Operation analysis report generation method and device, electronic equipment and storage medium

Abstract

The invention discloses an operation analysis report generation method and device, electronic equipment and a storage medium, and the method comprises the steps: obtaining user attention information and original data of a to-be-analyzed production operation project, and obtaining a target dimension index based on the original data; according to the user attention information, background purpose analysis data is generated based on the target dimension index by using a preset large language model; according to the target dimension index, carrying out drilling analysis on the original data based on the time trend and the code value to obtain a numerical analysis result; and based on the background objective analysis data and the numerical analysis result, performing data integration by using a preset large model to obtain an experience analysis report. According to the method, the operation data is automatically processed and analyzed through the large language model, and the operation analysis report is obtained by utilizing the text understanding and generating capability of the large language model. The method can efficiently and accurately realize operation analysis report generation, and can be widely applied to the technical field of data processing.

Description

Technical Field

[0001] The present invention relates to the field of data processing technology, and in particular to a method, device, electronic device and storage medium for generating a business analysis report. Background Art

[0002] With the rapid development of information technology, data analysis has become an indispensable part of the enterprise decision-making process. Traditional data analysis methods rely on manual operations, which are time-consuming and prone to errors. In recent years, with the development of big data technology, automated data analysis tools have emerged, improving the efficiency and accuracy of data processing. Especially in the field of artificial intelligence, the emergence of large language models (LLMs) marks a major advancement in natural language processing (NLP) technology. These models can understand and generate natural language text through deep learning technology, which has greatly promoted the development of machine reading comprehension, text summarization, language translation, and question-answering systems. However, despite the remarkable achievements of large language models in text understanding, existing data analysis systems still have limitations in processing complex data sets and generating in-depth analysis reports. These systems often lack the ability to deeply understand data and generate natural language, resulting in inaccurate or lack of depth in the generated reports. Mainly reflected in the limited ability to understand data: traditional data analysis tools are usually based on preset rules and algorithms, lacking a deep understanding of the meaning behind the data; report generation is not flexible enough: existing report generation systems often adopt a templated approach, lacking personalization and creativity; insufficient interactivity: the interaction between users and the system is usually mechanical, lacking the fluency and intuitiveness of natural language interaction. Summary of the invention

[0003] The present invention proposes a method, device, electronic device and storage medium for generating a business analysis report, aiming to solve at least one of the technical problems in the related art to a certain extent. The embodiments of the present invention can generate a business analysis report efficiently and accurately.

[0004] In one aspect, an embodiment of the present invention provides a method for generating a business analysis report, comprising:

[0005] Obtain the original data of user-focused information and production and operation items to be analyzed, and extract target dimension indicators based on the original data;

[0006] Based on the user's attention information, the preset large language model is used to generate background purpose analysis data based on the target dimension indicators;

[0007] According to the target dimension indicators, drill down analysis is performed on the original data based on time trends and code values ​​to obtain numerical analysis results;

[0008] Based on the background purpose, the data and numerical analysis results are analyzed, and the preset large model is used to integrate the data to obtain an empirical analysis report.

[0009] Optionally, the target dimension indicator includes a dimension classification result and an indicator classification result; extracting the target dimension indicator based on the original data includes the following steps:

[0010] Obtaining the data table to be analyzed as original data from the production and operation analysis database corresponding to the production and operation project;

[0011] Get the table structure and field information of the data table;

[0012] Based on the table structure and field information, the pre-trained language representation model is used to pre-classify dimensions and indicators to obtain dimension classification results and indicator classification results.

[0013] Optionally, the method further comprises the following steps:

[0014] Based on the preset dimension mapping table, the semantic judgment model is used to perform a first verification match on the dimension classification result output by the language representation model;

[0015] Among them, the expression of the first check match is:

[0016]

[0017] In the formula, c i Represents the i-th dimension information in the dimension mapping table, where i is the index value; C dim Indicates the dimension information contained in the preset dimension mapping table; d k represents the dimensional classification result output by the language representation model; V(·) represents the semantic judgment model, which is used to perform deep dimensional matching on the semantic dimension of the dimensional classification result output by the language representation model from the preset dimensional mapping table; η represents the preset judgment threshold;

[0018] Update the dimension classification result based on the result of the first verification match;

[0019] When the result of the first verification match is empty, obtain the historical user attention information;

[0020] Sequentially concatenate historical user attention information and user attention information, and then use language representation model, dimension mapping table and semantic judgment model to perform joint dimension information matching;

[0021] The expression for joint dimension information matching is:

[0022]

[0023] In the formula, ∪ represents the concatenation symbol, which is used to sequentially concatenate historical user attention information and user attention information; q k represents the kth historical user attention information, q o represents the user's attention information, a represents the total number of cumulative splicing; Bert(·) represents the language representation model;

[0024] Update the dimension classification result based on the result of the joint dimension information matching.

[0025] Optionally, the method further comprises the following steps:

[0026] Based on the preset indicator mapping table, the semantic judgment model is used to perform a second verification match on the indicator classification result output by the language representation model;

[0027] The expression for the second check match is:

[0028] b idx ={t i |V(d t ,t i )>η,t i ∈C index ,}

[0029] Where, t i Indicates the i-th indicator information in the indicator mapping table, where i is the index value; C index Indicates the indicator information contained in the preset indicator mapping table; d t represents the indicator classification result output by the language representation model; V(·) represents the semantic judgment model, which is used to perform deep dimension matching on the semantic dimension of the indicator classification result output by the language representation model from the preset indicator mapping table;

[0030] Obtaining a fuzzy indicator set according to the result of the second verification matching;

[0031] Based on the fuzzy indicator set, data table, field information and user attention information, the language representation model is used to perform secondary reasoning on the data table;

[0032] Update the indicator classification results based on the results of secondary reasoning.

[0033] Optionally, according to the user attention information, using a preset large language model to generate background purpose analysis data based on the target dimension index, including the following steps:

[0034] Based on a preset first prompt word template, generating first prompt word information according to user attention information and target dimension indicators;

[0035] The first prompt word information is input into a preset large language model for reasoning to generate background purpose analysis data.

[0036] Optionally, the original data represents a data table corresponding to the production and operation project; according to the target dimension indicator, the original data is drilled down for analysis based on the time trend and the code value to obtain a numerical analysis result, including the following steps:

[0037] Obtain the numerical data corresponding to the target dimension indicator from the data table; all data values ​​in the numerical data are arranged in time series based on their corresponding time points;

[0038] Perform month-on-month and comparative statistics on numerical data based on a preset time interval to obtain time trend analysis data;

[0039] Perform incremental statistics and increase statistics on numerical data to obtain code value analysis data;

[0040] According to the time trend analysis data and code value analysis data, single-dimensional drill-down statistical analysis and multi-dimensional drill-down statistical analysis are performed based on the target dimension indicators to obtain numerical analysis results.

[0041] Optionally, based on the background purpose analysis data and numerical analysis results, a preset large model is used to integrate the data to obtain an empirical analysis report, including the following steps:

[0042] Based on the preset second prompt word template, the second prompt word information is generated according to the background purpose analysis data and the numerical analysis result;

[0043] The second prompt word information is input into the preset large model for data integration to obtain an experience analysis report.

[0044] On the other hand, an embodiment of the present invention provides a business analysis report generating device, including:

[0045] The source data module is used to obtain the original data of the information that users are concerned about and the production and operation items to be analyzed, and extract the target dimension indicators based on the original data;

[0046] The semantic understanding module is used to generate background purpose analysis data based on the target dimension indicators according to the information that users pay attention to using the preset large language model;

[0047] The data analysis module is used to drill down and analyze the original data based on the target dimension indicators, time trends and code values ​​to obtain numerical analysis results;

[0048] The report generation module is used to analyze data and numerical analysis results based on background purposes, and use the preset large model to integrate data to obtain an empirical analysis report.

[0049] Optionally, the target dimension indicator includes a dimension classification result and an indicator classification result; the source data module is used to perform the following operations:

[0050] Obtaining the data table to be analyzed as original data from the production and operation analysis database corresponding to the production and operation project;

[0051] Get the table structure and field information of the data table;

[0052] Based on the table structure and field information, the pre-trained language representation model is used to pre-classify dimensions and indicators to obtain dimension classification results and indicator classification results.

[0053] Optionally, the device further includes a first module, configured to perform the following operations:

[0054] Based on the preset dimension mapping table, the semantic judgment model is used to perform a first verification match on the dimension classification result output by the language representation model;

[0055] Among them, the expression of the first check match is:

[0056]

[0057] In the formula, c i Represents the i-th dimension information in the dimension mapping table, where i is the index value; C dim Indicates the dimension information contained in the preset dimension mapping table; d k represents the dimensional classification result output by the language representation model; V(·) represents the semantic judgment model, which is used to perform deep dimensional matching on the semantic dimension of the dimensional classification result output by the language representation model from the preset dimensional mapping table; η represents the preset judgment threshold;

[0058] Update the dimension classification result based on the result of the first verification match;

[0059] When the result of the first verification match is empty, obtain the historical user attention information;

[0060] Sequentially concatenate historical user attention information and user attention information, and then use language representation model, dimension mapping table and semantic judgment model to perform joint dimension information matching;

[0061] The expression for joint dimension information matching is:

[0062]

[0063] In the formula, ∪ represents the concatenation symbol, which is used to sequentially concatenate historical user attention information and user attention information; q k represents the kth historical user attention information, q o represents the user's attention information, a represents the total number of cumulative splicing; Bert(·) represents the language representation model;

[0064] Update the dimension classification result based on the result of the joint dimension information matching.

[0065] Optionally, the device further includes a second module, configured to perform the following operations:

[0066] Based on the preset indicator mapping table, the semantic judgment model is used to perform a second verification match on the indicator classification result output by the language representation model;

[0067] The expression for the second check match is:

[0068] b idx ={t i |V(d t ,t i )>η,t i ∈C index ,}

[0069] Where, t i Indicates the i-th indicator information in the indicator mapping table, where i is the index value; C index Indicates the indicator information contained in the preset indicator mapping table; d t represents the indicator classification result output by the language representation model; V(·) represents the semantic judgment model, which is used to perform deep dimension matching on the semantic dimension of the indicator classification result output by the language representation model from the preset indicator mapping table;

[0070] Obtaining a fuzzy indicator set according to the result of the second verification matching;

[0071] Based on the fuzzy indicator set, data table, field information and user attention information, the language representation model is used to perform secondary reasoning on the data table;

[0072] Update the indicator classification results based on the results of secondary reasoning.

[0073] On the other hand, an embodiment of the present invention provides an electronic device, including: a processor and a memory; the memory is used to store programs; the processor executes the program to implement the above-mentioned business analysis report generation method.

[0074] On the other hand, an embodiment of the present invention provides a computer storage medium, in which a program executable by a processor is stored. When the program executable by the processor is executed by the processor, it is used to implement the above-mentioned business analysis report generation method.

[0075] The embodiment of the present invention obtains user-concerned information and the original data of the production and operation items to be analyzed, and obtains target dimension indicators based on the original data extraction; based on the user-concerned information, generates background purpose analysis data based on the target dimension indicators using a preset large language model; based on the target dimension indicators, drills down the original data based on time trends and code values ​​to obtain numerical analysis results; based on the background purpose analysis data and the numerical analysis results, uses a preset large model to integrate data to obtain an experience analysis report. The present invention uses a large language model to automatically process and analyze business data, and uses its text understanding and generation capabilities to obtain a business analysis report. The present invention can efficiently and accurately generate business analysis reports. BRIEF DESCRIPTION OF THE DRAWINGS

[0076] The accompanying drawings are used to provide a further understanding of the technical solution of the present invention and constitute a part of the specification. Together with the embodiments of the present invention, they are used to explain the technical solution of the present invention and do not constitute a limitation on the technical solution of the present invention.

[0077] Figure 1 A schematic diagram of an implementation environment for generating a business analysis report provided by an embodiment of the present invention;

[0078] Figure 2 A flowchart of a method for generating a business analysis report provided by an embodiment of the present invention;

[0079] Figure 3 A schematic diagram of a calculation process of a time trend calculation module provided in an embodiment of the present invention;

[0080] Figure 4 A schematic diagram of a calculation process of a dimension / code value analysis module provided in an embodiment of the present invention;

[0081] Figure 5 A schematic diagram of the overall process of the method for generating a business analysis report provided by an embodiment of the present invention;

[0082] Figure 6 A detailed flowchart of a method for generating a business analysis report provided by an embodiment of the present invention;

[0083] Figure 7 A schematic diagram of an example of background and purpose analysis of a business report provided by an embodiment of the present invention;

[0084] Figure 8 A schematic diagram of an example of computing requirements provided for an embodiment of the present invention;

[0085] Figure 9a A schematic diagram of an example of a business analysis report provided by an embodiment of the present invention;

[0086] Figure 9bA schematic diagram of another example of a business analysis report provided by an embodiment of the present invention;

[0087] Fig.10 A schematic diagram of the structure of an electronic device provided by an embodiment of the present invention. DETAILED DESCRIPTION

[0088] In order to make the purpose, technical solution and advantages of the present invention more clearly understood, the present invention is further described in detail below in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present invention and are not used to limit the present invention.

[0089] It should be noted that, although the functional modules are divided in the system schematic diagram and the logical order is shown in the flowchart, in some cases, the steps shown or described may be performed in a different order than the module division in the system or the order in the flowchart. The terms "first / S100", "second / S200", etc. in the specification, claims and the above-mentioned drawings are used to distinguish similar objects, and are not necessarily used to describe a specific order or sequence.

[0090] Reference to "embodiments" in the present invention means that a particular feature, structure, or characteristic described in conjunction with the embodiments may be included in at least one embodiment of the present invention. The appearance of the phrase in various places in the specification does not necessarily refer to the same embodiment, nor is it an independent or alternative embodiment that is mutually exclusive with other embodiments. It is explicitly and implicitly understood by those skilled in the art that the embodiments described in the present invention may be combined with other embodiments.

[0091] To facilitate the understanding of the technical solution of the present invention, the following is an explanation of the specific names of the technical features that may appear in the technical solution of the present invention:

[0092] LLM (Large Language Model): refers to a deep learning model trained with a large amount of text data that can generate natural language text or understand the meaning of language text. Large language models can handle a variety of natural language tasks, such as text classification, question answering, and dialogue, and are an important path to artificial intelligence.

[0093] like Figure 1 FIG. 1 is a schematic diagram of an implementation environment provided by an embodiment of the present invention. Figure 1 The implementation environment includes at least one terminal 102 and a server 101. The terminal 102 and the server 101 can be connected to a network wirelessly or wired to complete data transmission and exchange.

[0094] Server 101 can be an independent physical server, or a server cluster or distributed system composed of multiple physical servers. It can also be a cloud server that provides basic cloud computing services such as cloud services, cloud databases, cloud computing, cloud functions, cloud storage, network services, cloud communications, middleware services, domain name services, security services, CDN (Content Delivery Network), as well as big data and artificial intelligence platforms.

[0095] In addition, the server 101 can also be a node server in the blockchain network. Blockchain is a new application model of computer technologies such as distributed data storage, peer-to-peer transmission, consensus mechanism, and encryption algorithm.

[0096] The terminal 102 may be a smart phone, a tablet computer, a laptop computer, a desktop computer, a smart speaker, a smart watch, etc., but is not limited thereto. The terminal 102 and the server 101 may be directly or indirectly connected via wired or wireless communication, which is not limited in the embodiment of the present invention.

[0097] Based on the example Figure 1 In the implementation environment shown, an embodiment of the present invention provides a method for generating a business analysis report. The following is an example of applying the method for generating a business analysis report in a terminal 102. It can be understood that the method for generating a business analysis report can also be applied to a server 101.

[0098] Reference Figure 2 , Figure 2 The flowchart of the method for generating a business analysis report applied to a terminal provided in an embodiment of the present invention, the execution subject of the method for generating a business analysis report can be any of the aforementioned computer devices (including a server or a terminal). Figure 2 , the method comprises the following steps:

[0099] S100, obtaining user-focused information and original data of production and operation items to be analyzed, and extracting target dimension indicators based on the original data;

[0100] It should be noted that the target dimension indicators include dimension classification results and indicator classification results; in some embodiments, obtaining the target dimension indicators based on the original data extraction may include the following steps: obtaining the data table to be analyzed as the original data from the production and operation analysis database corresponding to the production and operation project; obtaining the table structure and field information of the data table; based on the table structure and field information, using a pre-trained language representation model to pre-classify dimensions and indicators to obtain dimension classification results and indicator classification results.

[0101] For example, in some specific implementations, by inputting source data (i.e., original data), generating indicators for analysis and performing dimensional classification, the specific implementation can be as follows:

[0102] Step S1-1: Access the production and operation analysis database and obtain the data table D to be analyzed t ;Specifically:

[0103] Step S1-1-1: Obtain the table structure and field information of the data table;

[0104] Step S1-1-2: Use the pre-trained Bert model (language representation model) to pre-classify the table field information into dimensions and indicators, where the Bert model is trained on business data to ensure the classification accuracy and relevance of indicators and table fields.

[0105] It should be noted that, in some embodiments, the method may further include the following steps:

[0106] Based on the preset dimension mapping table, the semantic judgment model is used to perform a first verification match on the dimension classification result output by the language representation model;

[0107] Among them, the expression of the first check match is:

[0108]

[0109] In the formula, c i Represents the i-th dimension information in the dimension mapping table, where i is the index value; C dim Indicates the dimension information contained in the preset dimension mapping table; d k represents the dimensional classification result output by the language representation model; V(·) represents the semantic judgment model, which is used to perform deep dimensional matching on the semantic dimension of the dimensional classification result output by the language representation model from the preset dimensional mapping table; η represents the preset judgment threshold;

[0110] Update the dimension classification result based on the result of the first verification match;

[0111] When the result of the first verification match is empty, obtain the historical user attention information;

[0112] Sequentially concatenate historical user attention information and user attention information, and then use language representation model, dimension mapping table and semantic judgment model to perform joint dimension information matching;

[0113] The expression for joint dimension information matching is:

[0114]

[0115] In the formula, ∪ represents the concatenation symbol, which is used to sequentially concatenate historical user attention information and user attention information; q k represents the kth historical user attention information, q o represents the user's attention information, a represents the total number of cumulative splicing; Bert(·) represents the language representation model;

[0116] Update the dimension classification result based on the result of the joint dimension information matching.

[0117] For example, in some specific implementations, taking the above specific implementation as an example, the following can be further implemented:

[0118] Step S1-1-3: Dimension classification based on step S1-1-2 dim , specifically:

[0119] Step S1-1-3-1: Dimension classification result of Bert model f dim Perform information verification and dimension matching. The specific calculation process is as follows:

[0120]

[0121] In the formula, c i is the specific dimension information in the dimension mapping table, i is the index value, C dim The preset dimension mapping table includes time dimension, trend dimension, etc. k It is the dimension information classified by the Bert model. V(·) is used to perform deep dimension matching on the semantic dimension from the preset dimension mapping table for the dimension result returned by the Bert model.

[0122] The above matching formula is used to improve the robustness of the Bert model output. When the Bert model classification is accurate, that is, d k ∈C dim When the Bert classification result is inaccurate, the semantic judgment model V(·) is used to return the Bert classification result from C dim Find the dimension with the highest semantic match greater than the threshold η as the output dimension information f dim .

[0123] Step S1-1-3-2: If the value returned by step S1-1-3-1 is dim If it is empty, it means that the current user input information does not contain obvious dimension semantic information. The user's historical input information is introduced for analysis and joint dimension information matching calculation is performed. The specific calculation process is as follows:

[0124]

[0125] In the formula, ∪ represents the input question concatenation symbol, and the historical input and current input are concatenated according to the user dialogue sequence. k represents the kth historical data, q o Indicates the current input source data, a is the current total number of splicing. To minimize the impact of historical data, a starts with the historical problem closest to the current problem and accumulates. Once f dim If it is not empty, stop accumulating.

[0126] It should be noted that, in some embodiments, the method further includes the following steps:

[0127] Based on the preset indicator mapping table, the semantic judgment model is used to perform a second verification match on the indicator classification result output by the language representation model;

[0128] The expression for the second check match is:

[0129] b idx ={t i |V(d t ,t i )>η,t i ∈C index ,}

[0130] Where, t i Indicates the i-th indicator information in the indicator mapping table, where i is the index value; C index Indicates the indicator information contained in the preset indicator mapping table; d t represents the indicator classification result output by the language representation model; V(·) represents the semantic judgment model, which is used to perform deep dimension matching on the semantic dimension of the indicator classification result output by the language representation model from the preset indicator mapping table;

[0131] Obtaining a fuzzy indicator set according to the result of the second verification matching;

[0132] Based on the fuzzy indicator set, data table, field information and user attention information, the language representation model is used to perform secondary reasoning on the data table;

[0133] Update the indicator classification results based on the results of secondary reasoning.

[0134] For example, in some specific implementations, taking the above specific implementation as an example, the following can be further implemented:

[0135] Step S1-1-4: Based on step S1-1-2, index information f idx Get, specifically:

[0136] The indicator information judgment annotation is basically the same as the dimension information judgment annotation in step S1-1-3. However, due to the complexity of the indicator information dimension, taking the amount as an example, it includes cost, income, and profit. The cost can be further refined into direct product / service cost, production cost, labor cost, etc., and the expenditure can be further refined into broadband income, server resource expenditure income, service operation income, etc. Profit can be further refined into gross profit and net profit. Therefore, when obtaining indicator information, it is necessary to use the current preset indicator mapping table C index Drill down based on the returned indicator information to obtain the data table D t Field specific information.

[0137] Step S1-1-4-1: Classify the Bert model dimension index information f idx Perform information verification and dimension matching. The specific calculation process is as follows:

[0138] b idx ={t i |V(d t ,t i )>η,t i ∈C index ,}

[0139] Where b idx According to the semantic model V(·) and C index The returned fuzzy index set, C index It is a preset indicator mapping table, including amount, bidding quantity, etc. i C index Corresponding specific indicator information.

[0140] Step S1-1-4-2: According to the fuzzy indicator set returned in step S-1-1-4-1, drill down to the dimension and confirm the final data table field dimension indicator. The specific steps are as follows:

[0141] S1-1-4-2-1 based on b idx Extract database field information, and the extraction calculation is as follows:

[0142] Using fuzzy index set b idx 、Known data table D t Field Information D c t , perform secondary construction of question input, and input the Bert model trained in step S1-1-2 for reasoning, and obtain the indicator information after drilling down into the specific data table z idx , the question template is constructed as follows:

[0143] {

[0144] Please help me find the most suitable field information based on the following known information:

[0145] Known issues: o ,

[0146] Limited indicator analysis dimension: b idx ,

[0147] Field to be filtered: D c t

[0148] }

[0149] In some specific application scenarios, the indicator information can be idx As the updated indicator information f idx . You can also perform further verification and matching through the following process:

[0150] Step S1-1-4-2-2 is based on the z returned by Bert inference idx , use the semantic analysis model to check the matching degree. The specific verification calculation is as follows (the verification principle is consistent with the first verification matching principle mentioned above, and the process logic details are not repeated):

[0151]

[0152] In the formula, m i For z idx Specific database field information.

[0153] S1-1-4-2-3: If step f idx If the returned value is empty, start from the closest to the current input question, gradually add historical questions, and repeat steps S1-1-4-2-1 and S1-1-4-2-2 to obtain the final indicator information f idx .

[0154] In some specific application scenarios, the pre-classification and mapping results of dimensions and indicators can be manually adjusted to form a data table business analysis model, namely:

[0155] D t |(f dim ,f idx );

[0156] S200, generating background purpose analysis data based on target dimension indicators using a preset large language model according to the user's attention information;

[0157] It should be noted that, in some embodiments, step S200 may include the following steps: based on a preset first prompt word template, generating first prompt word information according to user attention information and target dimension indicators; inputting the first prompt word information into a preset large language model for reasoning to generate background purpose analysis data.

[0158] For example, in some specific implementations, purpose and context identification can be implemented as follows:

[0159] Step S2-1: a user input module, used to accept instructions for generating a report and related requirements;

[0160] Step S2-2: Use LLM capabilities to identify the purpose and background of user input, and analyze the user's intention and needs in combination with the user's context information; specifically:

[0161] Step S2-2-1: Business analysis focuses on identifying analysis targets; specifically:

[0162] Step S2-2-1-1: Use prompt technology to identify and locate dimensions and indicators, and ensure efficient and accurate extraction of key information based on the natural language understanding capabilities of the large language model trained for data analysis

[0163] Step S2-2-1-2: Construct business analysis focusing on target identification prompt as follows:

[0164] {

[0165] You are a business analysis master. Please find out the indicators and dimensions that users are concerned about based on their questions.

[0166] User question:f question

[0167] Related indicators: f idx

[0168] Related dimension: f dim

[0169] }

[0170] Step S2-2-2: Generate background and purpose analysis of business analysis report; specifically:

[0171] Step S2-2-2-1: Based on the reasoning ability of the trained large language model, the prompt technology is used to realize the background purpose analysis and generation;

[0172] Step S2-2-2-2: Construct the background purpose analysis of the business analysis report. The prompt is as follows:

[0173] {

[0174] You are a business analysis master. Please generate a background purpose analysis based on the user's question.

[0175] Table information: f schema

[0176] User question:fquestion

[0177] Related indicators: f idx

[0178] Related dimension: f dim

[0179] }

[0180] Step S2-3: Get the relevant dimension f for this round of tasks dim 、Indicator f idx And analysis purpose tar , sent to the next step.

[0181] S300, according to the target dimension indicator, drill down analysis is performed on the original data based on the time trend and the code value to obtain a numerical analysis result;

[0182] It should be noted that the original data represents the data table corresponding to the production and operation project; in some embodiments, step S300 may include the following steps: obtaining numerical data corresponding to the target dimension indicator from the data table; arranging all data values ​​in the numerical data in a time series based on their corresponding time points; performing month-on-month statistics and comparative statistics on the numerical data based on a preset time interval to obtain time trend analysis data; performing incremental statistics and increase statistics on the numerical data to obtain code value analysis data; based on the time trend analysis data and the code value analysis data, performing single-dimensional drill-down statistical analysis and multi-dimensional drill-down statistical analysis based on the target dimension indicator to obtain numerical analysis results.

[0183] For example, in some specific implementations, data analysis can be performed by a time trend calculation module and a dimension / code value analysis module, which can be specifically implemented as follows:

[0184] Step S3-1: The time trend calculation module calculates (corresponding to the time trend analysis data); specifically:

[0185] Step S3-1-1: According to the characteristics of the business analysis, the time trend calculation includes: year-on-year, month-on-month, month-on-month, cumulative year-on-year, month-on-month and month-on-month. The calculation logic is as follows: Figure 3 As shown;

[0186] The year-on-year calculation method is as follows:

[0187]

[0188] The month-on-month calculation method is as follows:

[0189]

[0190] Among them, V curris the current base, such as this year, this month, and this year’s cumulative total; V same The base number for the same period, such as the current month of last year, the cumulative number of months of last year; V pre It is the base of the previous period, such as last year, last month, or last year’s cumulative amount.

[0191] Step S3-1-2: Input is data table content, time series, indicators and dimensions involved;

[0192] Step S3-1-3: Determine the time field / focus granularity;

[0193] Step S3-1-4: Determine the calculation type of the analysis indicator according to the indicator definition: mean, sum;

[0194] Step S3-1-5: Determine whether there is a daily continuity; specifically:

[0195] Step S3-1-5-1: If not, calculate the specified day comparison;

[0196] Step S3-1-5-2: If yes, calculate the day-on-day comparison;

[0197] Step S3-1-6: Determine whether there is a whole month; specifically:

[0198] Step S3-1-6-1: If not, the time trend calculation is completed;

[0199] Step S3-1-6-2: If yes, calculate the result for the whole month;

[0200] Step S3-1-7: Determine whether there is a month continuity; specifically:

[0201] Step S3-1-7-1: If not, calculate the specified month comparison;

[0202] Step S3-1-7-2: If yes, calculate the month-on-month / month-on-month comparison;

[0203] Step S3-1-8: Determine whether a whole year exists; specifically:

[0204] Step S3-1-8-1: If not, the time trend calculation is completed;

[0205] Step S3-1-8-2: If yes, calculate the result for the whole year;

[0206] Step S3-1-9: Determine whether there is a year continuity; specifically:

[0207] Step S3-1-9-1: If not, calculate the specified year comparison;

[0208] Step S3-1-9-2: If yes, calculate the year-on-year comparison;

[0209] Step S3-2: Dimension / code value analysis module calculation; specifically:

[0210] Step S3-2-1: According to the business characteristics of the business analysis, the dimension / code value analysis includes: the total amount, increment, and increase of the index aggregation results, and the calculation logic is as follows: Figure 4 As shown;

[0211] Step S3-2-2: Input is data table content, time series, indicators and dimensions involved;

[0212] Step S3-2-3: Calculate the result of the indicator calculation under each dimension (combination);

[0213] Step S3-2-4: Calculate the incremental result, the formula is as follows:

[0214] Y 增量 =V end -V base

[0215] Among them, V end is the amount of the indicator / dimension at the deadline, V base is the amount of the indicator / dimension at the base time;

[0216] Step S3-2-5: Calculate the amplification result, the formula is as follows:

[0217]

[0218] Among them, V ebd is the amount of the indicator / dimension at the deadline, V base The amount of the indicator / dimension at the base time

[0219] Step S3-3: Drill down analysis of a single dimension; specifically:

[0220] Step S3-3-1: Drill down the indicator in a single dimension, including: the index sum, index increase, and index increment head value results;

[0221] Step S3-4: Drill down analysis in multiple dimensions; specifically:

[0222] Step S3-4-1: Drill down the indicator in the nth dimension, including the head value results of the indicator sum, indicator increase, and indicator increment;

[0223] Step S3-4-2: Drill down results collection, including time analysis trend (overall analysis) whl , Single dimension analysis (focus analysis)f sig , Multi-dimensional analysis (focus analysis)f multi .

[0224] S400, analyzing data and numerical analysis results based on background objectives, integrating data using a preset large model, and obtaining an empirical analysis report.

[0225] It should be noted that, in some embodiments, step S400 may include the following steps: based on a preset second prompt word template, generating second prompt word information according to background purpose analysis data and numerical analysis results; inputting the second prompt word information into a preset large model for data integration to obtain an experience analysis report.

[0226] For example, in some specific implementations, content integration and business analysis report generation through a big model can be implemented as follows:

[0227] Step S4-1: Obtain the analysis objective f of step S2 tar , Step S3 time analysis trend (overall analysis) f whl , Single dimension analysis (focus analysis)f sig , Multi-dimensional analysis (focus analysis)f multi ;

[0228] Step S4-2: Utilize the reasoning ability of the large model and combine it with business data to generate modules one by one; specifically:

[0229] Step S4-2-1: Build a separate module prompt example as follows:

[0230] {

[0231] You are a business analysis master. Please generate a {module} analysis based on the data.

[0232] Related analysis data: f module

[0233] Related indicators: f idx

[0234] Related dimension: f dim

[0235] }

[0236] where f module f whl 、f sig or multi ;

[0237] Step S4-2-2: traverse all modules to be analyzed and perform step S4-2-1 prompt result output;

[0238] Step S4-2-3: Integrate the output results of step S4-2-2 in order;

[0239] Step S4-3: Use the big model to summarize and generate suggestions; specifically:

[0240] Step S4-3-1: Construct a summary suggestion module prompt example as follows:

[0241] {

[0242] You are a business analysis master. Please generate a summary analysis based on the data.

[0243] Related analysis data: f module

[0244] }

[0245] Step S4-3-2: Generate summary analysis content;

[0246] Step S4-3: Integrate all large model generation content and generate a business analysis report.

[0247] On the other hand, an embodiment of the present invention provides a business analysis report generating device, including:

[0248] The source data module is used to obtain the original data of the information that users are concerned about and the production and operation items to be analyzed, and extract the target dimension indicators based on the original data;

[0249] The semantic understanding module is used to generate background purpose analysis data based on the target dimension indicators according to the information that users pay attention to using the preset large language model;

[0250] The data analysis module is used to drill down and analyze the original data based on the target dimension indicators, time trends and code values ​​to obtain numerical analysis results;

[0251] The report generation module is used to analyze data and numerical analysis results based on background purposes, and use the preset large model to integrate data to obtain an empirical analysis report.

[0252] In some embodiments, the target dimension indicator includes a dimension classification result and an indicator classification result; the source data module can be used to perform the following operations:

[0253] Obtaining the data table to be analyzed as original data from the production and operation analysis database corresponding to the production and operation project;

[0254] Get the table structure and field information of the data table;

[0255] Based on the table structure and field information, the pre-trained language representation model is used to pre-classify dimensions and indicators to obtain dimension classification results and indicator classification results.

[0256] In some embodiments, the apparatus may further include a first module configured to perform the following operations:

[0257] Based on the preset dimension mapping table, the semantic judgment model is used to perform a first verification match on the dimension classification result output by the language representation model;

[0258] Among them, the expression of the first check match is:

[0259]

[0260] In the formula, c i Represents the i-th dimension information in the dimension mapping table, where i is the index value; C dim Indicates the dimension information contained in the preset dimension mapping table; d k represents the dimensional classification result output by the language representation model; V(·) represents the semantic judgment model, which is used to perform deep dimensional matching on the semantic dimension of the dimensional classification result output by the language representation model from the preset dimensional mapping table; η represents the preset judgment threshold;

[0261] Update the dimension classification result based on the result of the first verification match;

[0262] When the result of the first verification match is empty, obtain the historical user attention information;

[0263] Sequentially concatenate historical user attention information and user attention information, and then use language representation model, dimension mapping table and semantic judgment model to perform joint dimension information matching;

[0264] The expression for joint dimension information matching is:

[0265]

[0266] In the formula, ∪ represents the concatenation symbol, which is used to sequentially concatenate historical user attention information and user attention information; q k represents the kth historical user attention information, q o represents the user's attention information, a represents the total number of cumulative splicing; Bert(·) represents the language representation model;

[0267] Update the dimension classification result based on the result of the joint dimension information matching.

[0268] In some embodiments, the apparatus may further include a second module configured to perform the following operations:

[0269] Based on the preset indicator mapping table, the semantic judgment model is used to perform a second verification match on the indicator classification result output by the language representation model;

[0270] The expression for the second check match is:

[0271] b idx ={ti |V(d t ,t i )>η,t i ∈C index ,}

[0272] Where, t i Indicates the i-th indicator information in the indicator mapping table, where i is the index value; C index Indicates the indicator information contained in the preset indicator mapping table; d t represents the indicator classification result output by the language representation model; V(·) represents the semantic judgment model, which is used to perform deep dimension matching on the semantic dimension of the indicator classification result output by the language representation model from the preset indicator mapping table;

[0273] Obtaining a fuzzy indicator set according to the result of the second verification matching;

[0274] Based on the fuzzy indicator set, data table, field information and user attention information, the language representation model is used to perform secondary reasoning on the data table;

[0275] Update the indicator classification results based on the results of secondary reasoning.

[0276] The contents of the method embodiments of the present invention are all applicable to the device embodiments. The functions specifically implemented by the device embodiments are the same as those of the above method embodiments, and the beneficial effects achieved are also the same as those achieved by the above method embodiments.

[0277] In order to explain the principle of the technical solution of the present invention in detail, the overall process of the present invention is described below in combination with some specific embodiments. It is easy to understand that the following is an explanation of the technical principle of the present invention and cannot be regarded as a limitation of the present invention.

[0278] First of all, it should be explained that the present invention proposes a method for generating a business analysis report based on a large language model and dimensional drilling. The large model is used to learn the structure of the business data table, and at the same time, combined with user input, the demand background and purpose are analyzed; intelligent analysis indicators and dimensions are used, and a time trend calculation module and a dimension / code value analysis module are introduced to perform multi-dimensional data drilling on business data, so as to realize automatic integration of analysis results, generate a business analysis report including time trend analysis, single dimension analysis and multi-dimensional analysis, and display the key indicators of the report and their description and explanation. It mainly includes the following four modules:

[0279] 1. Source data module:

[0280] This module is used to receive raw data from the enterprise database, including but not limited to financial data, sales data, market data, etc. It also supports online analysis of accessed data, including intelligent analysis of indicators and dimensions in the database, and provides user instruction mapping capabilities. It ensures the accuracy and completeness of the data and provides a reliable basis for subsequent analysis.

[0281] 2. Semantic understanding module:

[0282] This module combines the user input with the data table to analyze the business background and purpose, and uses text generation technology to automatically generate the introduction part of the business analysis report, outlining the background and purpose of the analysis; at the same time, it provides directional guidance for other business analysis modules.

[0283] 3. Data analysis module:

[0284] This module includes a time trend calculation module and a dimension / code value analysis module. The time trend calculation module is used to calculate data indicators under different time series, including year-on-year, month-on-month, month-on-year, cumulative year-on-year for the month of the current year, and month-on-month and month-on-month for the current year; the dimension / code value analysis module is used to analyze the aggregation results of indicators, including the calculation of total amount, increment and increase. At the same time, advanced statistical methods and algorithms, such as time series analysis and cluster analysis, are applied to conduct in-depth analysis of the data. Multi-dimensional data drilling functions are realized, including but not limited to the calculation of indicator sum, increase and increment, and support the analysis of header value results.

[0285] 4. LLM report generation module:

[0286] This module uses the big model capabilities to automatically integrate analysis results and generate reports that include time trend analysis, single dimension analysis, and multi-dimensional analysis. The reports include the display of key indicators and detailed descriptions and explanations of data analysis.

[0287] like Figure 5 As shown, the main process of the scheme of the present invention is as follows:

[0288] S1: Collect raw data from various data sources of the enterprise (such as financial data, market data, sales data, etc.), clean and standardize them, use the big language model to train the processed data, learn the table structure and potential relationships of the data, and fine-tune the model for specific fields to improve the performance of the model in this field. Use the big language model to classify table fields and indicators, and map the dimension library and indicator library through the model and manual work.

[0289] S2: Use large language models to identify the purpose and context of user input to ensure effective information extraction and analysis. First, the user input module accepts instructions and related requirements for generating reports, and then uses LLM's natural language understanding capabilities to analyze the user's intentions and needs in combination with the user's contextual information. Specifically, business analysis focuses on target identification, using prompt technology to accurately identify dimensions and indicators, thereby efficiently extracting key information. In addition, LLM's reasoning capabilities are used to generate background purpose analysis, and a dedicated prompt is constructed to meet analysis needs. The important role of large language models in understanding user needs and generating high-quality analysis reports.

[0290] S3: Time trend and dimension / code value analysis of data. First, the time trend calculation module based on the large language model is used to analyze the data on a year-on-year, month-on-month and year-on-year basis; then the dimension / code value analysis module is used to calculate the total amount, increment and increase of indicators in each dimension; finally, single-dimensional and multi-dimensional drill-down analysis is performed to gradually gain a deeper understanding of the performance of indicators in different dimensions.

[0291] S4: Use the big model to integrate content and generate business analysis reports. First, obtain the analysis purpose and various analysis results of step S3 (time trend, single dimension, multi-dimensional analysis, etc.). Next, the big model generates analysis content module by module. By building modular prompts, relevant data, indicators and dimensions are input into the big model to generate corresponding analysis texts, which are integrated after traversing all modules. Finally, the big model summarizes the overall content and generates suggestions to form the final business analysis report. By utilizing the reasoning, generation and integration capabilities of the big model in this process, complex content generation and analysis can be completed efficiently.

[0292] like Figure 6 As shown, it is a flow chart of the detailed steps of the technical solution of the present invention, wherein the detailed steps of the technical solution of the present invention refer to the example description corresponding to the specific implementation methods of each step in the aforementioned method embodiment of the present invention, and will not be repeated here.

[0293] In some specific application scenarios, the present invention is used in the business analysis link. The following is a detailed description of the main analysis steps of business opportunity business analysis:

[0294] Step S1: Dimension f corresponding to the business analysis database dim With the index f idx ; Take "Help me analyze the contract signing amount in each province in each industry in 2022" as the user input (user attention information) as an example, specifically:

[0295] Step S1-1: Dimension f dim Get. Specifically:

[0296] The default dimension mapping table is C dim Including: date, bidding province, bid type, winning bidder, etc.;

[0297] Step S1-1-1: Inference results based on the Bert model of the user question: date, bidding province, and winning bidder;

[0298] Step S1-1-2: Results of deep matching based on the semantic matching model: date, bidding province;

[0299] Step S1-2: Index f idx Get. Specifically:

[0300] The default indicator mapping table is C index Including: quantity, amount, trend, etc.;

[0301] Data table fields: number of projects, number of bidding units, total amount of winning bids, total amount of sales, etc.

[0302] Step S1-2-1: Inference results of Bert model based on user questions: amount, trend;

[0303] Step S1-1-2: The construction of the new question has been Bert's secondary reasoning result:

[0304] The input is:

[0305] Please help me find the most suitable field information based on the following known information:

[0306] Known issue: Please help me analyze the contract signing situation in various industries and provinces in 2022.

[0307] Limit the indicator analysis dimensions: amount, trend;

[0308] Fields to be filtered: number of projects, number of bidding units, total amount of winning bids, etc.;

[0309] }

[0310] The returned result is: the total amount of the winning bid;

[0311] Step S1-1-3: The result of the second audit of the semantic model is: the total amount of the winning bid;

[0312] Step S2: User semantic understanding module is used for purpose and background recognition; specifically:

[0313] Step S2-1: User input: Please generate an operating analysis report based on the operating report scenario;

[0314] Step S2-2: Constructing business analysis focuses on analyzing target identification prompts as follows:

[0315] Step S2-2-1: The large model outputs {indicators: date, winning bidder, dimension: total amount of winning bid};

[0316] Step S2-3: Construct the background purpose analysis of the business analysis report to generate prompts as follows:

[0317] {

[0318] You are a business analysis master. Please generate a background purpose analysis based on the user's question.

[0319] Table information: f schema

[0320] User question: Please generate an operating analysis report based on the operating report scenario;

[0321] Related indicators: Date, bidding province;

[0322] Relevant dimensions: total amount of winning bids;

[0323] }

[0324] Step S2-3-1: Example of large model output results: Figure 7 As shown, it includes a title part 701, a background part 702 and a purpose part 703;

[0325] Step S3: Perform data analysis through the time trend calculation module and the dimension / code value analysis module; specifically:

[0326] Step S3-1: numerical calculation: mean; specifically:

[0327] Step S3-1-1: Business requirement: only the first 10 codes above the average are listed, and only the first 10 codes below the average are listed;

[0328] Step S3-1-2: Analysis module, numerical statistics input under dimension; specifically:

[0329] {Time granularity}: cumulative for the year, month;

[0330] {Time range}: January-April 2024, 202404;

[0331] {Analysis indicators}: contract signing amount;

[0332] {Analysis dimension one}: industry;

[0333] {Analysis Dimension 2}: Province;

[0334] Step S3-1-2: Analysis module, input of numerical description under dimension; specifically:

[0335] {Describe the grading criteria}: (default: average, max and / or min, mode, median);

[0336] {Description range}: the number of code values ​​n, or m% of the number of code values;

[0337] {description order}: default desc;

[0338] Step S3-2: Amplitude calculation: overall amplification; specifically:

[0339] Step S3-2-1: Business requirements: Pull in the same direction, and only list the first 10 code values ​​that are higher than the overall increase, and only list the first 10 code values ​​that are lower than the overall increase;

[0340] Step S3-2-2: Analysis module, inputting the code value increment statistics under the dimension; specifically:

[0341] {Time granularity}: cumulative for the year, month;

[0342] {Time range}: January-April 2024, 202404;

[0343] {Calculation type}: year-on-year and / or month-on-month (year-on-year granularity only);

[0344] {Analysis indicators}: contract signing amount;

[0345] {Analysis dimension one}: industry;

[0346] {Analysis Dimension 2}: Province;

[0347] Step S3-2-3: Analysis module, input of code value increment description under dimension; specifically:

[0348] {Describe the grading criteria}: (positive / negative, default: total growth rate, max and / or min, mode, median)

[0349] {Description range}: number of code values ​​n, or m% of the number of code values

[0350] {Description order}: default desc

[0351] Step S3-3: Increment calculation: overall increment; specifically:

[0352] Step S3-3-1: Business requirements: For the same direction, only the top 10 codes whose contribution ranking accounts for the top 30% of all code values ​​are listed; for the negative direction, only the top 10 codes whose contribution ranking accounts for the bottom 30% of all code values ​​are listed;

[0353] Step S3-3-2: Analysis module, dimension code value increment statistics input; specifically:

[0354] {Time granularity}: cumulative for the year, month;

[0355] {Time range}: January-April 2024, 202404;

[0356] {Calculation type}: year-on-year and / or month-on-month (year-on-year granularity only);

[0357] {Analysis indicators}: contract signing amount;

[0358] {Analysis dimension one}: industry;

[0359] {Analysis Dimension 2}: Province;

[0360] Step S3-2-2: Analysis module, input of code value increment description under dimension; specifically:

[0361] {Description grading standard}: (positive / negative, default: total share increase ratio, max and / or

[0362] min, mode, median);

[0363] {Description range}: the number of code values ​​n, or m% of the number of code values;

[0364] {description order}: default desc;

[0365] Based on the above specific implementation example, the calculation requirements are as follows: Figure 8 As shown, the part marked with background color is a specific example of the analysis result output according to the business analysis requirements;

[0366] Step S4: Use the big model to integrate content and generate business analysis reports; examples of business analysis reports include: Figure 9a and Figure 9b As shown, Figure 9a and Figure 9b The data structure of the business analysis report is mainly used to illustrate and display the data structure of the business analysis report. In order to avoid leakage of sensitive data, the specific numerical data are desensitized (the specific parameter values ​​will change with different application scenarios in actual applications, so the desensitization of numerical data cannot be regarded as insufficient disclosure of the technical solution of the present invention). The data structure mainly includes background and purpose (obtained based on the data results of step S2), data overview and overall analysis (obtained based on the data results of steps S3-1 and S3-2), key analysis (including Figure 9a Single dimension analysis and Figure 9b The cross-dimensional analysis in step S3-3 is obtained based on the data results of step S3-3) and the summary and suggestions (the summary and suggestions of the overall content are generated through the big model).

[0367] In summary, the present invention aims to provide a method for generating a business analysis report based on a large language model and dimensional drill-down, which is suitable for writing a business analysis report, automatically processing and analyzing business data through a large language model, and utilizing its text understanding and generation capabilities to obtain a business analysis report.

[0368] Compared with the prior art, the present invention has at least the following beneficial effects:

[0369] 1. In-depth data understanding: The system uses LLMs to conduct in-depth analysis of business data to understand the complex relationships and potential meanings between data.

[0370] 2. Flexible report generation: By utilizing the natural language generation capabilities of LLMs, the system can generate structured and content-rich reports based on different analysis requirements and business data characteristics.

[0371] 3. Enhanced interactivity: The system supports natural language queries and feedback, allowing users to interact with the system in a more natural way, improving the user experience.

[0372] Through the above technical improvements, the present invention can provide more accurate, in-depth and user-friendly business analysis reports to help enterprises better understand their operating conditions and make wise decisions.

[0373] On the other hand, an embodiment of the present invention further provides an electronic device, the electronic device includes a memory and a processor, the memory stores a computer program, and the processor implements the above sensitive information method when executing the computer program. The electronic device can be any intelligent terminal including a tablet computer, a car computer, etc.

[0374] It can be understood that the contents of the above method embodiments are all applicable to the present device embodiments, the functions specifically implemented by the present device embodiments are the same as those of the above method embodiments, and the beneficial effects achieved are also the same as those achieved by the above method embodiments.

[0375] like Fig.10 As shown, Fig.10 A specific example of the hardware structure of an electronic device 1000 according to an embodiment is shown. The electronic device 1000 includes:

[0376] The processor 1001 may be implemented by a general-purpose CPU (Central Processing Unit), a microprocessor, an application-specific integrated circuit (ASIC), or one or more integrated circuits, and is used to execute relevant programs to implement the technical solutions provided in the embodiments of the present invention.

[0377] The memory 1002 can be implemented in the form of a read-only memory (ROM), a static storage device, a dynamic storage device, or a random access memory (RAM). The memory 1002 can store an operating system and other application programs. When the technical solution provided in the embodiment of this specification is implemented by software or firmware, the relevant program code is stored in the memory 1002, and the processor 1001 calls and executes the network node population optimization method of the embodiment of the present invention;

[0378] Input / output interface 1003, used to implement information input and output;

[0379] The communication interface 1004 is used to realize the communication interaction between the device and other devices. The communication can be realized through a wired manner (such as USB, network cable, etc.) or a wireless manner (such as mobile network, WIFI, Bluetooth, etc.);

[0380] A bus 1005 , which transmits information between various components of the device (e.g., the processor 1001 , the memory 1002 , the input / output interface 1003 , and the communication interface 1004 );

[0381] The processor 1001 , the memory 1002 , the input / output interface 1003 and the communication interface 1004 are connected to each other in communication within the device via the bus 1005 .

[0382] The electronic device embodiments described above are merely illustrative, and the units described as separate components may or may not be physically separated, that is, they may be located in one place, or they may be distributed on multiple network units. Some or all of the modules may be selected according to actual needs to achieve the purpose of the solution of this embodiment.

[0383] The contents of the method embodiments of the present invention are all applicable to the electronic device embodiments. The functions specifically implemented by the electronic device embodiments are the same as those of the above method embodiments, and the beneficial effects achieved are also the same as those achieved by the above method embodiments.

[0384] Another aspect of an embodiment of the present invention further provides a computer-readable storage medium, the storage medium stores a program, and the program is executed by a processor to implement the above method.

[0385] It should be noted that the computer-readable medium shown in the embodiment of the present invention may be a computer-readable signal medium or a computer-readable storage medium or any combination of the above two. The computer-readable storage medium may be, for example, but not limited to, an electrical, magnetic, optical, electromagnetic, infrared, or semiconductor system, device or device, or any combination of the above. More specific examples of computer-readable storage media may include, but are not limited to: an electrical connection with one or more wires, a portable computer disk, a hard disk, a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM), a flash memory, an optical fiber, a portable compact disk read-only memory (Compact Disc Read to Only Memory, CD to ROM), an optical storage device, a magnetic storage device, or any suitable combination of the above. In the present invention, a computer-readable storage medium may be any tangible medium containing or storing a program, which may be used by or in combination with an instruction execution system, device or device. In the present invention, a computer-readable signal medium may include a data signal propagated in a baseband or as part of a carrier wave, which carries a computer-readable program code. Such propagated data signals may take a variety of forms, including but not limited to electromagnetic signals, optical signals, or any suitable combination of the above. Computer readable signal media may also be any computer readable medium other than computer readable storage media, which may send, propagate, or transmit programs for use by or in conjunction with an instruction execution system, apparatus, or device. The program code contained on the computer readable medium may be transmitted using any suitable medium, including but not limited to: wireless, wired, etc., or any suitable combination of the above.

[0386] The contents of the method embodiments of the present invention are all applicable to the computer-readable storage medium embodiments. The functions specifically implemented by the computer-readable storage medium embodiments are the same as those of the above method embodiments, and the beneficial effects achieved are also the same as those achieved by the above method embodiments.

[0387] The embodiment of the present invention also discloses a computer program product or a computer program, which includes a computer instruction stored in a computer-readable storage medium. A processor of a computer device can read the computer instruction from the computer-readable storage medium, and the processor executes the computer instruction, so that the computer device executes the above method.

[0388] The flow chart and block diagram in the accompanying drawings illustrate the possible architecture, function and operation of the system, method and computer program product according to various embodiments of the present invention. In this regard, each box in the flow chart or block diagram can represent a module, a program segment, or a part of a code, and the above-mentioned module, program segment, or a part of a code contains one or more executable instructions for realizing the specified logical function. It should also be noted that in some alternative implementations, the functions marked in the box can also occur in a different order from the order marked in the accompanying drawings. For example, two boxes represented in succession can actually be executed substantially in parallel, and they can sometimes be executed in the opposite order, depending on the functions involved. It should also be noted that each box in the block diagram or flow chart, and the combination of the boxes in the block diagram or flow chart can be implemented with a dedicated hardware-based system that performs a specified function or operation, or can be implemented with a combination of dedicated hardware and computer instructions.

[0389] It should be noted that, although several modules of the device for action execution are mentioned in the above detailed description, this division is not mandatory. In fact, according to an embodiment of the present invention, the features and functions of two or more modules or units described above can be embodied in one module or unit. On the contrary, the features and functions of one module or unit described above can be further divided into being embodied by multiple modules or units.

[0390] Through the description of the above implementation, it is easy for those skilled in the art to understand that the example implementation described here can be implemented by software, or by software combined with necessary hardware. Therefore, the technical solution according to the implementation of the present invention can be embodied in the form of a software product, which can be stored in a non-volatile storage medium (which can be a CD to ROM, a USB flash drive, a mobile hard disk, etc.) or on a network, including several instructions to enable a computing device (which can be a personal computer, a server, a touch terminal, or a network device, etc.) to execute the method according to the implementation of the present invention.

[0391] In some selectable embodiments, the function / operation mentioned in the block diagram may not occur in the order mentioned in the operation diagram. For example, depending on the function / operation involved, the two boxes shown in succession can actually be executed substantially simultaneously or the boxes can sometimes be executed in reverse order. In addition, the embodiment presented and described in the flow chart of the present invention is provided by way of example, for the purpose of providing a more comprehensive understanding of the technology. The disclosed method is not limited to the operation and logic flow presented by the present invention. Selectable embodiments are expected, wherein the order of various operations is changed and the sub-operation of a part described as a larger operation is performed independently.

[0392] In addition, although the present invention is described in the context of functional modules, it should be understood that, unless otherwise specified, one or more of the functions and / or features can be integrated into a single physical device and / or software module, or one or more functions and / or features can be implemented in separate physical devices or software modules. It is also understood that a detailed discussion of the actual implementation of each module is unnecessary for understanding the present invention. More specifically, in view of the properties, functions and internal relationships of various functional modules in the device disclosed in the present invention, the actual implementation of the module will be understood within the conventional skills of engineers. Therefore, those skilled in the art can implement the present invention set forth in the claims without excessive experimentation using ordinary techniques. It is also understood that the specific concepts disclosed are merely illustrative and are not intended to limit the scope of the present invention, which is determined by the full scope of the appended claims and their equivalents.

[0393] If the function is implemented in the form of 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 the present invention, or the part that contributes to the prior art, or the part of the technical solution, can be embodied in the form of a software product. The computer software product is stored in a storage medium, including several instructions for a computer device (which can be a personal computer, server, or network device, etc.) to perform all or part of the steps of the methods of each embodiment of the present invention. The aforementioned storage medium includes: U disk, mobile hard disk, read-only memory (ROM, Read Only Memory), random access memory (RAM, Random Access Memory), disk or optical disk, etc. Various media that can store program codes.

[0394] The logic and / or steps represented in the flowchart or otherwise described herein, for example, may be considered as an ordered list of executable instructions for implementing logical functions, and may be embodied in any computer-readable medium for use by an instruction execution device, apparatus or device (such as a computer-based device, a device including a processor, or other device that can fetch instructions from an instruction execution device, apparatus or device and execute the instructions), or in conjunction with such instruction execution device, apparatus or device. For purposes of this specification, "computer-readable medium" may be any device that can contain, store, communicate, propagate or transmit a program for use by an instruction execution device, apparatus or device, or in conjunction with such instruction execution device, apparatus or device.

[0395] More specific examples of computer-readable media (a non-exhaustive list) include the following: an electrical connection with one or more wires (electronic device), a portable computer disk case (magnetic device), a random access memory (RAM), a read-only memory (ROM), an erasable and programmable read-only memory (EPROM or flash memory), a fiber optic device, and a portable compact disk read-only memory (CDROM). In addition, the computer-readable medium may even be a paper or other suitable medium on which the program is printed, since the program may be obtained electronically, for example, by optically scanning the paper or other medium, followed by editing, deciphering or, if necessary, processing in another suitable manner, and then stored in a computer memory.

[0396] It should be understood that the various parts of the present invention can be implemented by hardware, software, firmware or a combination thereof. In the above-mentioned embodiments, a plurality of steps or methods can be implemented by software or firmware stored in a memory and executed by a suitable instruction execution device. For example, if implemented by hardware, as in another embodiment, it can be implemented by any one of the following technologies known in the art or their combination: a discrete logic circuit having a logic gate circuit for implementing a logic function for a data signal, a dedicated integrated circuit having a suitable combination of logic gate circuits, a programmable gate array (PGA), a field programmable gate array (FPGA), etc.

[0397] In the description of this specification, the description with reference to the terms "one embodiment", "some embodiments", "examples", "specific examples", or "some examples" means that the specific features, structures, materials or characteristics described in conjunction with the embodiment or example are included in at least one embodiment or example of the present invention. In this specification, the schematic representation of the above terms does not necessarily refer to the same embodiment or example. Moreover, the specific features, structures, materials or characteristics described may be combined in any one or more embodiments or examples in a suitable manner.

[0398] Although the embodiments of the present invention have been shown and described, it will be appreciated by those skilled in the art that various changes, modifications, substitutions and variations may be made to the embodiments without departing from the principles and spirit of the present invention, and that the scope of the present invention is defined by the claims and their equivalents.

[0399] The above is a specific description of the preferred implementation of the present invention, but the present invention is not limited to the embodiments. Those skilled in the art can make various equivalent modifications or substitutions without violating the spirit of the present invention. These equivalent modifications or substitutions are all included in the scope defined by the claims of the present invention.

Claims

1. A method for generating a business analysis report, characterized in that: The following steps are involved: Obtaining the original data of the user's concerned information and the production and operation items to be analyzed, and extracting the target dimension indicators based on the original data; According to the user attention information, using a preset large language model to generate background purpose analysis data based on the target dimension indicators; According to the target dimension indicator, the raw data is subjected to drill-down analysis based on time trends and code values ​​to obtain numerical analysis results; Based on the background purpose analysis data and the numerical analysis results, a preset large model is used to integrate the data to obtain an empirical analysis report.

2. The method for generating a business analysis report according to claim 1, characterized in that: The target dimension indicator includes a dimension classification result and an indicator classification result; the step of extracting the target dimension indicator based on the original data includes the following steps: Acquire a data table to be analyzed from a production and operation analysis database corresponding to the production and operation project as the original data; Obtaining the table structure and field information of the data table; Based on the table structure and the field information, a pre-trained language representation model is used to pre-classify dimensions and indicators to obtain the dimension classification results and the indicator classification results.

3. The method for generating a business analysis report according to claim 2, characterized in that: The method further comprises the following steps: Based on a preset dimension mapping table, a semantic judgment model is used to perform a first verification match on the dimension classification result output by the language representation model; The expression for the first verification match is: In the formula, c i Represents the i-th dimension information in the dimension mapping table, where i is the index value; C dim Indicates the dimension information contained in the preset dimension mapping table; d k represents the dimensional classification result output by the language representation model; V(·) represents the semantic judgment model, which is used to perform deep dimensional matching on the semantic dimension of the dimensional classification result output by the language representation model from the preset dimensional mapping table; η represents the preset judgment threshold; Update the dimension classification result based on the result of the first verification match; When the result of the first verification match is empty, obtaining historical user attention information; Sequentially concatenate the historical user attention information and the user attention information, and then use the language representation model, the dimension mapping table and the semantic judgment model to perform joint dimension information matching; The expression for matching the joint dimension information is: In the formula, ∪ represents the concatenation symbol, which is used to sequentially concatenate historical user attention information and user attention information; q k represents the kth historical user attention information, q o represents the user's attention information, a represents the total number of cumulative splicing; Bert(·) represents the language representation model; The dimension classification result is updated based on the result of the joint dimension information matching.

4. The method for generating a business analysis report according to claim 2, characterized in that: The method further comprises the following steps: Based on a preset indicator mapping table, a semantic judgment model is used to perform a second verification match on the indicator classification result output by the language representation model; The expression for the second check match is: b idx ={t i |V(d t ,t i )>η,t i ∈C index ,} Where, t i Indicates the i-th indicator information in the indicator mapping table, where i is the index value; C index Indicates the indicator information contained in the preset indicator mapping table; d t represents the indicator classification result output by the language representation model; V(·) represents the semantic judgment model, which is used to perform deep dimension matching on the semantic dimension of the indicator classification result output by the language representation model from the preset indicator mapping table; Obtaining a fuzzy indicator set according to the result of the second verification matching; Based on the fuzzy indicator set, the data table, the field information and the user attention information, using the language representation model to perform secondary reasoning on the data table; The indicator classification result is updated based on the result of the secondary reasoning.

5. The method for generating a business analysis report according to claim 1, characterized in that: The method of generating background purpose analysis data based on the target dimension indicator using a preset large language model according to the user attention information includes the following steps: Based on a preset first prompt word template, generating first prompt word information according to the user attention information and the target dimension indicator; The first prompt word information is input into the preset large language model for inference to generate the background purpose analysis data.

6. The method for generating a business analysis report according to claim 1, characterized in that: The original data represents a data table corresponding to the production and operation project; the drill-down analysis of the original data based on the target dimension indicator and the time trend and the code value to obtain the numerical analysis result includes the following steps: Acquire the numerical data corresponding to the target dimension indicator from the data table; arrange all data values ​​in the numerical data in time series based on their corresponding time points; Performing month-on-month statistics and comparative statistics on the numerical data based on a preset time interval to obtain time trend analysis data; Performing incremental statistics and increase statistics on the numerical data to obtain code value analysis data; According to the time trend analysis data and the code value analysis data, single-dimensional drill-down statistical analysis and multi-dimensional drill-down statistical analysis are performed based on the target dimension index to obtain the numerical analysis result.

7. The method for generating a business analysis report according to claim 1, characterized in that: The method of analyzing the background purpose data and the numerical analysis results and integrating the data using a preset large model to obtain an empirical analysis report includes the following steps: Based on a preset second prompt word template, generating second prompt word information according to the background purpose analysis data and the numerical analysis result; The second prompt word information is input into the preset large model for data integration to obtain the experience analysis report.

8. A business analysis report generating device, characterized in that: include: A source data module is used to obtain the original data of the user's concerned information and the production and operation items to be analyzed, and extract the target dimension indicators based on the original data; A semantic understanding module, used to generate background purpose analysis data based on the target dimension indicators using a preset large language model according to the user's attention information; A data analysis module, used to perform drill-down analysis on the original data based on the target dimension indicator, time trend and code value, to obtain numerical analysis results; The report generation module is used to analyze the data and the numerical analysis results based on the background purpose, integrate the data using a preset large model, and obtain an empirical analysis report.

9. An electronic device, characterized in that: including a processor and a memory; The memory is used to store programs; The processor executes the program to implement the method according to any one of claims 1 to 7.

10. A computer storage medium storing a program executable by a processor, characterized in that: The program executable by the processor is used to implement the method according to any one of claims 1 to 7 when executed by the processor.

Images