Network resource data processing method and device, electronic equipment and storage medium

Abstract

The invention provides a network resource data processing method and device, electronic equipment and a storage medium, and belongs to the technical field of data processing.The method comprises the steps that natural language query related to network resource data analysis is obtained; inputting the natural language query into a pre-constructed large model, determining a target symbolization formula corresponding to the natural language query by the large model, identifying the target symbolization formula to obtain an identification result, generating a business thinking chain based on the identification result, generating an instantiated cue word based on the business thinking chain, and generating a prompt word based on the instantiation. Processing the target network resource data, and outputting a data analysis result corresponding to the natural language query; wherein the large model is obtained by training based on the sample natural language query and a data analysis result label corresponding to the sample natural language query. The method improves the efficiency of network resource data analysis, is high in flexibility, and is suitable for complex application scenes.

Description

Technical Field

[0001] This invention relates to the field of data processing technology, and in particular to a method, apparatus, electronic device, and storage medium for processing network resource data. Background Technology

[0002] In the field of communication network operation and maintenance management, network resource data analysis is crucial. With the ever-increasing volume of network resource data, how to achieve efficient analysis of network resource data in complex application scenarios is a key issue that urgently needs to be addressed. Currently, existing network resource data processing methods rely on pre-configured reports or customized data query tools, which are inefficient, lack flexibility, and are difficult to adapt to complex application scenarios. Summary of the Invention

[0003] This invention provides a network resource data processing method, apparatus, electronic device, and storage medium to address the shortcomings of existing network resource data processing methods, such as low efficiency, poor flexibility, and difficulty in adapting to complex application scenarios.

[0004] This invention provides a method for processing network resource data, comprising: Obtain natural language queries related to network resource data analysis; The natural language query is input into a pre-built large model, which determines the target symbolic formula corresponding to the natural language query, identifies the target symbolic formula, obtains the identification result, generates a business thinking chain based on the identification result, generates instantiated prompt words based on the business thinking chain, processes the target network resource data based on the instantiated prompt words, and outputs the data analysis result corresponding to the natural language query. The large model is trained based on sample natural language queries and the corresponding data analysis result labels of the sample natural language queries.

[0005] In some embodiments, determining the target symbolic formula corresponding to the natural language query includes: The natural language query is matched with a set of preset symbolic formulas to obtain the matching results. The symbolic formulas include network resource entity symbols, network resource entity attribute symbols, and operation relation symbols. Based on the matching results, the target symbolic formula is determined from the plurality of symbolic formulas.

[0006] In some embodiments, identifying the target symbolic formula and obtaining the identification result includes: Based on the descriptive prompts of the target symbolic formula and the pre-built network resource knowledge base, the target symbolic formula is identified to obtain the identification result; The identification results include network resource entities, the attributes of network resource entities, and the relationships between different network resource entities.

[0007] In some embodiments, generating instantiated prompt words based on the business thought chain includes: Determine the execution prompt and execution example for the target symbolized formula; Based on the execution prompt words, the execution examples, and the business thought chain, instantiated prompt words are generated. The instantiated prompt words include the names and attributes of network resource entities, as well as the relationships between different network resource entities.

[0008] In some embodiments, processing the target network resource data includes: Construct a data link, wherein the data link uses network resource entities as nodes and the association relationships between different network resource entities as edges; Based on the data link, target network resource data is extracted from the network resource database; The target network resource data is filtered to determine the number of target network resource entities; Based on the attributes of the target network resource entities, the number of the target network resource entities is classified and statistically analyzed to obtain statistical results; The statistical results are formatted to obtain the data analysis results.

[0009] In some embodiments, after inputting the natural language query into a pre-built large model, the method further includes: Real-time monitoring of the data processing status of the large model; If the data processing status of the large model is determined to be abnormal, the network resource knowledge base is updated, or the computational strategy parameters of the large model are updated.

[0010] In some embodiments, the training process of the large model includes: Obtain sample natural language queries related to network resource data analysis; Determine the data analysis result labels corresponding to the natural language queries of the samples; The sample natural language query is input into a pre-built initial large model to obtain the data analysis and prediction results corresponding to the sample natural language query output by the initial large model. Based on the data analysis and prediction results corresponding to the sample natural language query, and the data analysis result labels corresponding to the sample natural language query, a loss function value is calculated. Based on the loss function value, the parameters of the initial large model are iteratively optimized to obtain the large model.

[0011] The present invention also provides a network resource data processing device, comprising: The acquisition unit is used to acquire natural language queries related to network resource data analysis. The processing unit is configured to input the natural language query into a pre-built large model, determine the target symbolization formula corresponding to the natural language query by the large model, identify the target symbolization formula, obtain the identification result, generate a business thinking chain based on the identification result, generate instantiated prompt words based on the business thinking chain, process the target network resource data based on the instantiated prompt words, and output the data analysis result corresponding to the natural language query. The large model is trained based on sample natural language queries and the corresponding data analysis result labels of the sample natural language queries.

[0012] The present invention also provides an electronic device, including a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the program to implement the network resource data processing method described above.

[0013] The present invention also provides a non-transitory computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the network resource data processing method as described above.

[0014] The network resource data processing method, apparatus, electronic device, and storage medium provided by this invention acquire natural language queries related to network resource data analysis; input the natural language queries into a pre-built large model, where the large model determines the target symbolization formula corresponding to the natural language query; identify the target symbolization formula to obtain the identification result; generate a business thinking chain based on the identification result; generate instantiated prompt words based on the business thinking chain; process the target network resource data based on the instantiated prompt words; and output the data analysis result corresponding to the natural language query. This improves the efficiency of network resource data analysis, is highly flexible, and is suitable for complex application scenarios. Attached Figure Description

[0015] To more clearly illustrate the technical solutions in this invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are some embodiments of this invention. For those skilled in the art, other drawings can be obtained from these drawings without creative effort.

[0016] Figure 1 This is a flowchart illustrating the network resource data processing method provided in an embodiment of the present invention.

[0017] Figure 2 This is a flowchart illustrating the process of determining the target symbolization formula corresponding to a natural language query, provided in an embodiment of the present invention.

[0018] Figure 3 This is a flowchart illustrating the training process of a large model provided in an embodiment of the present invention.

[0019] Figure 4 This is a schematic diagram of the network resource data processing device provided in an embodiment of the present invention.

[0020] Figure 5 This is a schematic diagram of the structure of the electronic device provided in an embodiment of the present invention. Detailed Implementation

[0021] To make the objectives, technical solutions, and advantages of this invention clearer, the technical solutions of this invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some, not all, of the embodiments of this invention. All other embodiments obtained by those skilled in the art based on the embodiments of this invention without creative effort are within the scope of protection of this invention.

[0022] Currently, existing network resource data processing methods, in complex application scenarios involving multiple resource entities, typically involve first querying data and then analyzing it. There is a disconnect between the two actions. For each resource entity, a scenario needs to be customized. Similar scenarios require users to perform multiple interface operations and manual analysis, which is time-consuming and labor-intensive. Furthermore, the support for mobile clients is poor.

[0023] To address this, embodiments of the present invention provide a network resource data processing method, apparatus, electronic device, and storage medium. This method involves acquiring natural language queries related to network resource data analysis; inputting the natural language queries into a pre-built large model; having the large model determine the target symbolization formula corresponding to the natural language query; identifying the target symbolization formula to obtain the identification result; generating a business thinking chain based on the identification result; generating instantiated prompt words based on the business thinking chain; processing the target network resource data based on the instantiated prompt words; and outputting the data analysis result corresponding to the natural language query. This improves the efficiency of network resource data analysis, offers high flexibility, and is suitable for complex application scenarios.

[0024] Figure 1 This is a flowchart illustrating a network resource data processing method provided in an embodiment of the present invention. Figure 1 As shown, a method for processing network resource data is provided, including the following steps: step 110 and step 120. This method's steps are merely one possible implementation of the present invention.

[0025] Step 110: Obtain natural language queries related to network resource data analysis; Natural language queries refer to instructions or questions issued by operations and maintenance personnel using spoken or written language from their daily work to describe their data analysis needs; natural language queries contain complete, multi-step analysis logic. For example, data centers with more than 5 baseband units (BBUs) are categorized and statistically analyzed by level.

[0026] Optionally, raw natural language queries input by the user can be received through one or more front-end interaction modules. These modules could be text input boxes in web applications, voice recognition interfaces in mobile operations and maintenance applications, or chatbots integrated into enterprise communication software.

[0027] Optionally, the original natural language query can be preprocessed by data cleaning, speech recognition, word segmentation, etc., to obtain the natural language query.

[0028] Step 120: Input the natural language query into the pre-built large model, the large model determines the target symbolization formula corresponding to the natural language query, identifies the target symbolization formula, obtains the identification result, generates a business thinking chain based on the identification result, generates instantiated prompt words based on the business thinking chain, processes the target network resource data based on the instantiated prompt words, and outputs the data analysis result corresponding to the natural language query. The large model is trained based on sample natural language queries and the corresponding data analysis result labels of the sample natural language queries.

[0029] Among them, the target symbolic formula is determined from a number of preset symbolic formulas. A symbolic formula is an intermediate logical representation that accurately expresses business intent. It uses general symbols to represent variable resource entities, attributes and operation relationships. Each symbolic formula is applicable to a type of business scenario with similar analysis logic.

[0030] The business thinking chain includes a structured, multi-step intermediate logical reasoning process. For example, the identified entities include BBU and data center. The BBU number is associated with the data center number, and the data center is classified and statistically analyzed according to its type.

[0031] Optionally, the large model includes a symbolization layer, a recognition layer, a thought chain generation layer, a prompt word generation layer, and an analysis layer; the symbolization layer is used to determine the target symbolization formula corresponding to the natural language query; the recognition layer is used to recognize the target symbolization formula and obtain the recognition result; the thought chain generation layer is used to generate a business thought chain based on the recognition result; the prompt word generation layer is used to generate instantiated prompt words based on the business thought chain; and the analysis layer is used to process the target network resource data based on the instantiated prompt words and output the data analysis results corresponding to the natural language query.

[0032] In this embodiment of the invention, a natural language query related to network resource data analysis is obtained; the natural language query is input into a pre-built large model, the large model determines the target symbolization formula corresponding to the natural language query, the target symbolization formula is identified, the identification result is obtained, a business thinking chain is generated based on the identification result, instantiated prompt words are generated based on the business thinking chain, the target network resource data is processed based on the instantiated prompt words, and the data analysis result corresponding to the natural language query is output. This improves the efficiency of network resource data analysis, is highly flexible, and is suitable for complex application scenarios.

[0033] Figure 2 This is a flowchart illustrating the process of determining the target symbolic formula corresponding to a natural language query, as provided in an embodiment of the present invention. Figure 2 As shown, in some embodiments, step 120, determining the target symbolic formula corresponding to the natural language query, includes: Step 121: Match the natural language query with multiple preset symbolic formulas to obtain the matching results. The symbolic formulas include network resource entity symbols, network resource entity attribute symbols, and operation relation symbols. Optionally, the symbolic formula library can be updated, and symbolic formulas can be added, deleted, or modified; natural language queries can be matched with multiple preset symbolic formulas in the symbolic formula library.

[0034] Optionally, the matching results may include multiple candidate symbolic formulas.

[0035] Step 122: Based on the matching results, determine the target symbolic formula from multiple symbolic formulas.

[0036] Symbolic formulas define a general logical analysis paradigm. When the system receives a natural language query, instead of hard-matching keywords like "BBU" or "data center," it matches the underlying analytical logic structure of the query. The system only needs to implement the parsing and execution capabilities of the symbolic formulas to generalize and support all business analyses that conform to this logical structure, greatly improving the system's applicability and flexibility, and avoiding endless customization.

[0037] In some embodiments, identifying the target symbolic formula and obtaining the identification result includes: Based on the descriptive prompts of the target symbolic formula and the pre-built network resource knowledge base, the target symbolic formula is identified, and the identification result is obtained. The identification results include network resource entities, the attributes of network resource entities, and the relationships between different network resource entities.

[0038] The network resource knowledge base includes at least a network resource entity knowledge base and a network resource relationship knowledge base. The network resource entity knowledge base includes at least network resource entities and their attributes, such as network equipment, equipment location, network access time, and ownership. The network resource relationship knowledge base includes the relationships between various network resource entities, such as the data center to which the BBU belongs.

[0039] Optionally, based on the descriptive prompts of the target symbolized formula, such as entity and attribute prompts, and in conjunction with a network resource entity knowledge base, the entity symbols and entity attribute symbols of the target symbolized formula are identified to determine the network resource entity represented by the entity symbol and the attribute of the network resource entity represented by the entity attribute symbol. For example, the network resource entities BBU and data center are identified, and the attribute of the data center includes data center level.

[0040] Optionally, based on the descriptive prompts of the target symbolic formula and the network resource relationship knowledge base, the relationships between different network resource entities, such as the relationship between BBU and data center, can be identified.

[0041] In some embodiments, instantiated prompt words are generated based on the business thought chain, including: Determine the execution prompt and execution example for the target symbolic formula; Based on execution prompts, execution examples, and business thought chains, instantiated prompts are generated. The instantiated prompts include the names and attributes of network resource entities, as well as the relationships between different network resource entities.

[0042] Optionally, based on the execution prompt words, execution examples, and business thought chains, the target execution process template is invoked to generate instantiated prompt words. The target execution process template is determined from multiple preset general execution process templates.

[0043] In this embodiment of the invention, by determining the execution prompt words and execution examples of the target symbolic formula, and generating instantiated prompt words based on the execution prompt words, execution examples, and business thought chains, the accuracy and reliability of the generated instructions are significantly enhanced.

[0044] In some embodiments, processing the target network resource data includes: Construct a data link, with network resource entities as nodes and the relationships between different network resource entities as edges; Based on the data link, target network resource data is extracted from the network resource database; Filter the target network resource data to determine the number of target network resource entities; Based on the attributes of the target network resource entities, the number of target network resource entities is classified and statistically analyzed to obtain statistical results; The statistical results are formatted to obtain the data analysis results.

[0045] Optionally, along the data link, a data retrieval request is initiated to the underlying network resource database, such as a relational database or a graph database, to extract the raw data of all relevant entities. For example, the data link includes two nodes: BBU and data center, and a related edge; information about all BBUs and their corresponding data centers can be extracted from the database.

[0046] Optionally, the statistical results can be converted into natural language descriptions or visual charts.

[0047] In this embodiment of the invention, by constructing a data link, target network resource data is extracted from the network resource database based on the data link; the target network resource data is filtered to determine the number of target network resource entities; based on the attributes of the target network resource entities, the number of target network resource entities is classified and statistically analyzed to obtain statistical results. This ensures the logical rigor of the complex calculation process, improves the accuracy of the calculation results, and enhances the transparency and debuggability of the entire analysis process.

[0048] In some embodiments, after inputting the natural language query into a pre-built large model, the method further includes: Real-time monitoring of the data processing status of large models; If the data processing status of the large model is found to be abnormal, update the network resource knowledge base or update the computational strategy parameters of the large model.

[0049] Abnormal states can be mainly divided into the following two categories: Knowledge mismatch anomaly: triggered when the planning of the large model contradicts the facts in the network resource knowledge base, such as missing entity relationships or non-existent entity attributes.

[0050] Anomalies in computational strategy: This occurs when the execution plan of a large model is logically feasible but infeasible or inefficient in reality. For example, the data volume exceeds the limit, and the query is expected to return a large amount of raw data far exceeding the memory processing capacity; or the computational complexity is too high, and the query is expected to cause the database to lock up for a long time or exhaust resources.

[0051] Optionally, the system can check the input, output, and intermediate states of each step in real time; when generating the business thought chain: check whether the entities, attributes, and relationships in the thought chain exist in the network resource knowledge base; when generating instantiation prompts: check whether the generated instantiation prompts conform to the syntax rules and estimate their computational complexity and resource consumption; when performing data processing: monitor the database response time, the amount of data returned, and whether any exceptions are thrown.

[0052] In this embodiment of the invention, by monitoring the data processing status of the large model in real time, and updating the network resource knowledge base or the computational strategy parameters of the large model when it is determined that the data processing status of the large model is abnormal, the robustness of the large model is improved, enabling the large model to provide reliable services in a long-term and stable manner in a dynamically changing network environment, thereby improving the user experience.

[0053] Figure 3 This is a flowchart illustrating the training process of a large model provided in an embodiment of the present invention. Figure 3 As shown, in some embodiments, the training process for a large model includes: Step 310: Obtain sample natural language queries related to network resource data analysis; Step 320: Determine the data analysis result labels corresponding to the sample natural language query; Step 330: Input the sample natural language query into the pre-built initial large model to obtain the data analysis and prediction results corresponding to the sample natural language query output by the initial large model; Step 340: Based on the data analysis and prediction results corresponding to the sample natural language queries, and the data analysis result labels corresponding to the sample natural language queries, calculate the loss function value. Based on the loss function value, iteratively optimize the parameters of the initial large model to obtain the large model.

[0054] Optionally, the sample natural language query is input into a pre-built initial large model to obtain the data analysis and prediction results corresponding to the sample natural language query output by the initial large model, including: The sample natural language query is input into the initial large model, which determines the sample symbolization formula corresponding to the sample natural language query. The sample symbolization formula is identified to obtain the prediction and recognition result. Based on the prediction and recognition result, a sample business thinking chain is generated. Based on the sample business thinking chain, instantiated sample prompt words are generated. Based on the instantiated sample prompt words, the sample network resource data is processed to output the data analysis and prediction result corresponding to the sample natural language query.

[0055] In this embodiment of the invention, by inputting sample natural language queries into a pre-built initial large model, the data analysis and prediction results corresponding to the sample natural language queries output by the initial large model are obtained; based on the data analysis and prediction results and data analysis result labels corresponding to the sample natural language queries, the loss function value is calculated; based on the loss function value, the parameters of the initial large model are iteratively optimized, thereby improving the deep logical reasoning and generalization capabilities of the large model.

[0056] The network resource data processing apparatus provided in the embodiments of the present invention is described below. The network resource data processing apparatus described below can be referred to in correspondence with the network resource data processing method described above.

[0057] Figure 4 This is a schematic diagram of the network resource data processing device provided in an embodiment of the present invention, as shown below. Figure 4 As shown, the network resource data processing device 400 includes: Acquisition unit 410 is used to acquire natural language queries related to network resource data analysis; The processing unit 420 is used to input natural language queries into a pre-built large model, determine the target symbolization formula corresponding to the natural language query by the large model, identify the target symbolization formula, obtain the identification result, generate a business thinking chain based on the identification result, generate instantiated prompt words based on the business thinking chain, process the target network resource data based on the instantiated prompt words, and output the data analysis result corresponding to the natural language query. The large model is trained based on sample natural language queries and the corresponding data analysis result labels of the sample natural language queries.

[0058] Optionally, determining the target symbolic formula corresponding to the natural language query includes: The natural language query is matched with multiple preset symbolic formulas to obtain the matching results. The symbolic formulas include network resource entity symbols, network resource entity attribute symbols, and operation relation symbols. Based on the matching results, the target symbolic formula is determined from multiple symbolic formulas.

[0059] Optionally, the target symbolic formula is identified to obtain the identification result, including: Based on the descriptive prompts of the target symbolic formula and the pre-built network resource knowledge base, the target symbolic formula is identified, and the identification result is obtained. The identification results include network resource entities, the attributes of network resource entities, and the relationships between different network resource entities.

[0060] Optionally, instantiated prompts are generated based on the business thought chain, including: Determine the execution prompt and execution example for the target symbolic formula; Based on execution prompts, execution examples, and business thought chains, instantiated prompts are generated. The instantiated prompts include the names and attributes of network resource entities, as well as the relationships between different network resource entities.

[0061] Optionally, the target network resource data is processed, including: Construct a data link, with network resource entities as nodes and the relationships between different network resource entities as edges; Based on the data link, target network resource data is extracted from the network resource database; Filter the target network resource data to determine the number of target network resource entities; Based on the attributes of the target network resource entities, the number of target network resource entities is classified and statistically analyzed to obtain statistical results; The statistical results are formatted to obtain the data analysis results.

[0062] Optionally, the network resource data processing device 400 further includes: The monitoring unit is used to monitor the data processing status of large models in real time. The update unit is used to update the network resource knowledge base or the computational strategy parameters of the large model when the data processing status of the large model is determined to be abnormal.

[0063] Optionally, the training process for large models includes: Obtain sample natural language queries related to network resource data analysis; Determine the labels for the data analysis results corresponding to the natural language queries in the sample; Input the sample natural language query into the pre-built initial large model to obtain the data analysis and prediction results corresponding to the sample natural language query output by the initial large model; Based on the data analysis and prediction results corresponding to the sample natural language queries, and the labels of the data analysis results corresponding to the sample natural language queries, the loss function value is calculated. Based on the loss function value, the parameters of the initial large model are iteratively optimized to obtain the large model.

[0064] It should be noted that the network resource data processing apparatus provided in this embodiment of the invention can implement all the method steps implemented in the above-described network resource data processing method embodiment and can achieve the same technical effect. Therefore, the parts and beneficial effects that are the same as those in the method embodiment will not be described in detail here.

[0065] Figure 5 This is a schematic diagram of the structure of an electronic device provided in an embodiment of the present invention, such as... Figure 5As shown, the electronic device may include a processor 510, a communication interface 520, a memory 530, and a communication bus 540, wherein the processor 510, the communication interface 520, and the memory 530 communicate with each other via the communication bus 540. The processor 510 can call logical instructions in the memory 530 to execute a network resource data processing method. This method includes: acquiring a natural language query related to network resource data analysis; inputting the natural language query into a pre-built large model, where the large model determines the target symbolization formula corresponding to the natural language query, identifies the target symbolization formula, obtains the identification result, generates a business thinking chain based on the identification result, generates instantiated prompt words based on the business thinking chain, processes the target network resource data based on the instantiated prompt words, and outputs the data analysis result corresponding to the natural language query; wherein the large model is trained based on sample natural language queries and the labels of the data analysis results corresponding to the sample natural language queries.

[0066] Furthermore, the logical instructions in the aforementioned memory 530 can be implemented as software functional units and, when sold or used as independent products, 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 a part of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods described in the various embodiments of the present invention. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.

[0067] In another aspect, the present invention also provides a non-transitory computer-readable storage medium storing a computer program thereon. When executed by a processor, the computer program implements the network resource data processing method provided by the above methods. The method includes: acquiring a natural language query related to network resource data analysis; inputting the natural language query into a pre-constructed large model, whereby the large model determines the target symbolic formula corresponding to the natural language query, identifies the target symbolic formula, obtains the identification result, generates a business thinking chain based on the identification result, generates instantiated prompt words based on the business thinking chain, processes the target network resource data based on the instantiated prompt words, and outputs the data analysis result corresponding to the natural language query; wherein the large model is trained based on sample natural language queries and the labels of the data analysis result corresponding to the sample natural language queries.

[0068] The device embodiments described above are merely illustrative. The units described as separate components may or may not be physically separate. The components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the modules can be selected to achieve the purpose of this embodiment according to actual needs. Those skilled in the art can understand and implement this without any creative effort.

[0069] Through the above description of the embodiments, those skilled in the art can clearly understand that each embodiment can be implemented by means of software plus necessary general-purpose hardware platforms, and of course, it can also be implemented by hardware. Based on this understanding, the above technical solutions, in essence or the part that contributes to the prior art, can be embodied in the form of a software product. This computer software product can be stored in a computer-readable storage medium, such as ROM / RAM, magnetic disk, optical disk, etc., and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute the methods described in the various embodiments or some parts of the embodiments.

[0070] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, and not to limit them; although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features; and these modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. A method for processing network resource data, characterized in that, include: Obtain natural language queries related to network resource data analysis; The natural language query is input into a pre-built large model, which determines the target symbolic formula corresponding to the natural language query, identifies the target symbolic formula, obtains the identification result, generates a business thinking chain based on the identification result, generates instantiated prompt words based on the business thinking chain, processes the target network resource data based on the instantiated prompt words, and outputs the data analysis result corresponding to the natural language query. The large model is trained based on sample natural language queries and the corresponding data analysis result labels of the sample natural language queries.

2. The network resource data processing method according to claim 1, characterized in that, Determining the target symbolization formula corresponding to the natural language query includes: The natural language query is matched with a set of preset symbolic formulas to obtain matching results. The symbolic formulas include network resource entity symbols, network resource entity attribute symbols, and operation relation symbols. Based on the matching results, the target symbolic formula is determined from the plurality of symbolic formulas.

3. The network resource data processing method according to claim 1, characterized in that, The process of identifying the target symbolic formula and obtaining the identification result includes: Based on the descriptive prompts of the target symbolic formula and the pre-built network resource knowledge base, the target symbolic formula is identified to obtain the identification result; The identification results include network resource entities, the attributes of network resource entities, and the relationships between different network resource entities.

4. The network resource data processing method according to claim 1, characterized in that, The instantiated prompt words generated based on the business thought chain include: Determine the execution prompt and execution example for the target symbolized formula; Based on the execution prompt words, the execution examples, and the business thought chain, instantiated prompt words are generated. The instantiated prompt words include the names and attributes of network resource entities, as well as the association relationships between different network resource entities.

5. The network resource data processing method according to claim 1, characterized in that, The processing of the target network resource data includes: Construct a data link, wherein the data link uses network resource entities as nodes and the association relationships between different network resource entities as edges; Based on the data link, target network resource data is extracted from the network resource database; The target network resource data is filtered to determine the number of target network resource entities; Based on the attributes of the target network resource entities, the number of the target network resource entities is classified and statistically analyzed to obtain statistical results; The statistical results are formatted to obtain the data analysis results.

6. The network resource data processing method according to claim 3, characterized in that, After inputting the natural language query into a pre-built large model, the process also includes: Real-time monitoring of the data processing status of the large model; If the data processing status of the large model is determined to be abnormal, the network resource knowledge base is updated, or the computational strategy parameters of the large model are updated.

7. The network resource data processing method according to claim 1, characterized in that, The training process of the large model includes: Obtain sample natural language queries related to network resource data analysis; Determine the data analysis result labels corresponding to the natural language queries of the samples; The sample natural language query is input into a pre-built initial large model to obtain the data analysis and prediction results corresponding to the sample natural language query output by the initial large model. Based on the data analysis and prediction results corresponding to the sample natural language query, and the data analysis result labels corresponding to the sample natural language query, a loss function value is calculated. Based on the loss function value, the parameters of the initial large model are iteratively optimized to obtain the large model.

8. A network resource data processing device, characterized in that, include: The acquisition unit is used to acquire natural language queries related to network resource data analysis. The processing unit is used to input the natural language query into a pre-built large model, determine the target symbolization formula corresponding to the natural language query by the large model, identify the target symbolization formula, obtain the identification result, generate a business thinking chain based on the identification result, generate instantiated prompt words based on the business thinking chain, process the target network resource data based on the instantiated prompt words, and output the data analysis result corresponding to the natural language query. The large model is trained based on sample natural language queries and the corresponding data analysis result labels of the sample natural language queries.

9. An electronic device comprising a memory, a processor, and a computer program stored in the memory and running on the processor, characterized in that, When the processor executes the computer program, it implements the network resource data processing method as described in any one of claims 1 to 7.

10. A non-transitory computer-readable storage medium having a computer program stored thereon, characterized in that, When the computer program is executed by the processor, it implements the network resource data processing method as described in any one of claims 1 to 7.

Images