Index fluctuation attribution analysis method and system fused with large language model, electronic equipment and storage medium

Through the indices fluctuation attribution analysis method of fusion large language model, semantic analysis requests are generated based on user questions and attribution dimension metadata, and mapped to the target tool for attribution analysis, solving the high threshold and low efficiency of attribution analysis in the existing technology. The problem of the problem has achieved an efficient and accurate summary of attribution results.

CN119003737AActive Publication Date: 2024-11-22SHENZHEN YISHIHUOLALA TECH CO LTD

Patent Information

Application Number
CN202411227577.6
Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Filing Date
2024-09-03
Publication Date
2024-11-22
Estimated Expiration
2044-09-03

AI Technical Summary

Technical Problem

The existing attribution analysis method for abnormal business indicators has a high threshold and is inefficient in use. Operators need to be familiar with multiple subsystems and tools, and the query configuration method of each tool is different, resulting in a long configuration time and complex comprehensive analysis of results.

Method used

Using the attribution analysis method of indicator fluctuation integrated with large language models, determine the attribution indicators and attribution dimension metadata through user questions, generate semantic analysis requests, and use large language model processing to generate DSL objects, and map them to target attribution. Analytical tools, attribution analysis, and ultimately use large language models to summarize the results to reduce the dependence and error risks of operational personnel.

Benefits of technology

It lowers the threshold for using attribution analysis tools, improves the efficiency and accuracy of attribution analysis, reduces the operator's dependence on each tool, and solves the error risk when large language models are used alone.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN119003737A_ABST
    Figure CN119003737A_ABST
Patent Text Reader

Abstract

The invention provides an index fluctuation attribution analysis method and system fused with a large language model, electronic equipment and a storage medium, and the method comprises the steps: determining a corresponding to-be-attributed index according to a user question sent by a user, and obtaining a pre-first text template and attribution dimension metadata corresponding to the to-be-attributed index; generating a semantic analysis request according to the user question, the first text template and the attribution dimension metadata; processing the semantic analysis request through a large language model to obtain a DSL object, and generating a corresponding attribution analysis request according to the DSL object; determining at least one target attribution analysis tool according to the question of the user, and mapping the attribution analysis request to each target attribution analysis tool; performing attribution analysis according to the attribution analysis request through each target attribution analysis to obtain at least one attribution analysis result; and outputting an attribution result summary of the to-be-attributed index through a big language model according to a result summary request generated by the second text template, the user question and the at least one attribution analysis result.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The present application relates to the field of computer technology, and more specifically, to an indicator fluctuation attribution analysis method, system, electronic device and storage medium integrating a large language model. Background Art

[0002] With the continuous development of enterprises, in order to better achieve sustainable growth, enterprises build corresponding multiple business indicators to summarize the key information in business activities related to the enterprise in a standardized manner; with the increase of business indicators, in order to avoid affecting the normal operation of the enterprise, real-time attention to changes in business indicators has gradually become the focus of daily operations of the enterprise.

[0003] At present, the attribution analysis of business indicator changes is generally carried out directly through the indicator dimension drill-down attribution, indicator topology layer attribution, indicator lineage attribution, etc. However, due to objective problems such as planning, interaction, and logical differences, the entrances of different attribution analysis tools need to be scattered in different subsystems, that is, the operation personnel need to be familiar with the product matrix in order to accurately deploy the corresponding attribution analysis tool to the corresponding subsystem; and, before using the attribution analysis tool, the corresponding query configuration needs to be performed in advance. Since the query configuration method of each attribution analysis tool is different, the operation personnel also need to have a certain understanding of each attribution, and the operation personnel need to spend a lot of time to complete the query configuration of each attribution analysis tool; after obtaining the corresponding analysis results through the attribution analysis tool, the operation personnel are also required to conduct a comprehensive analysis of the analysis results obtained by each different attribution analysis tool to determine the final attribution conclusion. It can be seen that the existing attribution analysis method for business indicator changes is not only high in threshold but also low in efficiency. Summary of the invention

[0004] In view of this, the present application provides an indicator fluctuation attribution analysis method, system, electronic device and storage medium integrating a large language model, with the purpose of lowering the user's usage threshold and improving the efficiency of indicator attribution analysis.

[0005] In a first aspect, the present application provides an indicator fluctuation attribution analysis method integrating a large language model, the method comprising:

[0006] Determine a corresponding to-be-attributed indicator according to the user question sent by the user, and obtain a preset first text template and attribution dimension metadata corresponding to the to-be-attributed indicator;

[0007] Generate a semantic analysis request according to the user question, the first text template, and the attribution dimension metadata of the indicator to be attributed;

[0008] Processing the semantic analysis request through a large language model to obtain a DSL object, and generating a corresponding attribution analysis request according to the DSL object;

[0009] Determine at least one target attribution analysis tool according to the user question, and map the attribution analysis request to each of the target attribution analysis tools;

[0010] Performing attribution analysis according to the attribution analysis request through each of the target attribution analyses to obtain at least one attribution analysis result;

[0011] A result summary request is generated according to a preset second text template, the user question and at least one of the attribution analysis results, and an attribution result summary of the indicator to be attributed is output according to the result summary request through the large language model.

[0012] Optionally, before generating a corresponding attribution analysis request according to the DSL object, the method further includes:

[0013] Verify the DSL object using preset specific business rules; wherein the specific business rules include DSL definition information and preset required parameters and dimension enumeration values ​​thereof;

[0014] If the DSL object verification passes, the step of generating a corresponding attribution analysis request according to the DSL object is performed.

[0015] Optionally, determining at least one target attribution analysis tool according to the user question, and mapping the attribution analysis request to the target attribution analysis tool includes:

[0016] Determining the user intention of the user according to the user question;

[0017] Determine whether the user intent is empty;

[0018] If the user intent is not empty, determining at least one target attribution analysis tool from various pre-set attribution analysis tools according to the intent enumeration and the indicator name indicated in the user intent;

[0019] The attribution analysis request is routed to the target attribution analysis tool.

[0020] Optionally, the method further includes:

[0021] If the user intent is empty, each pre-set attribution analysis tool is determined as a target attribution analysis tool.

[0022] Optionally, the attribution analysis request includes at least a dimension screening condition of the indicator to be attributed, a current time interval and a concurrent time interval of the indicator to be attributed, and;

[0023] The attribution analysis is performed by each target attribution analysis tool according to the attribution analysis request to obtain a corresponding attribution analysis result, including:

[0024] Attribution indicator data corresponding to the indicator to be attributed according to the dimension screening condition of the indicator to be attributed, the current time interval and the same period time interval of the indicator to be attributed;

[0025] According to the screening condition of each of the target attribution analysis tools, the target attribution indicator data satisfying each of the target attribution analysis tools are screened out from the attribution indicator data;

[0026] Attribution analysis is performed by each target attribution analysis tool according to corresponding target attribution indicator data to obtain an attribution analysis result.

[0027] Optionally, the screening conditions of the target attribution analysis tool include indicator fluctuation information, dimension influence coefficient and dimension item fluctuation contribution rate corresponding to the target attribution analysis tool;

[0028] The step of screening out target attribution indicator data satisfying each target attribution analysis tool from the attribution indicator data includes:

[0029] For each of the target attribution analysis tools, obtain the indicator fluctuation information, dimension influence coefficient and dimension item fluctuation contribution rate corresponding to the target attribution analysis tool;

[0030] According to the indicator fluctuation information, dimension influence coefficient and dimension item fluctuation contribution rate corresponding to each of the target attribution analysis tools, the target attribution indicator data satisfying each of the target attribution analysis tools is obtained from the attribution indicator data.

[0031] A second aspect of the present application provides an indicator fluctuation attribution analysis system integrating a large language model, the system comprising:

[0032] A user question receiving unit, configured to determine a corresponding to-be-attributed indicator according to the user question sent by the user, and obtain a preset first text template and attribution dimension metadata corresponding to the to-be-attributed indicator;

[0033] A semantic analysis unit, configured to generate a semantic analysis request according to the user question, the first text template and the attribution dimension metadata of the indicator to be attributed; process the semantic analysis request through a large language model to obtain a DSL object; generate a corresponding attribution analysis request according to the DSL object; determine at least one target attribution analysis tool according to the user question, and map the attribution analysis request to each of the target attribution analysis tools;

[0034] an attribution analysis unit, configured to perform attribution analysis according to the attribution analysis request through each of the target attribution analyses, and obtain at least one attribution analysis result;

[0035] An attribution result summary unit is used to generate a result summary request according to a preset second text template, the user question and at least one of the attribution analysis results, and output the attribution result summary of the indicator to be attributed according to the result summary request through the large language model.

[0036] Optionally, the semantic analysis unit is further used for:

[0037] The DSL object is verified using preset specific business rules; wherein the specific business rules include DSL definition information and preset mandatory parameters and dimension enumeration values ​​thereof; if the DSL object passes the verification, the step of generating a corresponding attribution analysis request according to the DSL object is executed.

[0038] The third aspect of the present application provides an electronic device, comprising: a processor and a memory, wherein the processor and the memory are connected via a communication bus; wherein the processor is used to call and execute a program stored in the memory; and the memory is used to store a program, wherein the program is used to implement the indicator fluctuation attribution analysis method integrating a large language model as provided in the first aspect of the present application.

[0039] The fourth aspect of the present application provides a computer-readable storage medium, in which computer-executable instructions are stored, and the computer-executable instructions are used to execute the indicator fluctuation attribution analysis method integrating a large language model as provided in the first aspect of the present application.

[0040] The present application provides an indicator fluctuation attribution analysis method, system, electronic device and storage medium integrating a large language model. When receiving a user question sent by a user, the to-be-attributed indicator that needs to be attribution analyzed can be determined according to the user question, and at the same time, the attribution dimension metadata corresponding to the preset first text and the to-be-attributed indicator can be obtained, and a semantic analysis request that meets the input of the large language model can be generated according to the user question, the first text template and the attribution dimension metadata of the to-be-attributed indicator, so as to utilize the information extraction capability of the large language model to generate a DSL object that can be used by a downstream attribution analysis tool according to the input semantic analysis request, without the need for operating personnel to spend a lot of time to complete the query configuration of each attribution analysis tool, thereby lowering the threshold for using the attribution analysis tool; the DSL object is encapsulated into an attribution analysis request, and at least one target attribution analysis tool is determined according to the user question, so as to realize the selection according to the user's needs. The corresponding attribution analysis tool is selected to realize the entrance integration of different scenarios of indicator fluctuation attribution analysis, and the attribution analysis request is mapped to at least one determined target attribution analysis tool, so that the target attribution analysis tool performs attribution analysis according to the mapped attribution analysis request to obtain at least one attribution analysis result, and finally the logical reasoning ability of the large language model is used to combine the attribution analysis results obtained according to each target attribution analysis tool, the user's question and the result summary request generated by the preset second text template to perform further heuristic analysis, and output the attribution result summary of the indicator to be attributed, that is, the summary and text generation capabilities of the large language model are combined to finally summarize the various attribution analysis results obtained, without the need for operators to conduct a comprehensive analysis of the results obtained by different attribution analysis tools, which can not only improve the efficiency of attribution analysis but also reduce the dependence on operators and improve the accuracy of attribution analysis. BRIEF DESCRIPTION OF THE DRAWINGS

[0041] In order to more clearly illustrate the embodiments of the present application or the technical solutions in the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are only embodiments of the present application. For ordinary technicians in this field, other drawings can be obtained based on the provided drawings without paying any creative work.

[0042] Figure 1 A flow chart of an indicator fluctuation attribution analysis method integrating a large language model provided in an embodiment of the present application;

[0043] Figure 2 A schematic diagram of the structure of an indicator fluctuation attribution analysis system integrating a large language model provided in an embodiment of the present application;

[0044] Figure 3 A schematic diagram of the structure of an electronic device provided in an embodiment of the present application. DETAILED DESCRIPTION

[0045] The following will be combined with the drawings in the embodiments of the present application to clearly and completely describe the technical solutions in the embodiments of the present application. Obviously, the described embodiments are only part of the embodiments of the present application, not all of the embodiments. Based on the embodiments in the present application, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of this application.

[0046] In this application, relational terms such as first and second, etc. are used only to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Moreover, the terms "comprise", "include" or any other variants thereof are intended to cover non-exclusive inclusion, so that a process, method, article or device including a series of elements includes not only those elements, but also other elements not explicitly listed, or also includes elements inherent to such process, method, article or device. In the absence of further restrictions, an element defined by the statement "comprises a ..." does not exclude the presence of other identical elements in the process, method, article or device including the element.

[0047] From the above background technology, it can be seen that the existing attribution analysis method for business indicator changes not only has a high threshold for use but also has low efficiency.

[0048] In order to solve the above problems, in the existing technology, many large language models such as ChatGPT (hereinafter referred to as large models) can be used to perform attribution analysis on changes in business indicators. However, it is found through practice that, firstly, due to the inaccurate understanding of internal business terms by large models, it is difficult to obtain satisfactory attribution results by relying solely on large models to complete the corresponding attribution analysis; secondly, accurate numerical calculation is a weak point in large models, and the attribution analysis process generally involves a large amount of numerical calculations, which increases the risk of errors in the attribution results obtained through large models.

[0049] Therefore, the present application provides an indicator fluctuation attribution analysis method, system, electronic device and storage medium integrating a large language model, which determines the to-be-attributed indicator that needs to be attribution analyzed according to the received user question, obtains the attribution dimension metadata corresponding to the preset first text and the to-be-attributed indicator, and generates a semantic analysis request that satisfies the input of the large language model according to the user question, the first text template and the attribution dimension metadata of the to-be-attributed indicator, so as to utilize the information extraction capability of the large language model to generate a DSL object that can be used by the downstream attribution analysis tool according to the input semantic analysis request, without the need for operators to spend a lot of time to complete the query configuration of each attribution analysis tool, thereby lowering the threshold for using the attribution analysis tool; encapsulates the DSL object into an attribution analysis request, and determines at least one target attribution analysis tool according to the user question, so as to select the corresponding attribution analysis tool according to the user's needs, realize the entry integration of different scenarios of indicator fluctuation attribution analysis, and convert the attribution analysis request into a DSL object. Mapped to at least one determined target attribution analysis tool, the target attribution analysis tool performs attribution analysis according to the mapped attribution analysis request, obtains at least one attribution analysis result, and finally utilizes the logical reasoning ability of the large language model in combination with the attribution analysis results obtained by each target attribution analysis tool, the user's question and the result summary request generated by the preset second text template, to perform further heuristic analysis, and output the attribution result summary of the indicator to be attributed, that is, combining the summary and text generation capabilities of the large language model, the obtained attribution analysis results are finally summarized, and there is no need for the operating personnel to conduct a comprehensive analysis of the results obtained by each different attribution analysis tool, which can not only improve the efficiency of the attribution analysis but also reduce the dependence on the operating personnel; and the present application completes the corresponding attribution analysis through each target attribution analysis tool in combination with the large language model, and can also solve the risk of errors caused by only performing attribution analysis through the large language model in the prior art, that is, improve the accuracy of the attribution analysis.

[0050] See also Figure 1 , shows a flow chart of an indicator fluctuation attribution analysis method integrating a large language model provided in an embodiment of the present application, the method specifically comprising the following steps:

[0051] S101: Determine a corresponding indicator to be attributed according to a user question sent by a user, and obtain a preset first text template and attribution dimension metadata corresponding to the indicator to be attributed.

[0052] In an embodiment of the present application, a corresponding first text template can be pre-set, wherein the first text template can be a predefined structured text; after receiving a user question, the corresponding indicator to be attributed can be determined according to the user question, and the pre-set text template and the attribution dimension metadata corresponding to the indicator to be attributed can be obtained, so that the user question and the attribution dimension metadata corresponding to the indicator to be attributed can be filled in the first text template subsequently, so as to assemble a semantic analysis request that meets the input of the large language model.

[0053] In some embodiments, the preset first text template may be a first prompt template, wherein the design structure of the first prompt template may be: "return legal JSON:${DSL} in the corresponding format, the identifiable dimension range includes ${dimension set}, wherein the dimension range may include dimension A, dimension B, ... and so on, correspondingly, dimension A is ${dimension definition}, that is, it can only be selected from ${dimension value range}, dimension B..., and user questions must be processed strictly in accordance with the above rules and reference examples, and the corresponding results must be output in the JSON:${DSL} format at the beginning, and the output results must be in the correct JSON format". The reference example is as follows:

[0054] Q:${Sample Question 1}, A:${Sample Answer 1},

[0055] Q:${Sample Question 2}, A:${Sample Answer 2},

[0056] .......

[0057] It should be noted that adding corresponding reference examples (example questions and example answers (model output results)) to the first text template can enable the large language model to output results in the required format based on the reference examples, thereby improving the output quality of the large language model; specifically, the few-shot prompting method can be adopted to add corresponding reference examples (example questions and example answers (model output results)) to the first prompt template (first text template).

[0058] In an embodiment of the present application, a corresponding dimension set can be pre-set, wherein the dimension set includes a subset of all dimensions of each indicator to be attributed, which is used to limit the scope of each indicator to be attributed; the dimension set includes multiple elements corresponding to each indicator to be attributed, and each element is composed of an indicator business dimension and all its dimension enumeration values, that is, each element contains an indicator business dimension and an optional range of its dimension values.

[0059] For example, if the indicator to be attributed is order volume, then the indicator business dimension corresponding to the indicator to be attributed can be "city", "business line", etc., and the enumeration values ​​of all dimensions corresponding to "city (indicator business dimension)" can be "Shenzhen, Guangzhou, Chongqing, etc.", and the enumeration values ​​of all dimensions corresponding to "business line (indicator business dimension)" can be "freight truck, freight car, etc.".

[0060] In the specific process of executing step S101, after receiving the user question sent by the user, a pre-set first text template can be obtained, and the user question can be analyzed to determine the indicator to be attributed related to the user question, so as to obtain the attribution dimension metadata related to the indicator to be attributed from the pre-set dimension set; wherein, the attribution dimension metadata includes each indicator business dimension corresponding to the indicator to be attributed and all dimension enumeration values ​​thereof.

[0061] For example, if the user question received is "Why has the order volume of freight trucks in Shenzhen decreased compared to last week?", by analyzing the user question, it can be determined that the corresponding indicator to be attributed can be "order volume", and then the attribution dimension metadata related to the indicator to be attributed can be obtained from the pre-set dimension set, that is, the business dimensions of each indicator related to the indicator to be attributed and all its dimension enumeration values ​​can be obtained.

[0062] S102: Generate a semantic analysis request according to the user question, the first text template, and the attribution dimension metadata of the indicator to be attributed.

[0063] In the specific process of executing step S102, after obtaining the user question, the first text template, and the attribution dimension metadata corresponding to the indicator to be attributed, the user question, the first text template, and the attribution dimension metadata corresponding to the attribution indicator can be spliced ​​to obtain a complete semantic analysis request that meets the input of the large language template. The semantic analysis request includes the user question, the first text template, and the attribution dimension metadata corresponding to the indicator to be attributed.

[0064] Specifically, if the first text template is the first prompt template, the user question and the attribution dimension metadata of the indicator to be attributed can be filled into the first prompt template according to the design structure of the first prompt template to obtain a complete semantic analysis request that meets the large language template input.

[0065] S103: Process the semantic analysis request through the large language model to obtain a DSL object, and generate a corresponding attribution analysis request according to the DSL object.

[0066] In the specific process of executing step S103, after obtaining the semantic analysis request, the semantic analysis request is input into the large language model, so that the large language model uses its own information extraction ability to extract the dimension screening conditions of the indicator to be attributed, the current time interval and the same period time interval of the indicator to be attributed according to the user question in the semantic analysis request and the attribution dimension metadata corresponding to the indicator to be attributed, and generates and outputs the JSON format of the domain-specific language (Domain-Specific A Language (DSL) object is generated; the DSL object is generated according to the agreement and the AIP interface parameters provided by the indicator fluctuation attribution analysis system of the large language model are integrated to obtain the corresponding attribution analysis request; wherein the DSL object includes the dimension screening conditions of the indicator to be attributed, the current time interval and the same period time interval of the indicator to be attributed, and the corresponding attribution analysis request at least includes the dimension screening conditions of the indicator to be attributed, the current time interval and the same period time interval of the indicator to be attributed; the dimension screening conditions of the indicator to be attributed include the indicator to be attributed and the indicator restriction conditions; the indicator restriction conditions may include the indicator business dimension corresponding to the indicator to be attributed and all dimension enumeration values ​​thereof;

[0067] For example, the output DSL object may include indicators to be attributed: order volume; city (indicator business dimension): Shenzhen (dimension enumeration value); business line (indicator business dimension): freight car (dimension enumeration value); current time: time: 2024-08-05; same period time: time: 2024-07-29.

[0068] It should be noted that DSL, also known as domain-specific language, refers to a computer language limited to a specific application field. A domain-specific language is a language with very specific goals in design and implementation. It is often used for business modeling to achieve specific goals in specific scenarios and is used by a group of computer programs that share this goal.

[0069] In an embodiment of the present application, after obtaining the corresponding DSL object, in order to avoid abnormal situations when performing attribution analysis using the attribution analysis request generated by the illegal DSL object, the DSL object can be firstly checked for legality before generating the corresponding attribution analysis request based on the DSL object; if the DSL object passes the check, the corresponding attribution analysis request can be generated based on the DSL object; if the DSL object fails the check, the DSL object can be filtered out to avoid abnormal situations when performing attribution analysis using the attribution analysis request generated by the illegal DSL.

[0070] Optionally, the process of checking the legality of the DSL object may specifically include: checking the DSL object using a preset specific business rule; if the DSL object fails the check, filtering the DSL object; if the DSL object passes the check, executing the step of generating a corresponding attribution analysis request according to the DSL object.

[0071] It should be noted that the specific business rule indicates whether the DSL object satisfies the DSL definition information and whether the DSL object includes preset mandatory parameters and dimension values ​​thereof. Specifically, after obtaining the DSL object, it can be determined whether the DSL object satisfies the DSL definition information and whether the DSL object includes preset mandatory parameters and dimension enumeration values ​​thereof. The preset parameter may be a pre-set indicator business dimension, which is not limited in the embodiment of the present application.

[0072] S104: Determine at least one target attribution analysis tool according to the user's question, and map the attribution analysis request to the target attribution analysis tool.

[0073] In the specific process of executing step S104, a corresponding attribution analysis module can be pre-set, wherein the attribution analysis module includes various pre-set attribution analysis tools; after obtaining the corresponding attribution analysis request, the user's question can be analyzed to obtain the user's user intention, and at least one target attribution analysis tool can be determined from the various pre-set attribution analysis tools according to the user intention, and finally the attribution analysis request can be mapped to each target attribution analysis tool.

[0074] It should be noted that the various attribution analysis tools may include indicator dimension drill-down attribution analysis tools, indicator topology level attribution analysis tools, indicator lineage attribution analysis tools, etc., which are not limited in the embodiments of the present application.

[0075] Optionally, the process of determining at least one target attribution analysis tool based on a user question and mapping the attribution analysis request to the target attribution analysis tool may specifically be: determining the user intent of the user based on the user question; determining whether the user intent is empty; if the user intent is not empty, determining at least one target attribution analysis tool from various pre-set attribution analysis tools based on the intent enumeration and indicator name indicated in the user intent; if the user intent is empty, determining each pre-set attribution analysis tool as the target attribution analysis tool; and routing the attribution analysis request to the target attribution analysis tool.

[0076] It should be noted that the intent enumeration can be dimension, topology, blood relationship, etc., and the indicator name can be the number of paired orders, the number of drivers accepting orders, etc., which is not limited in this embodiment of the present application.

[0077] It should also be noted that the various pre-set attribution analysis tools may include an indicator dimension drill-down attribution analysis tool, an indicator topology level attribution analysis tool, and an indicator lineage attribution analysis tool.

[0078] In actual application, users may or may not indicate their intentions in their questions. Therefore, the user questions may be analyzed to determine whether the user intention obtained from the analysis is empty. If it is empty, it means that the user did not indicate his or her intentions in the user questions. In this case, a fallback logic is adopted to determine all pre-set attribution analysis tools as target attribution analysis tools. If it is not empty, it means that the user indicated his or her intentions in the user questions. In this case, the attribution analysis tool corresponding to the intention enumeration indicated in the user intention and its indicator name may be determined as the target attribution analysis tool.

[0079] For example, if the user asks a question such as "Why has the order volume of freight trucks in Shenzhen decreased compared to last week according to the dimension analysis?", then by analyzing the user's question, it can be determined that the user is more concerned about the dimension analysis, that is, it can be determined that the intent enumeration indicated in the user's intent and its indicator name can be the dimension and its indicator name, and then the corresponding target attribution analysis tool can be determined as the indicator dimension drill-down attribution analysis tool; or, if the user asks a question such as "Why has the order volume of freight trucks in Shenzhen decreased compared to last week, use the dimension to drill down?", then by analyzing the user's question, it can be determined that the user has expressed his or her intent (dimension drill-down), that is, it can be determined that the intent enumeration indicated in the user's intent and its indicator name can be the dimension and its indicator name, and then the dimension drill-down attribution analysis tool can be determined as the target attribution analysis tool; or, if the user asks a question such as "Why has the order volume of freight trucks in Shenzhen decreased compared to last week?", then by analyzing the user's question, it can be determined that the user has not expressed his or her intent, that is, the user's intent is empty, and then all the pre-set attribution analysis tools can be determined as the target attribution analysis tools.

[0080] S105: Perform attribution analysis according to the attribution analysis request through each target attribution analysis to obtain at least one attribution analysis result.

[0081] In the process of specifically executing step S105, after determining at least one target attribution analysis tool, attribution indicator data corresponding to the indicator to be attributed can be further obtained according to the attribution analysis request, and the target attribution indicator data corresponding to each target attribution analysis tool can be screened out from the attribution indicator data, so that the target attribution analysis tool mapped with the attribution analysis request can perform attribution analysis according to the corresponding target attribution indicator data to obtain the corresponding attribution analysis result.

[0082] It should be noted that, since different attribution analysis tools focus on different contents, the filtering conditions corresponding to each attribution analysis tool may be pre-set according to the contents that each attribution analysis tool focuses on.

[0083] It is worth mentioning that in the dimension drill-down attribution analysis tool, the focus is mainly on the dimension set and dimension item details of the indicator to be attributed. Therefore, the corresponding indicator waveform information, dimension influence coefficient, and dimension item fluctuation contribution rate and other filtering conditions of the dimension drill-down attribution analysis tool can be set according to the content of the dimension drill-down attribution analysis tool; in the indicator topology level attribution analysis tool, the focus is mainly on the indicators that are topologically superior and subordinate to the indicator to be attributed. Therefore, the corresponding indicator waveform information, dimension influence coefficient, and dimension item fluctuation contribution rate and other filtering conditions of the indicator topology level attribution analysis tool can be set according to the content of the indicator topology level attribution analysis tool; in the indicator lineage attribution analysis tool, the focus is mainly on the data tables, fields, and task information that have a lineage relationship with the indicator to be attributed. Therefore, the corresponding indicator waveform information, dimension influence coefficient, and dimension item fluctuation contribution rate and other filtering conditions of the indicator lineage attribution analysis tool can be set according to the content of the indicator lineage attribution analysis tool. Among them, the indicator waveform information includes the waveform information corresponding to each indicator to be attributed, the dimension impact coefficient includes the impact coefficient corresponding to each indicator business dimension corresponding to each indicator to be attributed, and the dimension item fluctuation contribution rate includes the contribution rate corresponding to each indicator business dimension corresponding to the indicator to be attributed.

[0084] Optionally, the attribution analysis request includes at least the dimension screening conditions of the indicator to be attributed, the current time interval and the concurrent time interval of the indicator to be attributed, and each target attribution analysis tool performs attribution analysis according to the attribution analysis request to obtain the corresponding attribution analysis result. The specific process may be: according to the dimension screening conditions of the indicator to be attributed, the current time interval and the concurrent time interval of the indicator to be attributed, the attribution indicator data corresponding to the indicator to be attributed; according to the screening conditions of each target attribution analysis tool, the target attribution indicator data that meets each target attribution analysis tool is screened out from the attribution indicator data; and each target attribution analysis tool performs attribution analysis according to the corresponding target attribution indicator data to obtain the attribution analysis result.

[0085] In an embodiment of the present application, corresponding first attribution indicator data can be obtained based on the indicator to be attributed indicated in the dimension screening condition of the indicator to be attributed and the current time interval, and corresponding second attribution indicator data can be obtained based on the indicator to be attributed indicated in the dimension screening condition of the indicator to be attributed and the same period time interval, and finally, attribution indicator data of the indicator to be attributed is generated based on the first attribution indicator data and the second attribution indicator data.

[0086] As a preferred method of an embodiment of the present application, the screening conditions of the target attribution analysis tool include the indicator fluctuation information, dimension influence coefficient and dimension item fluctuation contribution rate corresponding to the target attribution analysis tool, and the process of screening out the target attribution indicator data that meets each target attribution analysis tool from the attribution indicator data can specifically be: for each target attribution analysis tool, obtaining the indicator fluctuation information, dimension influence coefficient and dimension item fluctuation contribution rate corresponding to the target attribution analysis tool; according to the indicator fluctuation information, dimension influence coefficient and dimension item fluctuation contribution rate corresponding to each target attribution analysis tool, obtaining the target attribution indicator data that meets each target attribution analysis tool from the attribution indicator data.

[0087] In the actual application of this process, for each target attribution analysis tool, the indicator fluctuation information, dimension influence coefficient and dimension item fluctuation contribution rate corresponding to the target attribution analysis tool are obtained, and the waveform information corresponding to the indicator to be attributed and the influence coefficient and contribution rate corresponding to each indicator business dimension corresponding to the indicator to be attributed indicated in the indicator limitation conditions are determined from the indicator fluctuation information, dimension influence coefficient and dimension item fluctuation contribution rate corresponding to the target attribution analysis tool, so as to obtain the target attribution data that meets the target attribution analysis tool from the attribution indicator data corresponding to the indicator to be attributed according to the waveform information corresponding to the indicator to be attributed and the influence coefficient and contribution rate corresponding to each indicator business dimension corresponding to the indicator to be attributed, so as to filter the attribution indicator data and obtain a topological analysis set (target attribution data) with the same fluctuation direction as the indicator to be attributed and meeting the target attribution analysis tool.

[0088] S106: Generate a result summary request according to a preset second text template, a user question and at least one attribution analysis result, and output an attribution result summary of the to-be-attributed indicator according to the result summary request through a large language model.

[0089] In the embodiments of the present application, the applicant has found that the attribution results output by the large model in the prior art are usually in the form of numerical values, vectors, and matrices, and such semantic features cannot be understood by computer programs and humans. This requires that after obtaining the attribution results, the downstream users must design corresponding display tools for the three semantic features rather than standardized computer voices, so as to convert the obtained attribution results into sentences that can be understood by computer programs and humans; therefore, in order to improve the accuracy of the summary of the obtained attribution results and to achieve the obtained attribution results as sentences that can be understood by computer programs and humans, after obtaining the attribution analysis results of each target attribution analysis tool, the logical reasoning ability of the large language model can be used, combined with the attribution analysis results of each target attribution analysis tool, to conduct further heuristic analysis; specifically, the attribution analysis results of each target attribution analysis tool can be combined with a pre-set second text template and a user question to generate a result summary request that meets the input of the large model, and the result summary request can be sent to the user. The large language model is sent to the large language model, so that the large language model uses its own logical reasoning ability to perform heuristic induction and summary according to the attribution analysis results of each target attribution analysis tool, the user's question and the second text model, and obtains the attribution result summary of the to-be-attributed indicator, without the user having to summarize and analyze the various attribution analysis results obtained, which can not only improve the efficiency of the attribution analysis but also improve the accuracy of the attribution analysis; and, after receiving the result summary request, the large language model can also perform attribution analysis according to the attribution analysis results of each target attribution analysis tool and the user's question in the result summary request according to the requirements in the second text template in the result summary request, and output the attribution result summary obtained by the attribution analysis according to the output format to be input in the second text template, thereby solving the problem in the prior art that after obtaining the attribution result, the downstream user must design the corresponding display tool for the three semantic features rather than the standardized computer voice, so as to realize the conversion of the obtained attribution result into a sentence that can be understood by a computer program and a person.

[0090] It should be noted that the preset second text model may be a prompt template (for the sake of distinction, the prompt template set here is referred to as a second prompt template), wherein the design of the second prompt template may be as follows:

[0091] Summarize and analyze the questions raised by users (user questions) based on the provided attribution data (attribution analysis results of each target attribution analysis tool), and explain the reasons for the data fluctuations of the attribution indicators that users are concerned about according to the user's intention. Note: 1. The main diagnostic dimensions include: the dimensions corresponding to the indicators to be attributed, such as "dimension A", "dimension B", etc.; 2. The given data (attribution analysis results of each target attribution analysis tool) has been sorted from high to low according to "influence"; 3. The dimension data needs to be retained when giving the diagnostic conclusion (attribution result summary); 4. If the actual diagnostic result (attribution result summary) does not match the user's question, it is necessary to analyze it according to the actual fluctuation results. The results need to be returned according to the following format rules, where the format rules can be: 1. Format rule A; 2. Format rule B; 3. Format rule C.

[0092] It should be noted that the format rules can generate text content (attribution result summary) that is easy for computer programs and humans to read and understand for the standardized large language model, including but not limited to the following format rules: 1. The year-on-year fluctuation results of {indicator}; 2. The diagnosis results show that according to the impact ranking, {dimension 1} is the most important dimension affecting the change of {indicator}; 3. In the impact ranking of {dimension 1}, it can be seen that {dimension value A}, {dimension value B}, and {dimension value C} have the greatest impact on the fluctuation of {indicator}. Among them, the indicator can be the indicator to be attributed, and dimension 1 can be the most important indicator business dimension that affects the change of the indicator to be attributed among the various indicator business dimensions of the indicator to be attributed; dimension value A, dimension value B, and dimension value C can be the enumeration values ​​of each dimension corresponding to dimension 1 (indicator business dimension). The above is only the preferred content of a format rule provided in the embodiment of the present application. Regarding the specific content of the format rule, it can be set according to the actual application, and the embodiment of the present application is not limited here.

[0093] It can be seen that after the large language model is structured into a result summary request, it can perform attribution analysis according to the attribution analysis results of each target attribution analysis tool in the result summary request and the user's question according to the requirements in the second text template in the result summary request, and output the attribution result summary obtained by the attribution analysis according to the output format to be input in the second text template, thereby solving the problem in the prior art that after obtaining the attribution result, the downstream user must design the corresponding display tool for three semantic features rather than standardized computer voice, so as to realize the conversion of the obtained attribution result into sentences that can be understood by computer programs and humans.

[0094] The present application provides an indicator fluctuation attribution analysis method integrating a large language model, which determines the to-be-attributed indicator that needs to be analyzed based on the received user question, obtains the attribution dimension metadata corresponding to the preset first text and the to-be-attributed indicator, and generates a semantic analysis request that satisfies the input of the large language model based on the user question, the first text template and the attribution dimension metadata of the to-be-attributed indicator, so as to utilize the information extraction capability of the large language model to generate a DSL object that can be used by a downstream attribution analysis tool based on the input semantic analysis request, without the need for operators to spend a lot of time to complete the query configuration of each attribution analysis tool, thereby lowering the threshold for using the attribution analysis tool; encapsulates the DSL object into an attribution analysis request, and determines at least one target attribution analysis tool based on the user question, so as to select the corresponding attribution analysis tool according to the user's needs, realize the entry integration of different scenarios of indicator fluctuation attribution analysis, and map the attribution analysis request to the determined at least On a target attribution analysis tool, the target attribution analysis tool performs attribution analysis according to the mapped attribution analysis request to obtain at least one attribution analysis result. Finally, the logical reasoning ability of the large language model is used to combine the attribution analysis results obtained by each target attribution analysis tool, the user's question and the result summary request generated by the pre-set second text template to perform further heuristic analysis, and output the attribution result summary of the indicator to be attributed, that is, the summary and text generation capabilities of the large language model are combined to finally summarize the various attribution analysis results obtained, without the need for operators to conduct a comprehensive analysis of the results obtained by different attribution analysis tools, which can not only improve the efficiency of attribution analysis but also reduce dependence on operators; and the present application completes the corresponding attribution analysis through each target attribution analysis tool in combination with the large language model, and can also solve the risk of errors caused by only performing attribution analysis through the large language model in the prior art, that is, improve the accuracy of attribution analysis.

[0095] Based on the above-mentioned indicator fluctuation attribution analysis method integrating a large language model provided in the embodiment of the present application, correspondingly, the embodiment of the present application also provides an indicator fluctuation attribution analysis system integrating a large language model, such as Figure 2 As shown, the indicator fluctuation attribution analysis system integrating the large language model includes a user question receiving unit, a semantic analysis unit, an attribution analysis unit, an attribution result summary unit and a large language model;

[0096] A user question receiving unit, used to determine a corresponding to-be-attributed indicator according to the user question sent by the user, and obtain a preset first text template and attribution dimension metadata corresponding to the to-be-attributed indicator;

[0097] A semantic analysis unit, configured to generate a semantic analysis request according to the user question, the first text template and the attribution dimension metadata of the indicator to be attributed; process the semantic analysis request through a large language model to obtain a DSL object; generate a corresponding attribution analysis request according to the DSL object; determine at least one target attribution analysis tool according to the user question, and map the attribution analysis request to each target attribution analysis tool;

[0098] an attribution analysis unit, configured to perform attribution analysis according to the attribution analysis request through each target attribution analysis, and obtain at least one attribution analysis result;

[0099] The attribution result summary unit is used to generate a result summary request according to a preset second text template, a user question and at least one attribution analysis result, and output an attribution result summary of the indicator to be attributed according to the result summary request through a large language model.

[0100] The specific principles and execution processes of each unit in the indicator fluctuation attribution analysis system for integrating a large language model disclosed in the above-mentioned embodiment of the present application are the same as the indicator fluctuation attribution analysis method for integrating a large language model disclosed in the above-mentioned embodiment of the present application. Please refer to the corresponding parts of the indicator fluctuation attribution analysis method for integrating a large language model disclosed in the above-mentioned embodiment of the present application, and no further details will be given here.

[0101] The present application provides an indicator fluctuation attribution analysis system integrating a large language model, which determines the to-be-attributed indicator that needs to be analyzed based on the received user question, obtains the attribution dimension metadata corresponding to the preset first text and the to-be-attributed indicator, and generates a semantic analysis request that satisfies the input of the large language model based on the user question, the first text template and the attribution dimension metadata of the to-be-attributed indicator, so as to utilize the information extraction capability of the large language model to generate a DSL object that can be used by a downstream attribution analysis tool based on the input semantic analysis request, without the need for operators to spend a lot of time to complete the query configuration of each attribution analysis tool, thereby lowering the threshold for using the attribution analysis tool; encapsulates the DSL object into an attribution analysis request, and determines at least one target attribution analysis tool based on the user question, so as to select the corresponding attribution analysis tool according to the user's needs, realize the entry integration of different scenarios of indicator fluctuation attribution analysis, and map the attribution analysis request to the determined at least On a target attribution analysis tool, the target attribution analysis tool performs attribution analysis according to the mapped attribution analysis request to obtain at least one attribution analysis result. Finally, the logical reasoning ability of the large language model is used to combine the attribution analysis results obtained by each target attribution analysis tool, the user's question and the result summary request generated by the pre-set second text template to perform further heuristic analysis, and output the attribution result summary of the indicator to be attributed, that is, the summary and text generation capabilities of the large language model are combined to finally summarize the various attribution analysis results obtained, without the need for operators to conduct a comprehensive analysis of the results obtained by different attribution analysis tools, which can not only improve the efficiency of attribution analysis but also reduce dependence on operators; and the present application completes the corresponding attribution analysis through each target attribution analysis tool in combination with the large language model, and can also solve the risk of errors caused by only performing attribution analysis through the large language model in the prior art, that is, improve the accuracy of attribution analysis.

[0102] Optionally, the semantic analysis unit provided in the embodiment of the present application is further used for:

[0103] The DSL object is verified using pre-set specific business rules, wherein the specific business rules include DSL definition information and preset mandatory parameters and dimension enumeration values ​​thereof; if the DSL object verification passes, the step of generating a corresponding attribution analysis request according to the DSL object is executed.

[0104] Optionally, at least one target attribution analysis tool is determined according to the user's question, and the attribution analysis request is mapped to a semantic analysis unit on the target attribution analysis tool, specifically for:

[0105] Determine the user intent of the user according to the user's question; determine whether the user intent is empty; if the user intent is not empty, determine at least one target attribution analysis tool from various pre-set attribution analysis tools according to the intent enumeration and indicator name indicated in the user intent; and route the attribution analysis request to the target attribution analysis tool.

[0106] Optionally, the semantic analysis unit provided in the embodiment of the present application is also used to: if the user intention is empty, determine each pre-set attribution analysis tool as a target attribution analysis tool.

[0107] Optionally, the attribution analysis request includes at least the dimension screening condition of the indicator to be attributed, the current time interval and the same period time interval of the indicator to be attributed, and each target attribution analysis tool performs attribution analysis according to the attribution analysis request to obtain an attribution analysis unit of a corresponding attribution analysis result, which is specifically used for:

[0108] According to the dimension screening conditions of the indicator to be attributed, the current time interval and the same period time interval of the indicator to be attributed, the attribution indicator data corresponding to the indicator to be attributed; according to the screening conditions of each target attribution analysis tool, the target attribution indicator data that meets each target attribution analysis tool is screened out from the attribution indicator data; attribution analysis is performed according to the corresponding target attribution indicator data through each target attribution analysis tool to obtain the attribution analysis result.

[0109] Optionally, the screening conditions of the target attribution analysis tool include indicator fluctuation information, dimension influence coefficient, and dimension item fluctuation contribution rate corresponding to the target attribution analysis tool, and the attribution analysis units that meet the target attribution indicator data of each target attribution analysis tool are screened out from the attribution indicator data, specifically used for:

[0110] For each target attribution analysis tool, obtain the indicator fluctuation information, dimension influence coefficient and dimension item fluctuation contribution rate corresponding to the target attribution analysis tool; according to the indicator fluctuation information, dimension influence coefficient and dimension item fluctuation contribution rate corresponding to each target attribution analysis tool, obtain the target attribution indicator data that meets each target attribution analysis tool from the attribution indicator data.

[0111] The present application embodiment provides an electronic device, such as Figure 3 As shown, the electronic device includes a processor 301 and a memory 302, the memory 302 is used to store program code and data for the indicator fluctuation attribution analysis of the device integrating a large language model, and the processor 301 is used to call the program instructions in the memory to execute the steps shown in the indicator fluctuation attribution analysis method integrating a large language model in the above embodiment.

[0112] An embodiment of the present application provides a storage medium, which includes a stored program, wherein when the program is running, the device where the storage medium is located is controlled to execute the indicator fluctuation attribution analysis method for integrating a large language model shown in the above embodiment.

[0113] Each embodiment in this specification is described in a progressive manner, and the same or similar parts between the embodiments can be referred to each other, and each embodiment focuses on the differences from other embodiments. In particular, for the system or system embodiment, since it is basically similar to the method embodiment, the description is relatively simple, and the relevant parts can be referred to the partial description of the method embodiment. The system and system embodiments described above are merely schematic, in which the units described as separate components may or may not be physically separated, and the components displayed as units may or may not be physical units, that is, they may be located in one place, or they may be distributed on multiple network units. Some or all of the modules can be selected according to actual needs to achieve the purpose of the scheme of this embodiment. Ordinary technicians in this field can understand and implement it without paying creative labor.

[0114] Professionals may further appreciate that the units and algorithm steps of each example described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, computer software, or a combination of the two. In order to clearly illustrate the interchangeability of hardware and software, the composition and steps of each example have been generally described in the above description according to function. Whether these functions are performed in hardware or software depends on the specific application and design constraints of the technical solution. Professionals and technicians may use different methods to implement the described functions for each specific application, but such implementation should not be considered to be beyond the scope of this application.

[0115] The above description of the disclosed embodiments enables those skilled in the art to implement or use the present application. Various modifications to these embodiments will be apparent to those skilled in the art, and the general principles defined herein may be implemented in other embodiments without departing from the spirit or scope of the present application. Therefore, the present application will not be limited to the embodiments shown herein, but will conform to the widest scope consistent with the principles and novel features disclosed herein.

[0116] The above are only preferred implementations of the present application. It should be pointed out that for ordinary technicians in this technical field, several improvements and modifications can be made without departing from the principles of the present application. These improvements and modifications should also be regarded as the scope of protection of the present application.

Claims

1. A method for attribution analysis of index fluctuations integrating a large language model, characterized in that: The method comprises: Determine a corresponding to-be-attributed indicator according to the user question sent by the user, and obtain a preset first text template and attribution dimension metadata corresponding to the to-be-attributed indicator; Generate a semantic analysis request according to the user question, the first text template, and the attribution dimension metadata of the indicator to be attributed; Processing the semantic analysis request through a large language model to obtain a DSL object, and generating a corresponding attribution analysis request according to the DSL object; Determine at least one target attribution analysis tool according to the user question, and map the attribution analysis request to each of the target attribution analysis tools; Performing attribution analysis according to the attribution analysis request through each of the target attribution analyses to obtain at least one attribution analysis result; A result summary request is generated according to a preset second text template, the user question and at least one of the attribution analysis results, and an attribution result summary of the indicator to be attributed is output according to the result summary request through the large language model.

2. The method according to claim 1, characterized in that Before generating a corresponding attribution analysis request according to the DSL object, the method further includes: Verify the DSL object using preset specific business rules; wherein the specific business rules include DSL definition information and preset required parameters and dimension enumeration values ​​thereof; If the DSL object verification passes, the step of generating a corresponding attribution analysis request according to the DSL object is performed.

3. The method according to claim 1, characterized in that Determining at least one target attribution analysis tool according to the user question, and mapping the attribution analysis request to the target attribution analysis tool, including: Determining the user intention of the user according to the user question; Determine whether the user intent is empty; If the user intent is not empty, determining at least one target attribution analysis tool from various pre-set attribution analysis tools according to the intent enumeration and the indicator name indicated in the user intent; The attribution analysis request is routed to the target attribution analysis tool.

4. The method according to claim 3, characterized in that The method further comprises: If the user intent is empty, each pre-set attribution analysis tool is determined as a target attribution analysis tool.

5. The method according to claim 1, characterized in that The attribution analysis request at least includes the dimension screening condition of the indicator to be attributed, the current time interval and the same period time interval of the indicator to be attributed, and the; The attribution analysis is performed by each target attribution analysis tool according to the attribution analysis request to obtain a corresponding attribution analysis result, including: Attribution indicator data corresponding to the indicator to be attributed according to the dimension screening condition of the indicator to be attributed, the current time interval and the same period time interval of the indicator to be attributed; According to the screening condition of each of the target attribution analysis tools, the target attribution indicator data satisfying each of the target attribution analysis tools are screened out from the attribution indicator data; Attribution analysis is performed by each target attribution analysis tool according to corresponding target attribution indicator data to obtain an attribution analysis result.

6. The method according to claim 5, characterized in that The screening conditions of the target attribution analysis tool include the indicator fluctuation information, dimension influence coefficient and dimension item fluctuation contribution rate corresponding to the target attribution analysis tool; The step of screening out target attribution indicator data satisfying each target attribution analysis tool from the attribution indicator data includes: For each of the target attribution analysis tools, obtain the indicator fluctuation information, dimension influence coefficient and dimension item fluctuation contribution rate corresponding to the target attribution analysis tool; According to the indicator fluctuation information, dimension influence coefficient and dimension item fluctuation contribution rate corresponding to each of the target attribution analysis tools, the target attribution indicator data satisfying each of the target attribution analysis tools is obtained from the attribution indicator data.

7. An indicator fluctuation attribution analysis system integrating a large language model, characterized in that: The system comprises: A user question receiving unit, configured to determine a corresponding to-be-attributed indicator according to the user question sent by the user, and obtain a preset first text template and attribution dimension metadata corresponding to the to-be-attributed indicator; A semantic analysis unit, configured to generate a semantic analysis request according to the user question, the first text template and the attribution dimension metadata of the indicator to be attributed; process the semantic analysis request through a large language model to obtain a DSL object; generate a corresponding attribution analysis request according to the DSL object; determine at least one target attribution analysis tool according to the user question, and map the attribution analysis request to each of the target attribution analysis tools; an attribution analysis unit, configured to perform attribution analysis according to the attribution analysis request through each of the target attribution analyses, and obtain at least one attribution analysis result; An attribution result summary unit is used to generate a result summary request according to a preset second text template, the user question and at least one of the attribution analysis results, and output the attribution result summary of the indicator to be attributed according to the result summary request through the large language model.

8. The system according to claim 7, characterized in that The semantic analysis unit is also used for: The DSL object is verified using preset specific business rules; wherein the specific business rules include DSL definition information and preset mandatory parameters and dimension enumeration values ​​thereof; if the DSL object passes the verification, the step of generating a corresponding attribution analysis request according to the DSL object is executed.

9. An electronic device, characterized in that: include: A processor and a memory, wherein the processor and the memory are connected via a communication bus; wherein the processor is used to call and execute a program stored in the memory; The memory is used to store a program, and the program is used to implement the indicator fluctuation attribution analysis method integrating a large language model as described in any one of claims 1-6.

10. A computer-readable storage medium, characterized in that: The computer-readable storage medium stores computer-executable instructions, and the computer-executable instructions are used to execute the indicator fluctuation attribution analysis method integrating a large language model as described in any one of claims 1 to 6.

Citation Information

Patent Citations

  • Attribution analysis method and device, computer equipment and storage medium

    CN115392799A

  • Regional and invasive enterprise risk early warning attribution analysis method, storage medium and related equipment

    CN116777604A

  • Knowledge graph enhanced power distribution network fault attribution analysis method based on large language model

    CN117312531A

  • Attribution analysis method, device and equipment and readable storage medium

    CN117764081A

  • Medical question and answer method and device based on large language model

    CN118133971A

Cited By

  • Index transaction attribution analysis method and device, computer equipment and storage medium

    CN119378689A

  • Method, device, computer equipment and storage medium for attributing abnormal movement of indicators

    CN119378689B

  • Method and device for data attribution, electronic equipment and program product

    CN119670897A

  • Method, apparatus, electronic device, and program product for data attribution

    CN119670897B

  • Tool constraint-based large model process formula increment difference modeling method and system

    CN122308280A