user's permission information in all branches, and it is necessary to combine it with entity information to determine whether the user has query permission. For example, the user has query permission for the Beijing Branch and the Nanjing Branch, but does not have query permission for the Guangzhou Branch. When the user wants to query the "bank indicators" related to the Beijing Branch, it can be determined that the user has query permission for this issue. When the user wants to query the bank indicators related to the "Guangzhou Branch", it can be determined that the user does not have query permission for this issue.

[0066] It should be noted that future-time questions refer to questions about financial indicators raised by users that involve a time point or period that has not yet arrived. In other words, the questions contain predictions, plans, or expectations about the future. For example, questions like "What will be the deposit growth rate of the Guangzhou branch next month?" or "How will the loan default rate of the Beijing branch change over the next year?" are all future-time questions. When users ask questions about financial indicators in the future, due to the lack of actual future data and deterministic results, it is impossible to directly provide accurate query results. Therefore, questions about financial indicators in the future will not be considered to meet the query conditions to avoid inaccurate or misleading results due to insufficient data.

[0067] It's important to note that data analysis questions refer to user-generated financial indicator questions that involve the statistical analysis, analysis, mining, or interpretation of existing data to derive valuable information or insights. These questions typically require the use of data processing and analysis techniques, such as data querying, data visualization, trend analysis, and correlation analysis, to draw conclusions. For example, questions like "What was the deposit growth rate of the Beijing branch over the past year?" and "How much more are the self-operated loans of the Beijing branch's Gold Card customers in the transportation industry than the self-operated deposits of the Shanghai branch's Diamond Card customers in the construction industry?" are all data analysis questions.

[0068] In a feasible implementation manner, after the steps of determining whether the user has query authority, determining whether the financial indicator problem is a data analysis problem, and determining whether the financial indicator problem is a future time problem, it also includes: when it is determined that the user does not have query authority or when it is determined that the financial indicator problem is a future time problem, returning an error message to the user through the big model; and / or, when it is determined that the financial indicator problem is not a data analysis problem, returning a prompt message to the user through the big model.

[0069] It should be noted that when the user does not have query permission, the query needs to be terminated and an error message needs to be returned to the user through the big model; when the financial indicator problem is a future time problem, an error message needs to be returned to the user through the big model; when the financial indicator problem is not a data analysis problem, a prompt needs to be returned through the big model according to the situation.

[0070] In actual implementation, for some non-data analysis questions, the large model will output some fixed answers. For example, if you are the AI assistant Xiaowhy, please answer the user's question according to the following requirements:

[0071] Require:

[0072] The current time is: {today}

[0073] Answer user questions simply and don’t make up data

[0074] Whenever I ask the following types of questions, I hope you can output fixed answers:

[0075] *When I ask "What is your function?" or similar questions, you should answer: "Xiao WHY is an intelligent assistant that helps you conduct business analysis. It supports the query and analysis of the bank's key operating indicators. You can ask questions about the indicators you are interested in, and Xiao WHY will answer your questions and conduct exploratory analysis on the indicator."

[0076] *When I ask you "How do I add new features" or similar questions, you should answer: "Contact the relevant administrator."

[0077] *When I ask "What problems can you handle or solve?" or similar questions, you should answer: "Little WHY is an intelligent assistant that helps you conduct business analysis. It supports querying and analyzing the bank's key operating indicators. You can ask questions about the indicators you are interested in, and Little WHY will answer your questions and conduct exploratory analysis on the indicator."

[0078] Step S30: When the judgment result satisfies the query condition, the entity information is subjected to intent recognition to generate a query statement;

[0079] In the specific implementation, when the judgment result meets the query conditions, a SQL query statement can be generated based on the database model to answer the following questions: [Input]

[0080] Database schema:

[0081] CREATE TABLE LM(

[0082] BK1_NM (branch name, such as Beijing Branch),

[0083] BRN_NM (branch name, which is the branch's subordinate institution (e.g., Beijing Changping Branch under Beijing Branch)),

[0084] C01_VAL (balance at that point in time - current day),

[0085] C02_VAL (balance at the time point - increase compared to the previous day),

[0086] C03_VAL (balance at the time point - increase compared to the end of the previous month),

[0087] C04_VAL (balance at the time point - increase compared to the end of the previous year),

[0088] C05_VAL (monthly average - current day),

[0089] C06_VAL (monthly average - increase compared to the previous day),

[0090] C07_VAL (monthly average - increase compared to the end of the previous month),

[0091] C08_VAL (monthly daily average - increase compared to the end of the previous year),

[0092] C09_VAL (annual daily average - current day),

[0093] C10_VAL (annual daily average - increase compared to the previous day),

[0094] C11_VAL (annual daily average - increase compared to the end of the previous month),

[0095] C12_VAL (annual daily average - increase compared to the end of the previous year),

[0096] CCY_CNV (conversion currency),

[0097] CCY_ORG (original currency),

[0098] DW_STAT_DT (data date, for example, 2022-07-31),

[0099] IND_CD string (indicator number),

[0100] IND_NM (Indicator Name) );

[0102] In a feasible implementation, the steps of performing intent recognition on the entity information and generating a query statement include: enumerating the entity information based on at least one of an organization key, an indicator key, a customer type key, a product type key, an industry type key, and a customer group type key to obtain enumeration information; identifying real combination problems in the financial indicator problem based on the enumeration information, and creating a key-value pair set for each real combination problem; placing all key-value pair sets into a user_intent list to generate a query statement.

[0103] It should be noted that since financial indicator questions may be multi-dimensional analysis questions, it is necessary to extract the user's real combination questions from the financial indicator questions based on the entity information, and create a JSON object (i.e., a key-value pair collection) containing the necessary information for each real question. These collections are then organized into a list called user_intent, where each list item represents an independent query intent.

[0104] In practice, when the financial indicator question is a multi-dimensional analysis question, intent recognition is needed to extract the user's true combination question. Specifically, a JSON-formatted output is created based on the financial indicator question, containing information about the combination question. The output should contain a "user_intent" list, each of which contains only one data query item. Each list has the following keys:

[0105] "branches":"<single organization>",

[0106] "indicators":"<single indicator>"

[0107] "customerType":"<single customer type>"

[0108] "productType":"<single product type>"

[0109] "industryTypes":"<single industry type>"

[0110] "customerGroupTypes":"<single customer group type>"

[0111] In the specific implementation, identify the organizations, indicators, customer types, and team types mentioned in the financial indicator questions, and then determine all the real combination questions in the financial indicator questions based on the identification results. Create a JSON object for each real combination question. Each JSON object contains the relevant "branches", "indicatorscustomerTypes", "productTypes", "industryTypes", and "customerGroupTypes" keys. Then put these JSON objects in the "user_intent" list and ensure that each JSON object contains all information related to the combination question in the user question.

[0112] In the specific implementation, we will use the financial indicator question "How much more are the self-operated loans of the Beijing branch's Gold Card customers in the transportation industry than the self-operated deposits of the Shanghai branch's Diamond Card customers in the construction industry?" as an example to illustrate. Specifically:

[0113] 1. Determine enumeration information based on entity information:

[0114] {"branches":["Beijing Branch","Shanghai Branch"],"indicators":["Proprietary Loans","Proprietary Deposits"],"customerTypes":[],"productTypes":[],"industrxTypes":["Transportation Industry"Construction Industry"],"customerGroupTypes":["Gold Card Customer Group","Diamond Customer Group"]}

[0115] 2. Then determine the real combination question based on the enumeration information, create a JSON object for each real combination question, and finally put all JSON objects into the "user_intent" list:

[0116] {{"user_intent":[{{"branches":["Beijing Branch"],"indicators":["Proprietary Loans"],"industryTypes":["Transportation Industry"],"customerGroupTypes":["Gold Card Customer Group"]}},{{"branches":["Shanghai Branch"],"indicators":["Proprietary Deposits"],"industryTypes":["Construction Industry"],"customerGroupTypes":["Diamond Customer Group"]}}]}}

[0117] In practice, here are some example user questions and their corresponding JSON output:

[0118] 1. Financial indicator question: What is the respective self-operated loan amount for gold card customers and silver card customers in the Shenyang branch?

[0119] Output: {{"user intent":[{{"branches":["Shenyang Branch"],"indicators":["Proprietary Deposits"]}},{{"branches":["Chengdu Branch"],"indicators":["Proprietary Deposits"]}}]}}

[0120] 2. Financial indicator question: What is the amount of proprietary deposits held by domestic insurance companies and national joint-stock commercial banks in their Chengdu branches?

[0121] Output: {{"user_intent":[{{"branches":["Chengdu Branch"],"customerGroupTypes":["Domestic Insurance Companies"],"indicators":["Proprietary Deposits"]}},{{"branches":["Chengdu Branch"],"customerGroupTypes":["National Joint-Stock Commercial Banks"],"indicators":["Proprietary Deposits"]}}]}}

[0122] 3. Financial indicator question: What is the proportion of fixed-term savings deposits of Dongguan Branch and the proportion of combination B-fixed deposits of Shenyang Branch?

[0123] Output: {{"user_intent":[{{"branches":["Dongguan Branch"],"indicators":["Percentage of fixed-term savings deposits"]}},{{"branches":["Shenyang Branch"],"productTypes":["Combination B - Fixed-term deposit"]}}]}}

[0124] Step S40, performing data query according to the query statement to obtain target data;

[0125] It should be noted that after querying data in the database according to the query statement, a query can be performed on the bank indicator problem to finally obtain the target data. Specifically, for example, when the deposits of the Beijing branch are required in the bank indicator problem, the deposit data of the Beijing branch can be queried from the database.

[0126] Step S50: Analyze the target data based on the target operator function to obtain analysis results, and generate answers to the financial indicator questions based on the analysis results through a large model.

[0127] It should be noted that data related to financial indicator problems can be retrieved from the database based on query statements, and then operators are called to perform statistical analysis on these target data to generate analysis results. Finally, these target data and analysis results are sent to the large model, which will organize them and uniformly output the answers to the financial indicator problems.

[0128] In the specific implementation, Figure 2 As shown, entity recognition is first performed to extract key entity information in the question, such as a specific branch or sub-branch. This entity information is then used to judge the question to determine whether it meets the query conditions. If the question does not meet the query conditions, such as when it involves data that is not accessible to permission, inquiries about data in the future, or non-data analysis questions, the system will directly call the large language model (LLM) to generate the answer. For questions that meet the query conditions, intent recognition is required before the query statement can be generated. After the analysis results are determined based on the query statement, the analysis results are input into the LLM (i.e., large model). The LLM generates the final answer based on these results and returns it to the user. This process ensures that the system can efficiently and accurately handle complex bank indicator query questions, while leveraging the natural language generation capabilities of the large model to provide answers that are easy to understand and meet business needs, thereby significantly improving user experience and work efficiency.

[0129] In the specific implementation, Figure 3 As shown, when the user enters "How much is the deposit of the Beijing branch?" in the question input field, the final generated "The deposit of the Beijing branch is xxx" is returned to the user through the large model.

[0130] This embodiment generates entity information by performing entity recognition on the user's financial indicator question; performs question judgment on the financial indicator question based on the entity information to obtain a judgment result; when the judgment result meets the query condition, performs intent recognition on the entity information to generate a query statement; performs data query based on the query statement to obtain target data; analyzes the target data based on the target operator function to obtain an analysis result, and generates an answer to the financial indicator question based on the analysis result through a large model, which solves the technical problem of unreliable answers when facing complex questions or incomplete database data. Compared with the existing technology, this application needs to decompose the financial indicator question into It is solved into multiple entities, and based on these entities, it is judged whether the financial indicator problem meets the query conditions. When the financial indicator problem meets the query conditions, the entity information is intended to be identified, and then a query statement is generated. Then, the target data for the financial indicator problem is determined based on the query statement, and the target operator function is called to analyze the target data to obtain the analysis results. Finally, the analysis results are sorted out through the large model and the answer is output. This method can avoid the "unknown number" problem that may occur when directly applying large models for problem analysis, thereby enhancing the accuracy and credibility of the analysis. At the same time, the natural language generation capability of the large model is utilized to make the final answer easier for users to understand and meet user needs.

[0131] Based on the first embodiment of the present application, in the second embodiment of the present application, the same or similar contents as those in the above embodiment 1 can be referred to the above introduction and will not be described in detail later. Figure 4 , step S40 further includes steps S401 to S403:

[0132] Step S401: determining a target operator function according to the financial indicator problem, wherein the target operator function includes at least one of a feature extraction function, a data aggregation function, a missing value supplementation function, a statistical analysis function, a time series analysis function, a financial business calculation function, and a multi-node analysis function;

[0133] It's important to note that each operator is an independent function, meaning it can be called and executed independently, resulting in a high degree of modularity. This independence makes operator services more flexible when it comes to functional expansion, maintenance, and optimization. Individual operators can be improved or updated based on specific needs without affecting the normal operation of other operators.

[0134] It's important to note that data feature extraction functions analyze basic data characteristics, including the presence of missing values, how data trends change over time, and trends and periodicity. When processing banking indicator data, data feature extraction can quickly reveal data integrity and regularity. Data aggregation functions are primarily used to aggregate data along the time dimension, for example, aggregating daily data into monthly or quarterly data. Banking indicator data is typically recorded and analyzed at different time granularities. Missing value filling functions address missing data, filling in missing values at specific time points or data points. In real-world banking data, missing data may occur. Missing value filling can restore data integrity and ensure proper use in subsequent analysis and calculations. Statistical analysis functions analyze data statistically, including distribution, central tendency, and dispersion. Time series analysis functions analyze time series data to determine characteristics such as stationarity, trend, and periodicity. Time series analysis can reveal patterns and trends in data over time, which is crucial for predicting future data changes, formulating business strategies, and managing risk. Financial business calculation functions are primarily used to perform specific financial business calculations, such as year-over-year and month-over-month calculations. These calculations are an integral part of bank data analysis, providing specific quantitative indicators for bank business assessment and decision-making, helping banks better understand business development trends. Multi-node analysis functions are primarily used to perform comparative and correlation analysis on sequence diagrams across multiple nodes (such as branches and regions). Multi-node analysis can reveal the business relationships and differences between different nodes at a macro level, providing a comprehensive perspective for comprehensive management and decision-making, and helping to optimize a bank's business structure and resource allocation.

[0135] Step S402: calling the target operator function to analyze the target data to obtain an analysis result;

[0136] It should be noted that target data can be extracted from the database based on the query statement, and this target data can be deeply analyzed using target operator functions to obtain valuable analytical results. Specifically, the query statement is first parsed to identify the entity information involved, such as "Beijing Branch." Based on this entity information, all relevant data can be accurately retrieved from the database, such as the Beijing branch's deposit balance, historical transaction records, customer information, etc., and then predefined target operator functions are called to analyze this target data. These operator functions may include various types such as feature extraction, data aggregation, missing value supplementation, statistical analysis, and time series analysis. They can process and analyze data from different perspectives, and ultimately generate analytical results through this process.

[0137] Step S403: Generate an answer to the financial indicator question based on the target data and the analysis results using a large model.

[0138] It is understandable that using the natural language generation capabilities of large models to generate answers based on target data and analysis results can make the final answers easier for users to understand and meet user needs.

[0139] In this implementation, you are the AI assistant Xiaowhy, specifically responsible for analyzing the bank's core indicators and providing corresponding explanations. Please answer according to the following requirements:

[0140] -The current time is: {today}. If today's data cannot be obtained, yesterday's data will be obtained.

[0141] - The original data provided shall not be modified in any way

[0142] Analyze the problem in detail one by one based on the following points:

[0143] 1. Based on the given statistical data, analyze the reasons for changes in indicator data (trends, basic data conditions, increments / increases).

[0144] 2. If you find extreme data, please tell me the maximum and minimum values and the dates on which they occurred.

[0145] ###Related indicator data:

[0146] {data}

[0147] ###User Question:

[0148] {question}

[0149] ###Answer example:

[0150] The detailed analysis is as follows:

[0151] 1. Statistical Data Analysis

[0152] 2. Extreme value.

[0153] In a specific implementation, hypothetical questions can be determined based on the entity information of the bank indicator question, and then pushed to the user for selection. Finally, the user's selected question is answered. The bank indicator question can also be refined based on the user's historical query information, so that the user's question can be answered based on the bank indicator question in one go, eliminating the need for the user to enter the question multiple times and affect the user experience. Specifically, for example, if a user enters "What is the deposit of the Beijing branch?", the bank indicator question can be refined based on the user's historical query information, for example, the question can be refined to "What was the deposit growth rate of the Beijing branch over the past year?" User historical query information can reflect a user's query preferences, making it possible to quickly refine the bank indicator question entered by the user in one go.

[0154] This embodiment determines the target operator function according to the financial indicator problem, wherein the target operator function includes at least one of a feature extraction function, a data aggregation function, a missing value supplement function, a statistical analysis function, a time series analysis function, a financial business calculation function, and a multi-node analysis function; calls the target operator function to analyze the target data to obtain an analysis result; and generates an answer to the financial indicator problem based on the target data and the analysis result through a large model. Through the above method, the present application can efficiently and accurately handle complex bank indicator query problems, provide accurate and in-depth analysis results, and at the same time use the powerful generation capability of the large model to generate answers that are easy for users to understand and meet business needs, thereby improving user experience and work efficiency.

[0155] It should be noted that the above examples are only used to understand this application and do not constitute a limitation on the financial indicator intelligent question-answering method of this application. More simple transformations based on this technical concept are all within the scope of protection of this application.

[0156] This application also provides a financial indicator intelligent question-answering device, please refer to Figure 5 , the financial indicator intelligent question-answering device includes:

[0157] A generation module 10 is used to perform entity recognition on the user's financial indicator question and generate entity information;

[0158] A determination module 20 is configured to determine the financial indicator problem based on the entity information and obtain a determination result;

[0159] The generating module 10 is further configured to perform intent recognition on the entity information and generate a query statement when the judgment result satisfies the query condition;

[0160] A query module 30 is used to perform data query according to the query statement to obtain target data;

[0161] The generating module 10 is further configured to analyze the target data based on the target operator function to obtain an analysis result, and generate an answer to the financial indicator question based on the analysis result through a large model.

[0162] The financial indicator intelligent question-answering device provided in this application, which utilizes the financial indicator intelligent question-answering method described in the aforementioned embodiments, can resolve the technical issue of unreliable answers when faced with complex questions or when database data is incomplete. Compared to the prior art, the financial indicator intelligent question-answering device provided in this application has the same beneficial effects as the financial indicator intelligent question-answering method described in the aforementioned embodiments. Other technical features of the financial indicator intelligent question-answering device are the same as those disclosed in the aforementioned embodiments and are not further elaborated here.

[0163] The present application provides a financial indicator intelligent question-answering device, which includes: at least one processor; and a memory communicatively connected to the at least one processor; wherein the memory stores instructions executable by the at least one processor, and the instructions are executed by the at least one processor to enable the at least one processor to execute the financial indicator intelligent question-answering method in the above-mentioned embodiment 1.

[0164] Reference below Figure 6 , which shows a schematic diagram of the structure of a financial indicator intelligent question-and-answer device suitable for implementing embodiments of the present application. The financial indicator intelligent question-and-answer device in the embodiments of the present application may include, but is not limited to, mobile terminals such as mobile phones, laptop computers, digital broadcast receivers, PDAs (Personal Digital Assistants), PADs (Portable Application Descriptions), PMPs (Portable Media Players), and in-vehicle terminals (e.g., in-vehicle navigation terminals), as well as fixed terminals such as digital TVs and desktop computers. Figure 6 The financial indicator intelligent question-and-answer device shown is merely an example and should not limit the functions and scope of use of the embodiments of the present application.

[0165] like Figure 6As shown, the financial indicator intelligent question-answering device may include a processing device 1001 (e.g., a central processing unit, a graphics processing unit, etc.), which can perform various appropriate actions and processes based on programs stored in a read-only memory 1002 or programs loaded from a storage device 1003 into a random access memory 1004. The random access memory 1004 also stores various programs and data required for the operation of the financial indicator intelligent question-answering device. The processing device 1001, the read-only memory 1002, and the random access memory 1004 are interconnected via a bus 1005. An input / output interface 1006 is also connected to the bus. Typically, the following systems may be connected to the input / output interface 1006: an input device 1007 including, for example, a touchscreen, touchpad, keyboard, mouse, image sensor, microphone, accelerometer, gyroscope, etc.; an output device 1008 including, for example, a liquid crystal display (LCD), speaker, vibrator, etc.; a storage device 1003 including, for example, a magnetic tape or hard disk; and a communication device 1009. Communication device 1009 can allow the financial indicator intelligent question-and-answer device to communicate wirelessly or wired with other devices to exchange data. Although the figure shows a financial indicator intelligent question-and-answer device with various systems, it should be understood that it is not required to implement or have all of the systems shown. More or fewer systems may be implemented or have alternatively.

[0166] In particular, according to the embodiments disclosed in the present application, the processes described above with reference to the flowcharts can be implemented as computer software programs. For example, the embodiments disclosed in the present application include a computer program product comprising a computer program carried on a computer-readable medium, the computer program comprising program code for executing the method shown in the flowchart. In such an embodiment, the computer program can be downloaded and installed from a network via a communication device, or installed from a storage device 1003, or installed from a read-only memory 1002. When the computer program is executed by the processing device 1001, the above-mentioned functions defined in the method of the embodiment disclosed in the present application are performed.

[0167] The financial indicator intelligent question-answering device provided in this application, employing the financial indicator intelligent question-answering method described in the aforementioned embodiments, can resolve the technical issue of unreliable answers to complex questions or when database data is incomplete. Compared to the prior art, the beneficial effects of the financial indicator intelligent question-answering device provided in this application are the same as those of the financial indicator intelligent question-answering method described in the aforementioned embodiments. Other technical features of this financial indicator intelligent question-answering device are the same as those disclosed in the aforementioned embodiments and are not further elaborated here.

[0168] It should be understood that the various parts disclosed in this application can be implemented using hardware, software, firmware, or a combination thereof. In the description of the above embodiments, specific features, structures, materials, or characteristics can be combined in any one or more embodiments or examples in a suitable manner.

[0169] The above description is merely a specific embodiment of the present application, but the scope of protection of the present application is not limited thereto. Any changes or substitutions that can be easily conceived by a person skilled in the art within the technical scope disclosed in this application should be included in the scope of protection of this application. Therefore, the scope of protection of this application should be based on the scope of protection of the claims.

[0170] The present application provides a computer-readable storage medium having computer-readable program instructions (i.e., computer programs) stored thereon, and the computer-readable program instructions are used to execute the financial indicator intelligent question-answering method in the above-mentioned embodiment.

[0171] The computer-readable storage medium provided in this application may be, for example, a USB flash drive, but is not limited to electrical, magnetic, optical, electromagnetic, infrared, or semiconductor systems or devices, or any combination thereof. More specific examples of computer-readable storage media may include, but are not limited to: an electrical connection with one or more wires, a portable computer disk, a hard disk, a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), an optical fiber, a portable compact disk read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination thereof. In this embodiment, the computer-readable storage medium may be any tangible medium that contains or stores a program that can be used by or in conjunction with an instruction execution system or device. The program code contained on the computer-readable storage medium may be transmitted using any appropriate medium, including but not limited to: wires, optical cables, RF (Radio Frequency), etc., or any suitable combination thereof.

[0172] The computer-readable storage medium may be included in the financial indicator intelligent question-answering device; or it may exist independently without being assembled into the financial indicator intelligent question-answering device.

[0173] The computer-readable storage medium carries one or more programs. When the one or more programs are executed by the financial indicator intelligent question-answering device, the financial indicator intelligent question-answering device: performs entity recognition on the user's financial indicator question to generate entity information; performs question judgment on the financial indicator question based on the entity information to obtain a judgment result; when the judgment result meets the query condition, performs intent recognition on the entity information to generate a query statement; performs data query based on the query statement to obtain target data; analyzes the target data based on the target operator function to obtain an analysis result, and generates an answer to the financial indicator question based on the analysis result through a large model.

[0174] Computer program code for performing the operations of the present application may be written in one or more programming languages, or a combination thereof, including object-oriented programming languages such as Java, Smalltalk, C++, and conventional procedural programming languages such as "C" or similar programming languages. The program code may be executed entirely on the user's computer, partially on the user's computer, as a stand-alone software package, partially on the user's computer and partially on a remote computer, or entirely on a remote computer or server. In cases involving a remote computer, the remote computer may be connected to the user's computer through any type of network, including a local area network (LAN) or a wide area network (WAN), or may be connected to an external computer (e.g., via the Internet using an Internet service provider).

[0175] The flow charts and block diagrams in the accompanying drawings illustrate the possible architecture, functions and operations of the systems, methods and computer program products according to various embodiments of the present application. In this regard, each box in the flow chart or block diagram can represent a module, program segment or a part of code, and the module, program segment or a part of code contains one or more executable instructions for realizing the specified logical function. It should also be noted that in some alternative implementations, the functions marked in the box can also occur in a different order than that marked in the accompanying drawings. For example, two boxes represented in succession can actually be executed substantially in parallel, and they can sometimes be executed in the opposite order, depending on the functions involved. It should also be noted that each box in the block diagram and / or flow chart, and the combination of the boxes in the block diagram and / or flow chart can be implemented by a dedicated hardware-based system that performs the specified function or operation, or can be implemented by a combination of dedicated hardware and computer instructions.

[0176] The modules described in the embodiments of the present application may be implemented in software or hardware, wherein the name of a module does not necessarily limit the unit itself.

[0177] The computer-readable storage medium provided in this application is a computer-readable storage medium that stores computer-readable program instructions (i.e., a computer program) for executing the aforementioned intelligent question-and-answer method for financial indicators. This computer-readable storage medium can address the technical issue of unreliable answers when faced with complex questions or when database data is incomplete. Compared to the prior art, the beneficial effects of the computer-readable storage medium provided in this application are the same as those of the intelligent question-and-answer method for financial indicators provided in the aforementioned embodiments, and are not further elaborated here.

[0178] The present application also provides a computer program product, including a computer program, which implements the steps of the above-mentioned financial indicator intelligent question-answering method when executed by a processor.

[0179] The computer program product provided in this application can address the technical issue of unreliable answers to complex questions or when incomplete database data is present. Compared to existing technologies, the beneficial effects of the computer program product provided in this application are similar to those of the intelligent question-and-answer method for financial indicators provided in the aforementioned embodiments, and are not further elaborated here.

[0180] The above description is only part of the embodiments of the present application and does not limit the patent scope of the present application. All equivalent structural transformations made by using the contents of the present application specification and drawings under the technical concept of the present application, or direct / indirect application in other related technical fields are included in the patent protection scope of the present application.

Claims

1. A financial indicator intelligent question-answering method, characterized in that: Applied to a financial indicator intelligent question-answering system, the method includes: Perform entity recognition on the user's financial indicator questions and generate entity information; Performing problem judgment on the financial indicator problem based on the entity information to obtain a judgment result; When the judgment result meets the query condition, the entity information is subjected to intent recognition to generate a query statement; Perform data query according to the query statement to obtain target data; Based on the target operator function, the target data is analyzed to obtain analysis results, and the answer to the financial indicator question is generated according to the analysis results through a large model.

2. The method according to claim 1, wherein The financial indicator intelligent question-answering system is configured with a target database; wherein the step of performing entity recognition on the user's financial indicator question and generating entity information includes: Reading a target data table in the target database, wherein the target data table is organized according to an access table specification, and the access table specification includes one of a narrow table format and a wide table format; Entity recognition is performed on the financial indicator problem according to field configuration information of the target data table to generate entity information, wherein the field configuration information includes alias configuration information.

3. The method according to claim 1, wherein The step of determining the financial indicator problem based on the entity information to obtain a determination result includes: Determining external authority information of the user; Based on the external authority information and the entity information, determining whether the user has query authority and whether the financial indicator problem is a data analysis problem or a future time problem; When it is determined that the user has query authority, the financial indicator problem is determined to be a data analysis problem, and the financial indicator problem is determined not to be a future time problem, it is determined that the judgment result meets the query condition.

4. The method according to claim 3, wherein After the steps of determining whether the user has query authority and determining whether the financial indicator problem is a data analysis problem or a future time problem based on the external authority information and the entity information, the method further includes: When it is determined that the user does not have query authority or when it is determined that the financial indicator question is a future time question, an error message is returned to the user through the large model; and / or, When it is determined that the financial indicator problem is not a data analysis problem, a prompt message is returned to the user through the big model.

5. The method according to claim 1, wherein The step of performing intent recognition on the entity information and generating a query statement includes: Enumerate the entity information based on at least one of the organization key, the indicator key, the customer type key, the product type key, the industry type key, and the customer group type key to obtain enumeration information; Identifying real combination problems in the financial indicator problems based on the enumeration information, and creating a key-value pair set for each real combination problem; Put all key-value pairs into the user_intent list and generate a query statement.

6. The method according to claim 1, wherein The steps of analyzing the target data based on the target operator function to obtain analysis results, and generating answers to the financial indicator questions based on the analysis results using a large model include: Determine a target operator function according to the financial indicator problem, wherein the target operator function includes at least one of a feature extraction function, a data aggregation function, a missing value supplementation function, a statistical analysis function, a time series analysis function, a financial business calculation function, and a multi-node analysis function; Calling the target operator function to analyze the target data to obtain an analysis result; An answer to the financial indicator question is generated based on the target data and the analysis results through a large model.

7. A financial indicator intelligent question-answering device, characterized in that: The device comprises: The generation module is used to perform entity recognition on the user's financial indicator questions and generate entity information; A determination module, configured to determine the financial indicator problem based on the entity information and obtain a determination result; The generating module is further configured to perform intent recognition on the entity information and generate a query statement when the judgment result satisfies the query condition; A query module, configured to perform data query according to the query statement to obtain target data; The generation module is further configured to analyze the target data based on the target operator function to obtain analysis results, and generate answers to the financial indicator questions based on the analysis results through a large model.

8. A financial indicator intelligent question-answering system, characterized in that: The system includes: a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the computer program is configured to implement the steps of the financial indicator intelligent question-answering method according to any one of claims 1 to 6.

9. A storage medium, characterized in that: The storage medium is a computer-readable storage medium, and a computer program is stored on the storage medium. When the computer program is executed by a processor, the steps of the financial indicator intelligent question-answering method according to any one of claims 1 to 6 are implemented.

10. A computer program product, characterized in that The computer program product includes a computer program, and when the computer program is executed by a processor, the steps of the financial indicator intelligent question and answer according to any one of claims 1 to 6 are implemented.

Images