Operation data analysis method and system, computer equipment and storage medium

Abstract

The invention provides an operation data analysis method and system, computer equipment and a storage medium, and belongs to the field of data processing, and the method comprises the steps: obtaining operation data of a plurality of shops; analyzing the operation data of the plurality of shops based on a preset short board index, and determining a short board type of each shop; sorting the plurality of stores according to the types of the short boards to obtain a processing priority sequence; according to the processing priority sequence, sequentially retrieving a strategy corresponding to each shop short board type from a preset knowledge base through an RAG technology; and constructing an operation suggestion of each shop according to the short board type of each shop and the corresponding strategy. According to the method, optimal configuration of analysis resources is realized, the timeliness of operation data analysis is further guaranteed, the generated operation suggestions have individuation and pertinence, and the operation data analysis effect is further guaranteed.

Description

Technical Field

[0001] This invention belongs to the field of data processing, and specifically relates to a business data analysis method, system, computer equipment, and storage medium. Background Technology

[0002] Operational data refers to a multi-dimensional, structured collection of data generated by retail outlets in their daily operations, reflecting their operational status and market performance. This data typically encompasses three main categories: store sales information, personnel information, and core operational information. Operational information is crucial for analysis and includes, but is not limited to: sales volume and sales revenue data such as month-on-month and year-on-year growth rates; profit indicators such as absolute gross profit; and quantitative and qualitative indicators such as inventory levels and inventory turnover ratios.

[0003] Currently, in the field of operational data analysis in industries such as tobacco, the mainstream approach relies on account managers or data analysts to manually process and judge data. This involves manually collecting and organizing operational data from various reports for each retail store, then using experience or personal analysis to identify potential operational problems and generate operational guidance suggestions. This method of generating operational data analysis and guidance suggestions not only struggles to provide personalized and targeted advice for stores with different weaknesses, but also, when there are too many stores in need, often leads to a fragmented allocation of operational analysis resources, making it difficult to identify stores that urgently need improvement, resulting in poor timeliness and low order dispatch efficiency. Summary of the Invention

[0004] To address the problems of fragmented resource allocation and poor timeliness in existing business analysis, this invention provides a business data analysis method, system, computer equipment, and storage medium.

[0005] To achieve the above objectives, the present invention provides the following technical solution: A business data analysis method, comprising: Obtain operational data from multiple stores; Based on preset weakness indicators, the operating data of multiple stores are analyzed to determine the weakness type of each store. The weakness type is the data category in the store's operating data that does not meet the corresponding standard of the weakness indicator. The multiple stores are then sorted according to the weakness type to obtain a processing priority sequence. According to the processing priority sequence, the strategies corresponding to each store's weakness type are retrieved from the preset knowledge base in sequence using RAG technology; and business suggestions for each store are constructed based on each store's weakness type and corresponding strategy.

[0006] Optionally, in the business data analysis method provided by the present invention, the types of shortcomings include one or more of the following: month-on-month increase in sales volume, absolute value of gross profit, inventory-to-sales ratio ranking, store rating, month-on-month increase in average price per item, and year-on-year increase in sales revenue.

[0007] Optionally, the operational data analysis method provided by the present invention further includes: The weakness type of each store is converted to obtain multiple query vectors; Following the priority sequence, cosine similarity is calculated for each query vector and text fragment in the knowledge base. The strategy corresponding to the weakness type is determined based on the text fragment with the highest cosine similarity result.

[0008] Optionally, the operational data analysis method provided by the present invention further includes: Following the processing priority sequence, the pre-trained LLM semantically matches the store's weakness type and corresponding strategy with the pre-set capability tags of the account manager, and assigns the account manager whose capability tag is closest to the weakness type to the store corresponding to the weakness type.

[0009] Optionally, the operational data analysis method provided by the present invention further includes: The matching degree of account managers is determined based on the type of shortcomings, the corresponding strategies, and the account manager's ability tags. The account manager with the highest matching degree is then assigned a task. When the account manager with the highest matching degree has no order assigned, the account manager with the highest matching degree will be assigned to the store corresponding to the weakness type, and the order assigned status of the account manager with the highest matching degree will be updated to "order assigned". When the account manager with the highest matching degree has already been assigned an order, the account manager with the highest matching degree will be removed from the list of account managers. The matching degree of the account manager will be determined based on the type of weakness, the corresponding strategy, and the ability tags of the account managers in the list of candidates.

[0010] Optionally, in the business data analysis method provided by the present invention, the business data of multiple stores are processed through the Dify workflow to obtain business suggestions for each store, and the account manager is assigned to the store; the Dify workflow includes an HTTP request node, a code execution node, a knowledge retrieval node, and an LLM node; Obtain operational data and account manager capability tags by requesting nodes via HTTP; The code execution nodes are used to determine the type of weakness and the priority sequence for handling each store from the operational data; The knowledge retrieval node retrieves strategies corresponding to the type of weakness for each store from a pre-set knowledge base; The LLM node assigns the account manager whose capability tag is closest to the weakness type to the store corresponding to the weakness type.

[0011] Optionally, the operational data analysis method provided by the present invention further includes: Based on the store's corresponding weaknesses, corresponding strategies, and account managers, an operational report is constructed and displayed.

[0012] This invention also provides an operational data analysis system, comprising: The business data acquisition module is used to acquire business data from multiple stores; The weakness sorting module is used to analyze the operating data of multiple stores based on preset weakness indicators, determine the weakness type of each store, where the weakness type is the data category in the store's operating data that does not meet the corresponding standard of the weakness indicator; and sort the multiple stores according to the weakness type to obtain a processing priority sequence. It is recommended to build a module that retrieves strategies corresponding to each store's weakness type from a preset knowledge base according to the processing priority sequence using RAG technology; and to build operational suggestions for each store based on its weakness type and corresponding strategy.

[0013] The present invention also provides a computer device, including a memory, a processor, and a computer program stored in the memory, wherein the processor executes the computer program to implement any of the steps in a business data analysis method.

[0014] The present invention also provides a computer-readable storage medium storing a computer program that, when loaded by a processor, can execute any step of a business data analysis method.

[0015] The business data analysis method, system, computer equipment, and medium provided by this invention have the following beneficial effects: Because the business data analysis method provided by this invention automates data analysis based on preset weakness indicators, it sorts stores according to the type and quantity of weaknesses to generate a processing priority sequence. Then, according to the priority sequence, it sequentially uses RAG technology to retrieve corresponding strategies from the knowledge base. Resources are prioritized for stores with the most prominent problems, achieving optimized allocation of analysis resources and ensuring the timeliness of business data analysis. Furthermore, by identifying the corresponding weakness for each store and matching it with appropriate strategies using RAG technology, the generated business suggestions are personalized and targeted, thus ensuring the effectiveness of business data analysis. Attached Figure Description

[0016] To more clearly illustrate the embodiments and design schemes of the present invention, the accompanying drawings required for this embodiment will be briefly described below. The drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0017] Figure 1This is a schematic diagram of an operational data analysis method provided in an embodiment of the present invention; Figure 2 This is an example of a problem-critical information extraction framework provided in an embodiment of the present invention; Figure 3 This is an example of a data statistical analysis framework provided in an embodiment of the present invention; Figure 4 This is an example of a data format conversion framework provided in an embodiment of the present invention; Figure 5 Example of a framework for generating guidance and suggestions provided for embodiments of the present invention; Figure 6 This is an example of a RAG framework provided in an embodiment of the present invention; Figure 7 This is an example of an intelligent order dispatch framework provided in an embodiment of the present invention; Figure 8 This is an example of the Agent principle provided in the embodiments of the present invention; Figure 9 This is an example of a Dify workflow provided in an embodiment of the present invention; Figure 10 This is an example of a report generation framework provided in an embodiment of the present invention; Figure 11 Example of the business guidance suggestions generated in this embodiment of the invention; Figure 12 This is an example of intelligent order dispatch results provided in an embodiment of the present invention. Detailed Implementation

[0018] To enable those skilled in the art to better understand and implement the technical solutions of the present invention, the present invention will be described in detail below with reference to the accompanying drawings and specific embodiments. The following embodiments are only used to more clearly illustrate the technical solutions of the present invention and should not be construed as limiting the scope of protection of the present invention.

[0019] Example 1 This application provides a method for analyzing business data, specifically as follows: Figure 1 As shown, it includes the following steps: Step 11: Obtain operating data from multiple stores.

[0020] Step 12: Analyze the operating data of multiple stores based on preset bottleneck indicators to determine the bottleneck type for each store. The bottleneck type is the data category in the store's operating data that does not meet the corresponding standard of the bottleneck indicator. Sort the multiple stores according to the bottleneck type to obtain a processing priority sequence. The bottleneck type includes one or more of the following: month-on-month sales growth, absolute gross profit, inventory-to-sales ratio ranking, store rating, month-on-month average price per item growth, and year-on-year sales growth.

[0021] Specifically, taking the analysis of operating data in the tobacco industry as an example, such as... Figure 2 As shown, the first step is to extract key information from the user's input question. For example, after a user inputs the question "Help me analyze the cigarette sales situation in a certain month", the first step is to extract the content needed for data analysis. For example, the time in the input question is extracted to determine which month's business situation needs to be analyzed, and then the corresponding monthly business data is obtained for statistical analysis.

[0022] Then, as Figure 3 As shown, monthly operating data is statistically analyzed from multiple perspectives based on pre-set bottleneck indicators to determine which stores meet the indicator requirements. Bottleneck indicators refer to a series of pre-set key performance parameters used to assess the health of a store's operations, such as, but not limited to, month-on-month sales growth and absolute gross profit. Bottleneck type refers to the specific operational dimension that a store has failed to meet expectations or is performing relatively poorly, identified by comparing and analyzing the store's operating data with the aforementioned bottleneck indicators. In other words, one bottleneck type corresponds to one poorly performing bottleneck indicator. For example, regarding the cigarette inventory data, sales volume and average price per carton are analyzed, and the store ranking last, the bottom 10% in month-on-month growth, and the bottom 10% in year-on-year growth are selected from multiple stores. For general store operation data, stores are selected based on sales revenue, gross profit, and inventory-to-sales ratio. For store specification inventory data, specifications nearing stockout and severely slow-moving specifications are selected based on inventory-to-sales ratio. For indicator score query data, stores with unsatisfactory evaluation responses are selected based on store scores.

[0023] After that, as Figure 4 As shown, based on the statistical analysis of the data, the statistical analysis results are converted into the form of stores and their corresponding shortcomings. The total number of stores with shortcomings is counted, and then specific shortcomings are counted for each store. Finally, the stores are sorted according to the number of shortcomings, and the results are output in the form of store number + shortcomings content.

[0024] Step 13: According to the processing priority sequence, retrieve the strategies corresponding to each store's weakness type from the preset knowledge base in sequence using RAG technology; construct business suggestions for each store based on each store's weakness type and corresponding strategy.

[0025] Step 13 includes: Step 131: Convert the weakness type of each store to obtain multiple query vectors.

[0026] Step 132: Following the processing priority sequence, calculate the cosine similarity between the query vector and the text fragments in the knowledge base one by one, and determine the strategy corresponding to the shortcoming type based on the text fragment with the highest cosine similarity calculation result.

[0027] Specifically, such as Figure 5 As shown, a strategy knowledge base corresponding to each weakness is pre-built. Based on each store's weakness, the corresponding strategy is matched through knowledge base retrieval. Then, the specific content of each strategy is retrieved from the strategy details knowledge base based on the strategy name. Furthermore, the weakness, strategy, and strategy details can all be input into the LLM model, allowing the LLM model to analyze, summarize, and generate corresponding business guidance suggestions.

[0028] Knowledge base retrieval can be achieved based on Retrieval Augmented Generation (RAG) technology. RAG is a technical framework that enhances the generation capabilities of large models by introducing external knowledge bases. Its core principle is to retrieve real-time or private information relevant to the user's query and use it as context input to the large model, thereby improving the accuracy, timeliness, and professionalism of the generated content. A framework diagram of RAG technology is shown below. Figure 6 As shown, the process includes data preparation and index building, retrieval, and generation. First, the text document is segmented into text blocks. Then, an embedding model is used for vectorization and index optimization, storing the segmented document content in a vector database. Next, the user's query is converted into a vector using the embedding model, and the most similar text fragments are retrieved from the vector database. A hybrid retrieval method can be used to determine the closest text fragments, integrating the advantages of traditional keyword retrieval and semantic vector retrieval to achieve complementary precise matching and semantic understanding. Finally, real-time or private information related to the user's query, text fragments calculated based on cosine similarity, and the query itself are input into the LLM model, which generates the answer based on the context.

[0029] Step 14: Following the processing priority sequence, the pre-trained LLM is used to semantically match the store's weakness type and corresponding strategy with the capability tags pre-set by the account manager. The account manager whose capability tag is closest to the weakness type is then assigned to the store corresponding to the weakness type.

[0030] Step 14 includes: Step 141: Determine the matching degree of account managers based on the type of weakness, the corresponding strategy, and the account manager's ability tags, and make a judgment on the order assignment for the account manager with the highest matching degree.

[0031] Step 142: When the account manager with the highest matching degree has no order assigned, assign the account manager with the highest matching degree to the store corresponding to the weakness type, and update the order assignment status of the account manager with the highest matching degree to "order assigned".

[0032] Step 143: When the account manager with the highest matching degree has been assigned an order, remove the account manager with the highest matching degree from the candidate list of account managers, and determine the matching of account managers based on the type of weakness, the corresponding strategy, and the ability tags of account managers in the candidate list.

[0033] Specifically, when performing semantic matching between stores and account managers, the list of account manager candidates for the next store to be assigned orders is determined based on the order assignment status. The candidate list is dynamically updated, and the account manager who has been assigned an order is deleted from the candidate list after each order is assigned. This process continues until all personnel have been assigned orders, at which point the candidate list is initialized and a new round of order assignment begins, thus avoiding the problem of uneven order assignment among account managers.

[0034] The Dify workflow processes operational data from multiple stores to generate operational suggestions for each store and assigns account managers to them. The Dify workflow includes HTTP request nodes, code execution nodes, knowledge retrieval nodes, LLM nodes, parameter extraction nodes, and Agent nodes. The HTTP request node retrieves operational data, account manager information, updates account manager information, and saves files, such as retrieving operational data and account manager capability tags. The code execution node performs data format conversion, content concatenation, and deconstruction of thought processes to determine the weakness type and processing priority sequence for each store from the operational data. The knowledge retrieval node retrieves the corresponding strategy for each store's weakness type from a pre-set knowledge base. The LLM node assigns the account manager whose capability tags are closest to the weakness type to the corresponding store. The parameter extraction and Agent nodes extract time parameters from the input and standardize the parameter format.

[0035] Specifically, such as Figure 7 As shown, once the strategy corresponding to the weakness type of each store is determined, the information file of the customer manager is read from the pre-built customer manager database. Based on the store's weakness and the corresponding business strategy, the most suitable customer manager is selected for task assignment, generating the task assignment results. In matching customer managers, not only are their corresponding capability tags considered, but also the balance of task assignment is taken into account. Each time a task is assigned, the assigned customer manager is removed from the list of pending customer managers until all customer managers have completed their assignments, then a new round of assignments begins, ensuring a uniform distribution of task assignment results. When there are multiple stores, customer managers are matched first to the store with the most weakness types, ensuring that the most severely affected stores receive priority in improving their business. Furthermore, once a customer manager is successfully matched, the customer manager's information file is updated, and the assigned customer manager is removed to avoid assigning the same customer manager to too many stores.

[0036] The Agent in this invention is an intelligent system capable of perceiving the environment, making autonomous decisions, and executing actions to achieve specific goals. Its core principle can be summarized as a closed-loop mechanism of "perception-decision-execution," such as... Figure 8 As shown, the agent integrates tools, memory, planning and decision-making, and action. Memory includes both short-term and long-term memory, used to support planning and decision-making, as well as reflection during the planning process. Planning and decision-making includes functions such as reflection, self-criticism, thought chains, and sub-goal decomposition. Tools include calendars, calculators, code interpreters, and search engines, used to assist in implementing specific actions. In summary, the agent perceives the environment by integrating tools, utilizes memory to store experiences, makes complex planning decisions, and ultimately executes actions to interact with the environment and achieve goals.

[0037] The matching of account managers can be achieved through LLM technology. The model performs a comprehensive analysis based on the input of the store's weaknesses and strategies, combined with the account manager's competency tags, to select the most suitable account manager for the current store. To ensure that a suitable candidate is still available if the assigned account manager is unable to participate in the work, multiple backup account managers are added on top of the assigned order, for example, two backup account managers, for users to comprehensively evaluate based on actual needs.

[0038] A complete operational data analysis methodology can be built on, for example... Figure 9 The Dify workflow shown uses nodes such as document extractor, agent, parameter extractor, HTTP, code execution, knowledge retrieval, LLM, iteration, and template conversion to identify the shortcomings in the store's operating data, search for corresponding strategies for those shortcomings, and match the store with the corresponding account manager.

[0039] Specifically, the Dify workflow starts with a question and basic information input by the customer. Then, a parameter extraction node extracts the customer's needs from the basic information and uses an Agent node, such as a deepseek model based on the "Functioncalling" strategy, to extract time parameters and standardize the parameter format.

[0040] Next, the code execution node analyzes the store operation data within the time parameter range and converts the format of the analyzed data to determine the weakness type and processing priority sequence for each store. After the stores are sorted, multiple parameter extraction nodes extract the store number, customer-related information, and the weakness type corresponding to the store number. Then, the knowledge retrieval node matches the store with the strategy based on a preset strategy matching mechanism. Next, the HTTP node extracts customer manager-related information, and the LLM node performs semantic matching between the store-strategy matching results and the customer manager-related information to achieve customer manager recommendation. Then, the parameter extraction node determines the name of the customer manager assigned to the store, and the HTTP node updates the customer manager information, removing the assigned customer manager from the candidate list. From the remaining customer managers, one or more most suitable alternative customer managers are selected. Finally, based on the customer manager, strategy details, store, and recommendations, the store operation data strategy and customer manager recommendation file are constructed through title generation and content concatenation, generating and saving the final report.

[0041] Step 15: Based on the store's corresponding weakness type, corresponding strategy, and account manager, construct an operation report and output the operation report for display.

[0042] Specifically, such as Figure 10 As shown, after determining the type of weakness, the corresponding strategy, and the account manager for the store, the system generates a business guidance suggestion title and a smart order dispatch title. Then, it combines the title, basic customer information, weakness analysis, strategy suggestions, and smart order dispatch results. Finally, it calls the interface for generating and saving a pre-written fixed-format PDF document to save the generated report in the specified location.

[0043] In summary, most of the current operational guidance suggestions for the tobacco industry are generated according to fixed templates based on manual data analysis, lacking diversity and being time-consuming; order assignments are mostly based on regional divisions and staff availability based on past experience, without scientific analysis, and lack specificity.

[0044] The business data analysis method provided by this invention, based on Dify's business guidance suggestion generation and intelligent order dispatch system, can determine the shortcomings and severity of each store based on a large amount of data analysis, and then provide targeted and reasonable suggestions. It can make reasonable recommendations and dispatch orders based on the comprehensive capabilities of account managers, and can conduct targeted order dispatch according to the specific problems of the stores, thereby improving the rationality and scientific nature of order dispatch. At the same time, the addition of AI analysis can improve the efficiency of suggestion generation and order dispatch, and free up manpower.

[0045] Specifically, the operational data analysis method provided by this invention statistically analyzes tobacco business data to identify the weaknesses of each store, ranks the stores according to the number of weaknesses, performs strategy matching and content matching through knowledge base retrieval, and then generates complete operational guidance suggestions by summarizing this information through a model. Based on existing account manager capabilities, combined with store weaknesses and corresponding strategies, the model recommends the most suitable account manager for each store, while also recommending two alternative account managers. The account manager information file is dynamically updated after each order assignment, thus achieving both even order assignment and assigning the most capable account manager to the store with the most problems.

[0046] Example 2 Based on Example 1, such as Figure 11 and Figure 12 As shown, this application also provides a specific example of tobacco business data analysis: For store 620102107046, its monthly operating data for May 2025 was extracted. Weaknesses were identified, including: lowest sales volume (last in the rankings), lowest month-on-month sales volume growth rate (last in the 10% range), lowest year-on-year sales volume growth rate (last in the 10% range), lowest month-on-month average price per carton growth rate (last in the 10% range), lowest sales revenue (last in the rankings), lowest year-on-year sales revenue growth rate (last in the 10% range), lowest gross profit margin (last in the rankings), lowest month-on-month gross profit margin growth rate (last in the 10% range), lowest year-on-year gross profit margin growth rate (last in the 10% range), lowest inventory-to-sales ratio (last in the 10% range), and an unsatisfactory store rating. Subsequently, based on the RAG technology in the Dify workflow of this invention, corresponding suggestions for cigarette display and reasonable ordering were retrieved from the business strategy knowledge base and input into the LLM model, outputting as follows: Figure 11 As shown.

[0047] Once the shortcomings of store 620102107046 and the corresponding strategies are determined, such as Figure 12 As shown, the Agent technology in the Dify workflow of this invention matches multiple account managers from the account manager database. For example, based on data analysis capabilities, system processing capabilities, and market research capabilities, one account manager and two alternative account managers are matched to construct the corresponding order dispatch results.

[0048] Example 3 This application also provides an operational data analysis system, including: The business data acquisition module is used to acquire business data from multiple stores; The weakness sorting module is used to analyze the operating data of multiple stores based on preset weakness indicators, determine the weakness type of each store, where the weakness type is the data category in the store's operating data that does not meet the corresponding standard of the weakness indicator; and sort the multiple stores according to the weakness type to obtain a processing priority sequence. It is recommended to build a module that retrieves strategies corresponding to each store's weakness type from a preset knowledge base according to the processing priority sequence using RAG technology; and to build operational suggestions for each store based on its weakness type and corresponding strategy.

[0049] The present invention also provides a computer device, including a memory, a processor, and a computer program stored in the memory. The processor executes the computer program to implement the steps in an embodiment of a business data analysis method. Specific implementation methods can be found in the method embodiments, and will not be repeated here.

[0050] Furthermore, the present invention also provides a non-transitory computer-readable storage medium containing instructions on which a computer program is stored. For example, a memory containing instructions that can be executed by a processor of a computer device to perform the aforementioned method. For example, the non-transitory computer-readable storage medium may be a ROM, random access memory (RAM), CD-ROM, magnetic tape, floppy disk, and optical data storage device, etc. When the computer program is executed by the processor, it can implement the steps in an embodiment of a business data analysis method. Specific implementation methods can be found in the method embodiments, which will not be repeated here.

[0051] Those skilled in the art will understand that embodiments of the present invention can provide methods, systems, or computer program products. Therefore, the present invention can take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, the present invention can take the form of a computer program product embodied on one or more computer-usable storage media (including, but not limited to, disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0052] This invention is described with reference to flowchart illustrations and / or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of the invention. It will be understood that each block of the flowchart illustrations and / or block diagrams, as well as combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, special-purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, generate instructions for implementing the flowchart. Figure 1 One or more processes and / or boxes Figure 1 A device that provides the functions specified in one or more boxes.

[0053] These computer program instructions may also be stored in a computer-readable storage medium that can direct a computer or other programmable data processing device to function in a particular manner, such that the instructions stored in the computer-readable storage medium produce an article of manufacture including instruction means, which are implemented in a process Figure 1 One or more processes and / or boxes Figure 1 The function specified in one or more boxes.

[0054] These computer program instructions may also be loaded onto a computer or other programmable data processing equipment to cause a series of operational steps to be performed on the computer or other programmable equipment to produce a computer-implemented process, thereby providing instructions that execute on the computer or other programmable equipment for implementing the process. Figure 1 One or more processes and / or boxes Figure 1 The steps of the function specified in one or more boxes.

[0055] It should be noted that the specific embodiments described above enable those skilled in the art to more fully understand the present invention, but do not limit the present invention in any way. Therefore, although the present invention has been described in detail in this specification and embodiments, those skilled in the art should understand that modifications or equivalent substitutions can still be made to the present invention; and all technical solutions and improvements that do not depart from the spirit and scope of the present invention are covered within the protection scope of the present invention patent. No reference numerals in the claims should be construed as limiting the scope of the claims. Any simple variations or equivalent substitutions of technical solutions that can be readily obtained by those skilled in the art within the scope of the technology disclosed in the present invention are within the protection scope of the present invention.

Claims

1. A method of analyzing business data, characterized by, The method comprises: obtaining operation data of a plurality of stores; analyzing the operation data of the plurality of stores based on preset short board indicators to determine a short board type of each store, wherein the short board type is a data category in the operation data of the store that does not meet the corresponding standard of the short board indicators; and sorting the plurality of stores according to the short board type to obtain a processing priority sequence; according to the processing priority sequence, retrieving a strategy corresponding to the short board type of each store from a preset knowledge base in sequence through RAG technology; and constructing an operation suggestion for each store according to the short board type and the corresponding strategy of each store.

2. The method of claim 1, wherein, The short board type includes one or more of sales growth, gross profit absolute value, inventory-sales ratio ranking, store rating, average price growth, and sales growth.

3. The method of claim 1, wherein, Retrieving the strategy corresponding to the short board type of each store from the preset knowledge base in sequence through RAG technology comprises: converting the short board type of each store to obtain a plurality of query vectors; according to the processing priority sequence, performing cosine similarity calculation between the query vector and the text segment in the knowledge base one by one based on the cosine similarity calculation result, and determining the strategy corresponding to the short board type based on the text segment with the highest cosine similarity calculation result.

4. The method of claim 1, wherein, After retrieving the strategy corresponding to the short board type of each store from the preset knowledge base in sequence through RAG technology, the method further comprises: according to the processing priority sequence, performing semantic matching between the short board type and the corresponding strategy and the ability label of the client manager pre-set by the pre-trained LLM one by one, and assigning the client manager with the closest ability label to the store corresponding to the short board type.

5. The method of claim 4, wherein, Performing semantic matching between the short board type and the corresponding strategy and the ability label of the client manager pre-set by the pre-trained LLM one by one, and assigning the client manager with the closest ability label to the store corresponding to the short board type comprises: determining the matching degree of the client manager according to the short board type and the corresponding strategy and the ability label of the client manager, and determining the order of the client manager with the highest matching degree; when the order of the client manager with the highest matching degree is not assigned, assigning the client manager with the highest matching degree to the store corresponding to the short board type, and updating the order of the client manager with the highest matching degree to assigned; when the order of the client manager with the highest matching degree is assigned, removing the client manager with the highest matching degree from the candidate list of the client manager, and determining the matching degree of the client manager according to the short board type and the corresponding strategy and the ability label of the client manager in the candidate list.

6. The method of claim 5, wherein, processing the operation data of the plurality of stores through a Dify workflow to obtain an operation suggestion for each store, and assigning a client manager to the store; the Dify workflow comprises an HTTP request node, a code execution node, a knowledge retrieval node, and an LLM node; obtaining the operation data and the ability label of the client manager through the HTTP request node; determining the short board type and the processing priority sequence of each store from the operation data through the code execution node; retrieve, by the knowledge retrieval node, a strategy corresponding to the short-board type of each store from a preset knowledge base; assign, by the LLM node, the customer manager closest to the capability label and the short-board type to the store corresponding to the short-board type.

7. The method of claim 4, wherein the method further comprises: After assigning the customer manager closest to the capability label and the short-board type to the store corresponding to the short-board type, further comprising: construct an operation report according to the short-board type, the corresponding strategy and the customer manager of the store, and output the operation report for display.

8. An operating data analysis system characterized by, Comprise: an operation data acquisition module configured to acquire operation data of a plurality of stores; a short-board sorting module configured to analyze the operation data of the plurality of stores based on preset short-board indicators, determine a short-board type of each store, wherein the short-board type is a data category in the operation data of the store that does not meet a corresponding standard of the short-board indicators, and sort the plurality of stores according to the short-board type to obtain a processing priority sequence; a suggestion construction module configured to retrieve a strategy corresponding to the short-board type of each store from a preset knowledge base in sequence by RAG technology according to the processing priority sequence, and construct an operation suggestion for each store according to the short-board type and the corresponding strategy of each store.

9. A computer device comprising a memory, a processor, and a computer program stored on the memory, wherein the computer program comprises instructions that, when executed by the processor, cause the processor to perform the method of any one of claims 1-8. The processor executes the computer program to implement the steps of the operation data analysis method of any one of claims 1 to 7.

10. A computer-readable storage medium having stored thereon a computer program, characterized in that, The computer program can implement the steps of the operation data analysis method of any one of claims 1 to 7 when loaded by the processor.

Images