Fund product screening method, electronic equipment, storage medium and program product

Abstract

The embodiment of the invention provides a fund product screening method, electronic equipment, a storage medium and a program product, and relates to the technical field of financial science and technology. The method comprises the steps that a query statement and a fund database are obtained, the query statement is used for indicating fund product requirements of a user, and the fund database comprises a fund product data table and structure information corresponding to the fund product data table; processing the query statement through the target large language model to obtain a structured query demand, the structured query demand being used for indicating a screening condition of a fund product; according to the structured query demand and a fund database, generating an executable SQL query statement; according to the SQL query statement, a target fund product is determined from a fund database, and the target fund product meets the fund product requirement indicated by the query statement. The method can improve the accuracy of the screening result.

Description

Technical Field

[0001] This application relates to the field of financial technology, and in particular to a fund product screening method, electronic device, storage medium, and program product. Background Technology

[0002] In online marketing scenarios for financial products, users submit personalized product requests to the banking system via mobile devices or web pages.

[0003] In related technologies, keywords are typically extracted using natural language processing (NLP) and then combined with the mapping relationship between user feature tags and product tags to filter the product pool. However, with the surge in the number of financial products and the diversification of user needs, the filtering results generated by the above methods suffer from low accuracy. Summary of the Invention

[0004] This application provides a fund product screening method, electronic device, storage medium, and program product to improve the accuracy of screening results.

[0005] In a first aspect, embodiments of this application provide a method for screening fund products, including:

[0006] Retrieve the query statement and fund database. The query statement indicates the user's fund product needs. The fund database includes fund product data tables and the corresponding structure information of the fund product data tables.

[0007] The query statement is processed by the target large language model to obtain structured query requirements, which are used to indicate the screening conditions for fund products.

[0008] Based on the structured query requirements and the fund database, generate executable SQL query statements;

[0009] Based on the SQL query, the target fund product is determined from the fund database. The target fund product meets the fund product requirements indicated by the query.

[0010] In one possible implementation, the query statement is processed through a target large language model to obtain structured query requirements, including:

[0011] The query statement is processed by the target large language model to obtain intent information, which includes at least one of the following: time range, numerical constraints, and logical relationships.

[0012] Transform intent information into structured query requirements that include database fields.

[0013] In one possible implementation, based on the structured query requirements and the fund database, an executable SQL query statement is generated, including:

[0014] Determine the mapping relationship between structured query requirements and structured information, where structured information includes table name, field name, and field description;

[0015] Determine the query logic based on the structured query requirements;

[0016] Determine the SQL query statement based on the query logic and mapping relationship.

[0017] In one possible implementation, the target fund product is determined from a fund database based on an SQL query statement, including:

[0018] Execute the SQL query to obtain at least one candidate fund product;

[0019] The candidate fund products are deduplicated to obtain the target fund products. The deduplication process is used to remove redundant fund products from the candidate fund products.

[0020] In one possible implementation, obtaining the query statement includes:

[0021] Obtain user input data, and the format of user input data includes at least one of the following: text data, voice data, image data, and table data;

[0022] The user input data is standardized to obtain a query statement, wherein the data format of the query statement is text data, and the standardization process includes at least one of the following: converting voice data into text data through a speech recognition module, extracting data from image data through an image recognition module, and extracting data from table data through a structured table parsing module.

[0023] In one possible implementation, after generating an executable SQL query statement based on the structured query requirements and the fund database, the method further includes:

[0024] Call the SQL syntax checking tool via the interface;

[0025] Use an SQL syntax checker to validate the SQL query statement and obtain the validation results;

[0026] If the verification result indicates that the verification failed, the SQL query statement is modified.

[0027] In one possible implementation, after determining the target fund product from the fund database based on an SQL query, the method further includes:

[0028] Receive feedback results, which are used to indicate the user's need for adjustments to the target fund product;

[0029] When feedback indicates that the target fund product needs to be revised, the structured query requirements are dynamically adjusted based on the feedback results, and the target fund product is re-determined based on the adjusted structured query requirements.

[0030] Secondly, embodiments of this application provide a fund product screening device, comprising:

[0031] The acquisition module is used to acquire query statements and fund databases. The query statements indicate the user's fund product needs, and the fund database includes fund product data tables and the corresponding structure information of the fund product data tables.

[0032] The first processing module is used to process the query statement through the target large language model to obtain the structured query requirements, which are used to indicate the screening conditions for fund products.

[0033] The second processing module is used to generate executable SQL query statements based on structured query requirements and the fund database.

[0034] The determination module is used to identify target fund products from the fund database based on SQL query statements. The target fund products must meet the fund product requirements indicated by the query statements.

[0035] In one possible implementation, the first processing module is specifically used for:

[0036] The query statement is processed by the target large language model to obtain intent information, which includes at least one of the following: time range, numerical constraints, and logical relationships.

[0037] Transform intent information into structured query requirements that include database fields.

[0038] In one possible implementation, the second processing module is specifically used for:

[0039] Determine the mapping relationship between structured query requirements and structured information, where structured information includes table name, field name, and field description;

[0040] Determine the query logic based on the structured query requirements;

[0041] Determine the SQL query statement based on the query logic and mapping relationship.

[0042] In one possible implementation, the determining module is specifically used for:

[0043] Execute the SQL query to obtain at least one candidate fund product;

[0044] The candidate fund products are deduplicated to obtain the target fund products. The deduplication process is used to remove redundant fund products from the candidate fund products.

[0045] In one possible implementation, the acquisition module is specifically used for:

[0046] Obtain user input data, and the format of user input data includes at least one of the following: text data, voice data, image data, and table data;

[0047] The user input data is standardized to obtain a query statement, wherein the data format of the query statement is text data, and the standardization process includes at least one of the following: converting voice data into text data through a speech recognition module, extracting data from image data through an image recognition module, and extracting data from table data through a structured table parsing module.

[0048] In one possible implementation, the fund product screening device is also used for:

[0049] Call the SQL syntax checking tool via the interface;

[0050] Use an SQL syntax checker to validate the SQL query statement and obtain the validation results;

[0051] If the verification result indicates that the verification failed, the SQL query statement is modified.

[0052] In one possible implementation, the fund product screening device is also used for:

[0053] Receive feedback results, which are used to indicate the user's need for adjustments to the target fund product;

[0054] When feedback indicates that the target fund product needs to be revised, the structured query requirements are dynamically adjusted based on the feedback results, and the target fund product is re-determined based on the adjusted structured query requirements.

[0055] Thirdly, embodiments of this application provide an electronic device, including: a memory and a processor;

[0056] The memory stores instructions that the computer executes;

[0057] The processor executes computer execution instructions stored in memory, causing the processor to perform the first aspect and / or various possible implementations of the first aspect as described above.

[0058] Fourthly, embodiments of this application provide a computer-readable storage medium storing computer-executable instructions, which, when executed by a processor, are used to implement the first aspect and / or various possible implementations of the first aspect.

[0059] Fifthly, embodiments of this application provide a computer program product, including a computer program that, when executed by a processor, implements the first aspect and / or various possible implementations of the first aspect.

[0060] The fund product screening method, electronic device, storage medium, and program product provided in this application employ a target large language model to semantically parse query statements and generate structured query requirements, which can accurately extract complex screening conditions and logical relationships in the user's intent. By dynamically generating SQL query statements by combining structured query requirements and database table structure information, it can flexibly adapt to various non-pre-defined personalized query scenarios and get rid of dependence on fixed expert rules. By executing the generated SQL statements to perform conditional queries in the fund product data table, it can accurately locate and return target fund products that fully meet the user's described needs, thereby improving the accuracy of query results. Attached Figure Description

[0061] The accompanying drawings, which are incorporated in and form part of this specification, illustrate embodiments consistent with this application and, together with the description, serve to explain the principles of this application.

[0062] Figure 1 This is a schematic diagram of a related technical method process provided in an embodiment of this application;

[0063] Figure 2 Flowchart of the fund product screening method provided for this application Figure 1 ;

[0064] Figure 3 Flowchart of the fund product screening method provided for this application Figure 2 ;

[0065] Figure 4 A schematic diagram of the fund product screening device provided for this application;

[0066] Figure 5 A schematic diagram of the structure of the electronic device provided in this application.

[0067] The accompanying drawings illustrate specific embodiments of this application, which will be described in more detail below. These drawings and descriptions are not intended to limit the scope of the concept in any way, but rather to illustrate the concept of this application to those skilled in the art through reference to particular embodiments. Detailed Implementation

[0068] Exemplary embodiments will now be described in detail, examples of which are illustrated in the accompanying drawings. When the following description relates to the drawings, unless otherwise indicated, the same numbers in different drawings denote the same or similar elements. The embodiments described in the following exemplary embodiments do not represent all embodiments consistent with this application. Rather, they are merely examples of apparatuses and methods consistent with some aspects of this application as detailed in the appended claims.

[0069] It should be noted that the user information (including but not limited to user device information, user personal information, etc.) and data (including but not limited to data used for analysis, data stored, data displayed, etc.) involved in this application are all information and data authorized by the user or fully authorized by all parties. Furthermore, the collection, storage, use, processing, transmission, provision, disclosure, and application of the relevant data all comply with the relevant laws, regulations, and standards of the relevant countries and regions, have taken necessary confidentiality measures, do not violate public order and good morals, and provide corresponding operation access points for users to choose to authorize or refuse.

[0070] Furthermore, the technical solution involved in this application, which involves big data analysis of user information (including but not limited to personal biometrics, identity data, consumption data, asset data, electronic terminal operation data, etc.) and the use of artificial intelligence technology for automated decision-making, and makes decisions that have a significant impact on personal rights based on the results of automated decision-making, provides users with corresponding operation entry points for users to choose to agree to or reject the results of automated decision-making; if the user chooses to reject, the process will proceed to the expert decision-making process.

[0071] As a core scenario of internet marketing, personalized product recommendations to users have always been a top priority for online marketing. Finding the right products for specific user segments without face-to-face interaction has become a crucial challenge for operations personnel. The sale of financial products, due to its high level of expertise, has encountered significant obstacles in the process of digitizing services, leading to a current dilemma in the operation of banking and financial products: no offline traffic and no online performance. Therefore, the ability to quickly find tailor-made products for users online is of paramount importance.

[0072] With the booming development of mobile internet and the increasing richness of online services, traditional offline traffic has gradually shifted online. Online product recommendations have mainly gone through the following stages:

[0073] (1) Enterprises subjectively categorize all products and display them to customers in the form of a showcase, which is equivalent to bringing the offline showcase online. After entering the function, users need to select the products they need themselves. The limitation of this solution is that users do not know much about the products, and there is a lack of personnel to introduce them online, which results in users being unable to make a purchase decision quickly despite the abundance of products.

[0074] (2) Enterprises label customers and products and map the labels using certain expert rules. After users enter the function, the system matches and displays products with corresponding characteristics based on the customer's feature labels. This solution can initially achieve the "one thousand people, many faces" display for customers, that is, customers with different characteristics can see products with corresponding characteristics first, which reflects the advantages of personalized recommendation to a certain extent. However, due to the limited number of feature labels, enterprises are still some distance away from providing "one thousand people, one face" service.

[0075] (3) To further clarify user needs, enterprises use natural language recognition technology to support users in inputting their needs through natural language, obtain user needs through semantic recognition capabilities, and provide recommended products based on user characteristics and product characteristics. This solution can basically achieve "personalized service for each user". After the emergence of large language models, this model has gradually become the main tool for enterprises' AI product operation.

[0076] Although the above scheme (3) has basically achieved a high degree of matching different products to different users, the matching between features in this method requires label mapping through expert rules. Suppose a user asks: I want to find fund products with an annualized return of more than 10% in the past 5 years. After semantic recognition, the user's keywords are identified as: 5 years, annualized return, fund. Based on these keywords, it is not enough to truly understand the user's real needs. Even if combined with user features, it can only return some fund products with relatively high returns, and cannot accurately find the corresponding product, because the expert rule base cannot exhaustively list the user's needs. Therefore, unless the user's needs are predicted in advance, that is, a pool of fund products with an annualized return of more than 10% in the past 5 years is prepared in advance, the matching will definitely fail.

[0077] Figure 1 This is a schematic diagram of a related technical method flow provided in an embodiment of this application. For example... Figure 1 As shown, it includes:

[0078] S101, User inputs product requirements.

[0079] S102. Use semantic recognition capabilities to analyze the product requirements input by the user and extract keywords.

[0080] S103. Obtain user characteristics.

[0081] S104. Obtain the mapping relationship between user characteristics and product characteristic tags.

[0082] S105. Obtain product features based on user characteristics.

[0083] S106. Find the product pool through product characteristics.

[0084] S107. Use keywords obtained from semantic recognition analysis to match product attributes and product features.

[0085] S108, Return Product.

[0086] The limitations of the above scheme are mainly reflected in the following three aspects: (1) Limited semantic recognition capability: It can only extract keywords and cannot understand the deep logic of user needs (such as "the annualized return in the past five years has been higher than 10%). (2) Reliance on expert rules: The tag mapping relationship needs to be preset manually, which cannot cover the user's unpredicted needs, resulting in deviation or omission of matching results. (3) Wide range of tag matching: The tags can only filter out a coarse-grained product pool and cannot accurately locate products that fully match the user's needs (such as funds with specific historical return conditions).

[0087] Based on this, this application provides a method for screening fund products. By employing a target large language model to semantically parse query statements and generate structured query requirements, it can accurately extract complex screening conditions and logical relationships from user intent. By dynamically generating SQL query statements by combining structured query requirements and database table structure information, it can flexibly adapt to various non-pre-defined personalized query scenarios, eliminating reliance on fixed expert rules. By executing the generated SQL statements to perform conditional queries in the fund product data table, it can accurately locate and return target fund products that fully meet the user's described needs, thereby improving the accuracy of query results.

[0088] The technical solution of this application and how the technical solution of this application solves the above-mentioned technical problems are described in detail below with specific embodiments. These specific embodiments can be combined with each other, and the same or similar concepts or processes may not be described again in some embodiments. The embodiments of this application will now be described with reference to the accompanying drawings.

[0089] Figure 2 Flowchart of the fund product screening method provided for this application Figure 1 The execution entity in this application embodiment can be a processor. The processor can be implemented through software or a combination of software and hardware. For example... Figure 2 As shown, the method includes:

[0090] S201. Obtain the query statement and fund database. The query statement is used to indicate the user's fund product requirements. The fund database includes fund product data tables and the corresponding structure information of the fund product data tables.

[0091] A query statement is a text-based statement generated based on a user's query request. Users can initiate query requests via mobile devices or web pages. The processor retrieves the query request through an Application Programming Interface (API) and converts it into a query statement. The user's query request can be one or more of text data, voice data, or chart data. The processor can standardize the query request to obtain the query statement. For example, a query statement might be: "Are there any high-yield and stable investment funds with an annualized return exceeding 10% over the past 5 years?"

[0092] A fund database is a database used in a banking system to store fund products. It can be a relational or non-relational database and includes multiple fund products. The database also includes fund product data tables, which store attribute information about the fund products, such as fields for fund code, annualized return, and risk level. Structure information indicates the structure of the fund product data tables and may include table names, field names, and field descriptions.

[0093] The fund database can be obtained through database interface calls.

[0094] S202. The query statement is processed through the target large language model to obtain structured query requirements, which are used to indicate the screening conditions for fund products.

[0095] The target large language model is a pre-trained machine learning model that can parse user natural language input and generate structured query requirements, such as a pre-trained language model or a natural language generation model.

[0096] Structured query requirements refer to converting user natural language input into text descriptions that include fields, numerical constraints, and logical relationships, such as "annualized rate of return has been higher than 10% for five consecutive years".

[0097] You can call the target language model through its interface and input the query statement into it. The target language model will then perform semantic analysis on the query statement to obtain the structured query requirements.

[0098] For example, the structured query requirement corresponding to the above query statement can be: (1) First, filter out funds from the fund parameter table that have an annualized return of more than 10% in each of the past five years (2020-2024). (2) Main query: Obtain detailed information of these fund products from the fund product data table, such as fund code and name.

[0099] Statement (1) is used to ensure that each fund has complete data records over the past five years, that is, there is return data for each year and all of them are higher than 10%. Statement (2) is used to return all fund products with an annualized return of more than 10% over the past five years, helping users find high-yield and stable funds.

[0100] S203. Based on the structured query requirements and the fund database, generate executable SQL query statements.

[0101] SQL query statements are query statements that conform to database syntax and are used to extract data that meets specific conditions from a fund product data table.

[0102] In one possible implementation, SQL query statements can be generated by calling the primary language model. The target language model can extract the structural information corresponding to the fund data tables in the fund database that are relevant to the structured query requirements, and convert the extracted information into SQL query statements. The primary language model can be the aforementioned target language model, or it can be a different language model.

[0103] For example, the SQL query statement corresponding to the above structured query requirement can be: SELECT fund_id,fund_name FROM fund_info WHERE fund_id IN ( SELECT fund_id FROM annual_return WHERE year >= 2019 AND year <= 2023 AND annual_return > 10 GROUP BY fund_id HAVING COUNT(*) = 5 ).

[0104] Here, `fund_id` is the unique identifier of the fund product, usually the fund code; `fund_name` is the fund name; `fund_info` is the fund product data table; `annual_return` is the annual return table of the fund product; `year >= 2019 AND year <= 2023` means filtering out data between 2019 and 2023; `annual_return > 10` means the annualized return of the fund product is higher than 10%; `GROUP BY fund_id` means grouping the filtered data by `fund_id`, that is, aggregating the filtered data by each fund, so that subsequent statistics can be performed on each group of data; `HAVINGCOUNT(*) = 5` is the filtering condition for the grouped results, that is, counting the number of records in each fund group that meet the above conditions. This number of records must be equal to 5 to ensure that there is data in every year of the past five years and that the annualized return of each record is higher than 10%.

[0105] S204. Based on the SQL query statement, determine the target fund product from the fund database. The target fund product meets the fund product requirements indicated by the query statement.

[0106] The target fund products are the fund products that are ultimately provided to the user and can meet the user's query needs. The target fund products may include multiple fund products, as well as the corresponding fund codes and fund names for each fund product.

[0107] The target fund product can be obtained by executing an SQL query in the fund database.

[0108] The fund product screening method provided in this application employs a target large language model to semantically parse query statements and generate structured query requirements, which can accurately extract complex screening conditions and logical relationships in the user's intent. By dynamically generating SQL query statements by combining structured query requirements and database table structure information, it can flexibly adapt to various non-pre-defined personalized query scenarios and get rid of dependence on fixed expert rules. By executing the generated SQL statements to perform conditional queries in the fund product data table, it can accurately locate and return target fund products that fully meet the user's described needs, thereby improving the accuracy of query results.

[0109] Optionally, the query statement is processed through the target large language model to obtain structured query requirements, including: performing intent recognition processing on the query statement through the target large language model to obtain intent information, the intent information including at least one of the following: time range, numerical constraints, logical relationships; and converting the intent information into structured query requirements containing database fields.

[0110] Intent recognition processing refers to the use of semantic reasoning capabilities by the target large language model to perform deep logical analysis on query statements, such as understanding the time range and continuity constraints in "the annualized return for the past five years has been higher than 10%".

[0111] Intent information refers to the information obtained after the target large language model performs intent recognition processing on the query statement, which is used to indicate the time range, numerical constraints, and logical relationships corresponding to the query statement.

[0112] Numerical constraints refer to restrictions on numerical conditions in a query statement, such as "annualized return is higher than 10%" or "volatility is lower than 5%".

[0113] Logical relationships refer to the combined logic between fields in a query statement, such as "annualized return is higher than 10% and volatility is lower than 5%".

[0114] By leveraging the semantic reasoning capabilities of the target large language model, query statements (such as "Recommend funds with annualized returns exceeding 10% over the past five years") are parsed into intent information including time range (such as 2020-2024), numerical constraints (such as annual_return > 10), and logical relationships (such as GROUP BY fund_id HAVING COUNT(*) = 5). Subsequently, these parsed results are converted into structured query requirements containing database fields (such as "Filter funds with annualized returns exceeding 10% over the past five years"), providing a basis for generating subsequent SQL query statements.

[0115] The above method leverages the semantic reasoning capabilities of a large model to accurately extract the time range, numerical constraints, and logical relationships from the query statement, ensuring that the generated structured query requirements are highly consistent with the user's actual needs. For example, when a user inputs "annualized returns for the past five years have all been higher than 10%", the system can parse out the time range and continuity constraints, and convert them into GROUP BY and HAVING logic, thereby improving the accuracy of the recommendation results.

[0116] Optionally, based on the structured query requirements and the fund database, an executable SQL query statement is generated, including: determining the mapping relationship between the structured query requirements and the structured information, wherein the structured information includes table names, field names, and field descriptions; determining the query logic based on the structured query requirements; and determining the SQL query statement based on the query logic and the mapping relationship.

[0117] The mapping relationship is the correspondence between the intent information in the structured query requirement and the table name and field name in the structure information. For example, mapping the annualized rate of return to annual_return, mapping the fund product data table to fund_info, etc.

[0118] The query logic is information determined based on the logical relationships in the structured query requirements, used to indicate the combination methods and execution order between the mapping relationships.

[0119] Based on the mapping relationship and query logic, and in accordance with SQL syntax rules, a basic SQL framework is constructed, and core statements are concatenated to generate SQL query statements.

[0120] The above method ensures that the generated SQL query statement is fully compatible with the database fields by dynamically matching the database table structure information. For example, "annualized rate of return is higher than 10%" is mapped to annual_return > 10, and combined with GROUPBY and HAVING logic, query failures caused by field mismatches are avoided.

[0121] Optionally, the target fund product is determined from the fund database based on the SQL query statement, including: executing the SQL query statement to obtain at least one candidate fund product; and performing screening processing on the candidate fund products to obtain the target fund product, wherein the screening processing is used to deduplicate the candidate fund products and / or verify the screening conditions of the candidate fund products.

[0122] Candidate fund products are a collection of fund products obtained after executing the SQL query statement, which includes all fund products that meet the query conditions listed in the SQL query statement.

[0123] Deduplication is used to remove duplicate funds from a collection of fund products to identify the target fund product. Filtering processes include, for example, using the DISTINCT statement or a unique identifier (such as fund_id) to eliminate duplicate records.

[0124] This step involves deduplicating the fund products in the execution result after the SQL query is executed. For example, the DISTINCT fund_id statement is used to retain unique fund products, and it is ensured that they meet the time range and numerical constraints (such as an annualized return of more than 10% for five consecutive years).

[0125] The above method eliminates redundant data through deduplication, ensuring the uniqueness and accuracy of the recommendation results. For example, it avoids the problem of duplicate recommendations caused by multiple records corresponding to the same fund product.

[0126] Figure 3 Flowchart of the fund product screening method provided for this application Figure 2 ,like Figure 3 As shown, in this embodiment... Figure 2 Based on the examples, the fund product selection method is described in detail, which includes:

[0127] S301. Obtain the query statement and fund database.

[0128] For example, obtaining a query statement may include: obtaining user input data, the format of which includes at least one of the following: text data, voice data, image data, and table data; standardizing the user input data to obtain a query statement, wherein the data format of the query statement is text data, and the standardization process includes at least one of the following: converting voice data into text data through a voice recognition module, extracting data from image data through an image recognition module, and extracting data from table data through a structured table parsing module.

[0129] User input data refers to data entered by users via mobile devices or web pages. User input data can include various formats, such as an image data entry (e.g., a trend chart of returns) combined with voice input to provide a voice message explaining a structured query request, such as recommending several suitable investment funds based on the image data.

[0130] The speech recognition module is a technology module that converts speech signals into text, such as an API based on Automatic Speech Recognition (ASR).

[0131] The image recognition module is a technology module for extracting key data from charts, such as an API based on Optical Character Recognition (OCR).

[0132] The structured table parsing module is a technical module for parsing tabular data.

[0133] The multimodal input standardization module (speech recognition, image recognition, and structured form parsing) converts user speech, charts, or tables into text descriptions, which are then used as input to the target large language model. For example, after a user uploads a profit trend chart, OCR extracts the year and profit rate data from the chart, and combined with supplementary explanations from the voice input, generates a structured query requirement.

[0134] The methods described above support multimodal input, enabling users to express complex needs through voice, charts, or structured forms. For example, after a user uploads a chart, key data can be extracted and combined with voice input to generate more precise query criteria.

[0135] S302. Process the query statement through the target large language model to obtain structured query requirements.

[0136] S303. Based on the structured query requirements and the fund database, generate executable SQL query statements.

[0137] S304. Perform syntax validation on the SQL query statement and obtain the validation result.

[0138] Syntax validation of SQL query statements includes: calling an SQL syntax checking tool via an interface; using the SQL syntax checking tool to perform syntax validation on the SQL query statement and obtaining the validation results.

[0139] SQL syntax checking tools are tools used to detect syntax errors in SQL query statements, such as SQL Lint.

[0140] The SQL query statement is verified using an SQL syntax checker. The verification result is "verification passed" when the SQL query statement is grammatically correct and "verification failed" when the SQL query statement is grammatically incorrect.

[0141] Optionally, after verifying the syntactic correctness of the SQL query statement, its executability can also be verified through simulated execution. Simulated execution verifies the logical feasibility of the SQL query statement in a virtual environment before actual execution. If the SQL query statement is syntactically correct and executable, the verification result is "verification passed"; if the SQL query statement is syntactically incorrect or unexecutable, the verification result is "verification failed".

[0142] S305. Determine whether the verification result is satisfactory.

[0143] If the verification result indicates that the verification failed, proceed to step S306; if the verification result indicates that the verification passed, proceed to step S307.

[0144] Furthermore, when the verification result indicates a failure, the verification result also includes the reason for the failure.

[0145] S306. Correct the SQL query statement.

[0146] Based on the reasons for failure indicated in the verification results, the SQL query statement is modified accordingly to obtain the final SQL query statement.

[0147] The above method ensures that the generated SQL query statement can be executed directly through syntax and logic verification, avoiding query failure due to syntax errors or logic vulnerabilities.

[0148] S307. Determine the target fund product from the fund database based on the SQL query statement.

[0149] S308, Receive feedback results.

[0150] The feedback results are used to indicate the user's need for adjustments to the target fund product.

[0151] Once the target fund is identified, it can be presented to the user in a visual format, allowing the user to provide real-time feedback via mobile devices. This real-time feedback can include adding new filter criteria or modifying previous ones.

[0152] S309. Determine whether the target fund product needs to be modified based on the feedback results.

[0153] If the feedback result indicates that the target fund product needs to be modified, proceed to step S310. If the feedback result indicates that the target fund product does not need to be modified, end the current screening and determine the target query result determined in step S307 as the final query result.

[0154] S310. Dynamically adjust the structured query requirements based on the feedback results.

[0155] Dynamically adjust structured query requirements. For example, after receiving real-time feedback from users on target fund products (such as "desiring volatility below 5%), adjust logical constraints based on the feedback results (such as adding the condition "volatility < 5" to the original SQL script).

[0156] Based on the adjusted structured query requirements, re-execute step S303 and subsequent steps to determine the adjusted target fund products until the feedback result shows that the target fund products meet the user's needs, and then end this screening.

[0157] The above method dynamically optimizes query conditions through a real-time feedback mechanism, making the target fund products more closely match the user's final needs. For example, users can gradually refine their requirements, thereby dynamically adjusting the SQL script logic.

[0158] Optionally, solutions from related technologies can be retained as a fallback option. If no corresponding fund product is found using this method, results can be returned to the user using tag matching relationships to improve the user experience.

[0159] Figure 4 A schematic diagram of the fund product screening device provided in this application is shown below. Figure 4 As shown, the fund product screening device 20 provided in this embodiment includes:

[0160] The acquisition module 21 is used to acquire query statements and fund databases. The query statements are used to indicate the user's fund product needs. The fund database includes fund product data tables and the corresponding structure information of the fund product data tables.

[0161] The first processing module 22 is used to process the query statement through the target large language model to obtain the structured query requirements, which are used to indicate the screening conditions for fund products.

[0162] The second processing module 23 is used to generate executable SQL query statements based on the structured query requirements and the fund database.

[0163] The determination module 24 is used to determine the target fund product from the fund database based on the SQL query statement. The target fund product meets the fund product requirements indicated by the query statement.

[0164] In one possible implementation, the first processing module 22 is specifically used for:

[0165] The query statement is processed by the target large language model to obtain intent information, which includes at least one of the following: time range, numerical constraints, and logical relationships.

[0166] Transform intent information into structured query requirements that include database fields.

[0167] In one possible implementation, the second processing module 23 is specifically used for:

[0168] Determine the mapping relationship between structured query requirements and structured information, where structured information includes table name, field name, and field description;

[0169] Determine the query logic based on the structured query requirements;

[0170] Determine the SQL query statement based on the query logic and mapping relationship.

[0171] In one possible implementation, the determining module 24 is specifically used for:

[0172] Execute the SQL query to obtain at least one candidate fund product;

[0173] The candidate fund products are deduplicated to obtain the target fund products. The deduplication process is used to remove redundant fund products from the candidate fund products.

[0174] In one possible implementation, the acquisition module 21 is specifically used for:

[0175] Obtain user input data, and the format of user input data includes at least one of the following: text data, voice data, image data, and table data;

[0176] The user input data is standardized to obtain a query statement, wherein the data format of the query statement is text data, and the standardization process includes at least one of the following: converting voice data into text data through a speech recognition module, extracting data from image data through an image recognition module, and extracting data from table data through a structured table parsing module.

[0177] In one possible implementation, the fund product screening device 20 is also used for:

[0178] Call the SQL syntax checking tool via the interface;

[0179] Use an SQL syntax checker to validate the SQL query statement and obtain the validation results;

[0180] If the verification result indicates that the verification failed, the SQL query statement is modified.

[0181] In one possible implementation, the fund product screening device 20 is also used for:

[0182] Receive feedback results, which are used to indicate the user's need for adjustments to the target fund product;

[0183] When feedback indicates that the target fund product needs to be revised, the structured query requirements are dynamically adjusted based on the feedback results, and the target fund product is re-determined based on the adjusted structured query requirements.

[0184] The fund product screening device provided in this embodiment can execute the method provided in the above method embodiment. Its implementation principle and technical effect are similar, and will not be described in detail here.

[0185] Figure 5 A schematic diagram of the structure of the electronic device provided in this application. Figure 5 As shown, the electronic device 30 provided in this embodiment includes at least one processor 301 and a memory 302. Optionally, the electronic device 30 further includes a communication component 303. The processor 301, memory 302, and communication component 303 are connected via a bus.

[0186] In the specific implementation process, at least one processor 301 executes computer execution instructions stored in memory 302, causing at least one processor 301 to execute the above-described fund product screening method embodiment.

[0187] The specific implementation process of processor 301 can be found in the above method embodiments, and its implementation principle and technical effect are similar, so it will not be repeated here. In the above embodiments, it should be understood that the processor can be a Central Processing Unit (CPU), or other general-purpose processors, digital signal processors (DSPs), application-specific integrated circuits (ASICs), etc. A general-purpose processor can be a microprocessor or any conventional processor. The steps of the method disclosed in this invention can be directly manifested as execution by a hardware processor, or execution by a combination of hardware and software modules within the processor.

[0188] The memory may include random access memory (RAM) and may also include non-volatile memory (NVM), such as at least one disk storage device.

[0189] The bus can be an Industry Standard Architecture (ISA) bus, a Peripheral Component Interconnect (PCI) bus, or an Extended Industry Standard Architecture (EISA) bus, etc. Buses can be categorized as address buses, data buses, control buses, etc. For ease of illustration, the buses shown in the accompanying drawings are not limited to a single bus or a single type of bus.

[0190] This application also provides a computer program product, including a computer program that, when executed by a processor, implements the above-described fund product screening method.

[0191] This application also provides a computer-readable storage medium storing computer-executable instructions, which, when executed by a processor, implement the aforementioned fund product screening method.

[0192] The aforementioned readable storage medium can be implemented by any type of volatile or non-volatile storage device or a combination thereof, such as static random access memory (SRAM), electrically erasable programmable read-only memory (EEPROM), erasable programmable read-only memory (EPROM), programmable read-only memory (PROM), read-only memory (ROM), magnetic storage, flash memory, magnetic disk, or optical disk. The readable storage medium can be any available medium accessible to a general-purpose or special-purpose computer.

[0193] An exemplary readable storage medium is coupled to a processor, enabling the processor to read information from and write information to the readable storage medium. Of course, the readable storage medium can also be a component of the processor. The processor and the readable storage medium can reside in an Application Specific Integrated Circuit (ASIC). Alternatively, the processor and the readable storage medium can exist as discrete components in the device.

[0194] The division of units is merely a logical functional division; in actual implementation, there may be other division methods. For example, multiple units or components may be combined or integrated into another system, or some features may be ignored or not executed. Furthermore, the coupling or direct coupling or communication connection shown or discussed may be indirect coupling or communication connection through some interfaces, devices, or units, and may be electrical, mechanical, or other forms.

[0195] The units described as separate components may or may not be physically separate. 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 the units can be selected to achieve the purpose of this embodiment according to actual needs.

[0196] In addition, the functional units in the various embodiments of the present invention can be integrated into one processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit.

[0197] If a function is implemented as a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of this invention, or the part that contributes to the prior art, or a part of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods of the various embodiments of this invention. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.

[0198] Those skilled in the art will understand that all or part of the steps of the above-described method embodiments can be implemented by hardware related to program instructions. The aforementioned program can be stored in a computer-readable storage medium. When executed, the program performs the steps of the above-described method embodiments; and the aforementioned storage medium includes various media capable of storing program code, such as ROM, RAM, magnetic disks, or optical disks.

[0199] Finally, it should be noted that other embodiments of the invention will readily occur to those skilled in the art upon consideration of the specification and practice of the invention disclosed herein. This invention is intended to cover any variations, uses, or adaptations of the invention that follow the general principles of the invention and include common knowledge or customary techniques in the art not disclosed herein, and is not limited to the precise structures described above and shown in the accompanying drawings, and various modifications and changes can be made without departing from its scope. The scope of the invention is limited only by the appended claims.

Claims

1. A method of screening for a fund product, characterized by, The method includes: Obtain a query statement and a fund database. The query statement is used to indicate the user's fund product needs. The fund database includes a fund product data table and the corresponding structure information of the fund product data table. The query statement is processed by the target large language model to obtain structured query requirements, which are used to indicate the screening conditions for fund products. Based on the structured query requirements and the fund database, generate an executable SQL query statement; Based on the SQL query statement, a target fund product is determined from the fund database, and the target fund product meets the fund product requirements indicated by the query statement.

2. The method of claim 1, wherein, The process of processing the query statement using the target large language model to obtain structured query requirements includes: The query statement is processed by the target large language model to obtain intent information, which includes at least one of the following: time range, numerical constraints, and logical relationships. The intent information is transformed into a structured query requirement that includes database fields.

3. The method of claim 1, wherein, The step of generating an executable SQL query statement based on the structured query requirements and the fund database includes: Determine the mapping relationship between the structured query requirements and the structure information, wherein the structure information includes table name, field name, and field description; Based on the structured query requirements, determine the query logic; The SQL query statement is determined based on the query logic and the mapping relationship.

4. The method of claim 1, wherein, The step of determining the target fund product from the fund database based on the SQL query statement includes: Executing the SQL query statement yields at least one candidate fund product; The candidate fund products are deduplicated to obtain the target fund product, wherein the deduplication process is used to remove redundant fund products from the candidate fund products.

5. The method of claim 1, wherein, The process of obtaining the query statement includes: Acquire user input data, wherein the format of the user input data includes at least one of the following: text data, voice data, image data, and table data; The user input data is standardized to obtain the query statement, wherein the data format of the query statement is text data, and the standardization process includes at least one of the following: converting voice data into text data through a speech recognition module, extracting data from image data through an image recognition module, and extracting data from table data through a structured table parsing module.

6. The method of claim 1, wherein, After generating an executable SQL query statement based on the structured query requirements and the fund database, the method further includes: Call the SQL syntax checking tool via the interface; The SQL syntax checking tool is used to perform syntax verification on the SQL query statement, and the verification result is obtained. If the verification result indicates that the verification failed, the SQL query statement is corrected.

7. The method of claim 1, wherein, After determining the target fund product from the fund database based on the SQL query statement, the method further includes: Receive feedback results, which are used to indicate the user's need for adjustments to the target fund product; When the feedback result indicates that the target fund product needs to be modified, the structured query requirement is dynamically adjusted according to the feedback result, and the target fund product is re-determined according to the adjusted structured query requirement.

8. A fund product screening device, characterized in that, include: The acquisition module is used to acquire a query statement and a fund database. The query statement is used to indicate the user's fund product needs, and the fund database includes a fund product data table and the corresponding structure information of the fund product data table. The first processing module is used to process the query statement through the target large language model to obtain a structured query requirement, which is used to indicate the screening conditions for fund products. The second processing module is used to generate executable SQL query statements based on the structured query requirements and the fund database. The determination module is used to determine the target fund product from the fund database based on the SQL query statement, wherein the target fund product meets the fund product requirements indicated by the query statement.

9. An electronic device, characterized in that, include: Memory, processor; The memory stores computer-executed instructions; The processor executes computer execution instructions stored in the memory, causing the processor to perform the method as described in any one of claims 1-7.

10. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores computer-executable instructions, which, when executed by a processor, are used to implement the method as described in any one of claims 1-7.

11. A computer program product, characterized in that, Includes a computer program that, when executed by a processor, implements the method as described in any one of claims 1-7.

Images