Intelligent question and answer result comparison and traceability method and system for electricity marketing

Abstract

The invention discloses an intelligent question and answer result comparison and traceability method and system for power marketing, and relates to the technical field of intelligent question and answer, and the method comprises the steps: S1, generating a plurality of intelligent responses according to the consultation content; s2, identifying key information associated with the power marketing business according to the intelligent response and performing information comparison; s3, querying a power marketing background business system in real time and performing conflict judgment according to the key information after comparison is completed, and obtaining target business data directly associated with the key information; s4, in the generated response, embedding a traceability link pointing to the target business data based on a traceability mechanism, and performing data traceability according to the traceability link; according to the method, specific business data and calculation logic can be traced, the credibility and transparency of the question and answer result are enhanced, and the reliability of the answer result is improved.

Description

Technical Field

[0001] This application relates to the field of intelligent question-answering technology, specifically to an intelligent question-answering result comparison and tracing method and system for electricity marketing. Background Technology

[0002] In the field of electricity marketing, with the deepening of digital transformation, customer service, electricity billing, and policy interpretation require increasingly higher accuracy and efficiency in information acquisition. Intelligent question-and-answer systems, especially those based on large language processing units (MLUs), are being widely adopted to provide a fast and natural interactive experience. However, electricity marketing data is highly specialized and sensitive, and the reliability of the question-and-answer results directly affects service quality and corporate credibility. Currently, mainstream intelligent question-and-answer solutions typically rely on a single large language processing unit to provide answers. This makes it difficult for users to intuitively compare the strengths and weaknesses of different processing units in terms of professional knowledge, logical reasoning, and content generation, thus hindering their ability to choose the most reliable answer.

[0003] Furthermore, existing systems generally lack a transparent display of the answer generation process, and the underlying data sources on which their conclusions are based are unclear, making it impossible for users to quickly trace back to specific business systems for data verification. This "black box" model casts doubt on the credibility of question-and-answer results, especially in high-value or high-risk business decisions, thus limiting the practical value and application depth of intelligent question-and-answer tools.

[0004] In summary, conventional intelligent question-answering systems rely on a single large language processing unit, making it difficult for users to compare their merits to choose a reliable answer. Furthermore, the answer generation process is opaque, the underlying data source is unclear, and it is impossible to trace and verify, resulting in a lack of credibility in the results. Summary of the Invention

[0005] The purpose of this application is to address the problem of low reliability of conventional intelligent question-and-answer systems in the power industry due to the lack of traceability of generated answers and the inflexibility of answer selection. It proposes an intelligent question-and-answer result comparison and traceability method and system for power marketing. By generating multiple responses and identifying key information in the responses, and performing conflict resolution after comparison, the accuracy of the data is improved. The traceability mechanism allows for tracing back to specific business data and calculation logic, enhancing the credibility and transparency of the question-and-answer results and improving the reliability of the response results.

[0006] To achieve the above objectives, the technical solutions adopted in the embodiments of this application are as follows: In a first aspect, embodiments of this application provide a method for comparing and tracing the results of intelligent question-and-answer sessions in electricity marketing, the method comprising: S1. Generate multiple intelligent responses based on the consultation content; S2. Based on the intelligent response, identify key information related to electricity marketing business and perform information comparison; S3. Based on the key information completed by the comparison, query the power marketing back-end business system in real time and make conflict resolution to obtain the target business data directly related to the key information. S4. In the generated response, a traceability link pointing to the target business data is embedded based on the traceability mechanism, and data traceability is performed based on the traceability link.

[0007] This solution generates multiple intelligent responses, breaking the reliance on a single language processing unit. Users can intuitively evaluate the merits of different responses and choose the most reliable answer. By identifying and comparing key information from electricity marketing operations, it reduces interference from invalid information, providing precise information targets for subsequent integration with backend systems and conflict resolution, thus improving data query efficiency. Conflict resolution addresses the issue of unclear data sources, ensuring that response data is based on real business systems and complies with current effective electricity policies. This avoids unreliable results caused by contradictions in multi-source data, improving the professionalism, accuracy, and reliability of responses. Data traceability enhances the credibility and transparency of question-and-answer results, breaking the "black box" model. Users can trace back to specific business data and reenact calculation steps, resolving doubts about the credibility of results and further improving data reliability.

[0008] Preferably, the step of querying the power marketing back-end business system in real time based on the key information completed by comparison to obtain target business data directly related to the key information includes: continuously extracting business data related to the key information from multiple power marketing back-end business systems; performing timestamp alignment and business logic calibration on the extracted business data to obtain first business data; performing semantic standardization processing on the first business data to obtain second business data with unified semantic standards; and adjudicating conflicting data in the second business data according to preset power business priority rules and data freshness assessment to obtain target business data.

[0009] Preferably, the step of resolving conflicts in the second business data and obtaining target business data based on preset power business priority rules and data freshness assessment includes: dynamically matching the currently effective power policy version according to the business type and timestamp in the second business data to obtain a matching power policy version; loading the corresponding business scenario adjudication rule set according to the matching power policy version to obtain an adapted business scenario adjudication rule set; and adjudicating conflicting data in the second business data according to the matching power policy version and the adapted business scenario adjudication rule set to obtain the target business data.

[0010] Preferably, the step of embedding a traceability link pointing to the target business data in the generated response based on the traceability mechanism, and performing data traceability based on the traceability link, includes: intercepting the click request of the traceability link and parsing the information to be verified in the request; Obtain target business data that matches the information to be verified; load the business rules and calculation logic that generate the information to be verified; replay the calculation process from the target business data to the information to be verified based on the loaded business rules and calculation logic; and respond to the user with a step-by-step explanation of the replayed calculation process to achieve traceability of intelligent question answering.

[0011] Preferably, the reenactment of the calculation process is presented to the user in a step-by-step manner, including: extracting the business type and time range of the information to be verified and determining the initial display granularity from a preset display strategy; aggregating the reenactment of the calculation process according to the initial display granularity, and using interactive operation, allowing the user to drill down or scroll up to view data and corresponding calculation details at different granularities as needed; when the user drills down, dynamically loading and displaying the target business data, business rules, and calculation steps corresponding to the fine-grained level, and highlighting or marking key business rules and data points in the current calculation path.

[0012] Preferably, the step of displaying key business rules and data points in the current calculation path through highlighting or specific markers includes: obtaining a power business rule influence factor configuration that matches the current user's role type, business scenario, and timestamp of the information to be verified; evaluating the influence weight of each business rule and data point in the current calculation path on the final result based on the power business rule influence factor configuration; adjusting the intensity and type of the highlighting or specific markers based on the influence weights; and displaying the adjusted highlighting or specific markers.

[0013] Preferably, the step of evaluating the influence weights of each business rule and data point in the current calculation path on the final result according to the power business rule influence factor configuration includes: parsing the current calculation path, identifying the business rules and data points contained in the calculation path, and constructing a business rule dependency graph, wherein the business rule dependency graph represents the logical association and calculation order between each business rule and data item; identifying multiple business rules and data points as inputs to other business rules according to the business rule dependency graph, wherein the multiple business rules include business rules with multiple inputs or multiple outputs, and quantifying the independent influence of the multiple business rules and data points as inputs to other business rules in the current calculation path; and using the independent influence in combination with the power business rule influence factor configuration as the influence weights of each business rule and the data point on the final result.

[0014] Preferably, the step of using the independent influence combined with the power business rule influence factor configuration as the influence weight of each business rule and the data point on the final result includes: establishing a power business rule influence factor configuration version library, wherein the configuration version library stores power business rule influence factor configurations effective for different time periods; retrieving power business rule influence factor configuration versions that match the business type and timestamp in the configuration version library according to the business type and timestamp to be verified, and determining the target configuration version according to a preset configuration version priority rule; and merging the quantified independent influence with the target configuration version as the influence weight of each business rule and the data point on the final result.

[0015] Preferably, determining the target configuration version according to a preset configuration version priority rule includes: when several power business rule impact factor configuration versions are retrieved in the configuration version library, obtaining the timestamps of conflicting configuration versions and identifying the effective and expiration dates of power business policies; constructing a policy lifecycle map based on the effective and expiration dates of power business policies and marking policy transition periods; determining whether the timestamp of the information to be verified falls within the policy transition period; when the timestamp of the information to be verified falls within the policy transition period, selecting a configuration version that matches the policy transition rule or a specific business scenario exception clause; identifying the business scenario tag in the information to be verified and adjusting the priority of the configuration version according to the business scenario tag; adding the authority level of the data source associated with the conflicting configuration version to the configuration version priority rule, so as to select the configuration version with the highest priority as the target configuration version based on the authority weight of the data source.

[0016] Secondly, embodiments of this application provide an intelligent question-and-answer result comparison and tracing system for electricity marketing, comprising: a consultation receiving module, used to generate multiple intelligent responses based on the consultation content; an information identification module, used to identify key information related to electricity marketing business based on the intelligent responses; a data query module, used to query the electricity marketing back-end business system in real time based on the key information and perform conflict resolution to obtain target business data directly related to the key information; and a data tracing module, used to embed a tracing link pointing to the target business data in the generated responses based on a tracing mechanism, and perform data tracing based on the tracing link.

[0017] The beneficial effects of this application are: 1. By querying the power marketing back-end business system in real time to obtain directly related target business data, it is possible to effectively integrate and standardize multi-source heterogeneous power marketing business data, so as to make conflict resolution through priority rules and data freshness assessment, ensuring the accuracy and timeliness of the obtained target business data, thereby improving the reliability of the question and answer results; 2. By dynamically matching the power policy version and the business scenario adjudication rule set, refined adjudication of conflicting data was achieved, ensuring the accuracy and compliance of the adjudication results and further enhancing the authority of the Q&A results; 3. Through the data traceability mechanism, the system can intercept traceability link requests and replay the calculation process to show users the generation logic and data source of the answer in a step-by-step explanation. This greatly enhances the transparency and credibility of the intelligent question-answering results and solves the problem of the "black box" mode. Users can drill down or roll up to view the data and calculation details as needed, and highlight key information to significantly improve the user's understanding efficiency and experience of the traceability process. 4. By constructing a business rule dependency graph and quantifying independent impacts, combined with the configuration of impact factors, we can achieve an accurate assessment of the impact weights of each business rule and data point, thereby more efficiently adapting to the dynamic changes in power business rules. Attached Figure Description

[0018] Other features, objects, and advantages of this application will become more apparent from the following detailed description of non-limiting embodiments with reference to the accompanying drawings. The drawings are for illustrative purposes only and are not intended to limit the scope of this application. Furthermore, the same reference numerals denote the same parts throughout the drawings.

[0019] Figure 1 A flowchart illustrating a method for comparing and tracing intelligent question-and-answer results in electricity marketing, as provided in this application embodiment.

[0020] Figure 2 This is a schematic diagram of a smart question-and-answer result comparison and tracing system module for electricity marketing, provided in an embodiment of this application. Detailed Implementation

[0021] To make the objectives, technical solutions, and advantages of this application clearer, the following detailed description of this application is provided in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely one preferred embodiment of this application and are only used to explain this application. They do not limit the scope of protection of this application. All other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this application.

[0022] Some of the terms or terms that appear in the description of the embodiments of this application shall be interpreted as follows: Intelligent response refers to a reply generated by an artificial intelligence system based on the content of a user's inquiry. This reply can be in the form of text, voice, or other multimedia content. In the context of electricity marketing, intelligent response typically involves professional knowledge such as electricity bill calculation, policy interpretation, and business processing procedures.

[0023] Example 1: As Figure 1 As shown, a method for comparing and tracing the results of intelligent question-and-answer sessions for electricity marketing includes the following steps: S1. Generate multiple intelligent responses based on the consultation content; S2. Based on the intelligent response, identify key information related to electricity marketing business and perform information comparison; S3. Based on the key information completed by the comparison, query the power marketing back-end business system in real time and make conflict resolution to obtain the target business data directly related to the key information. S4. In the generated response, a traceability link pointing to the target business data is embedded based on the traceability mechanism, and data traceability is performed based on the traceability link.

[0024] This embodiment aims to address the challenges of assessing answer reliability and lacking data traceability in existing intelligent question-and-answer systems in the power marketing field by employing mechanisms such as multi-intelligent response generation, key information comparison, backend business system query and conflict resolution, and embedded traceability links. This will enhance the transparency, credibility, and practical value of intelligent question-and-answer results, providing users with more accurate and verifiable power marketing consulting services.

[0025] Specifically, key information refers to data points extracted from intelligent responses that are closely related to electricity marketing operations and have decision-making or verification value, such as customer number, meter reading, business type, and policy terms. The electricity marketing back-office business system refers to various information systems supporting the daily operations of the power company, including but not limited to customer relationship management systems (CRM), billing systems, electricity consumption information collection systems, and marketing management systems. These systems store a large amount of business data and rules.

[0026] In some embodiments, step S1 includes: Multiple independent Large Language Processing Units (LLMs) or intelligent question-answering models can be used to process the same inquiry content in parallel. For example, a Transformer-based model can be configured to generate general answers, while a model fine-tuned with a power industry knowledge graph can be configured to generate specialized answers. Each model independently generates an intelligent response based on its training data and algorithm logic.

[0027] In other embodiments, step S1 includes: The system integrates different knowledge bases or data sources to generate intelligent responses. For example, one response might be primarily based on official policy documents related to electricity marketing, while another might be based primarily on a database of historical customer service cases. The system dynamically selects the appropriate knowledge base for retrieval and generation based on the type and complexity of the inquiry content.

[0028] As an optional implementation, step S2 includes: By utilizing natural language processing techniques (such as named entity recognition or relation extraction), entities related to electricity marketing business and the relationships between these entities are automatically identified from each smart response, thereby standardizing the representation of key information; Key information extracted from different intelligent responses is compared, including structured comparison and semantic comparison.

[0029] In some embodiments, the entities related to electricity marketing business include at least the customer name, electricity bill amount, policy name, and business processing date. The relationships between entities include the customer and the corresponding business type, for example, the queried electricity bill is for customer A's business type.

[0030] Furthermore, structured comparison includes comparing whether numerical values, dates, and encodings are consistent, while semantic comparison includes comparing whether the meanings of descriptive texts are consistent.

[0031] As an optional implementation, step S3 includes: Continuously extract business data related to the key information from multiple power marketing back-end business systems; The extracted business data is timestamped and the business logic is calibrated to obtain the first business data. Perform semantic standardization processing on the first business data to obtain second business data with a unified semantic standard; Based on preset power business priority rules and data freshness assessment, conflicting data in the second business data are adjudicated to obtain the target business data.

[0032] In some embodiments, continuously extracting business data related to the key information from multiple power marketing back-end business systems means that the system is configured to periodically or when triggered by specific events acquire business data directly related to the key information identified in user inquiries from multiple power marketing back-end business systems, such as customer relationship management systems (CRM), billing systems, electricity consumption information collection systems, and marketing activity management systems. This ensures that the acquired data is comprehensive and real-time, providing a foundation for subsequent data processing.

[0033] Furthermore, the extracted business data undergoes timestamp alignment and business logic calibration to obtain the first business data. Specifically, timestamp alignment refers to uniformly calibrating business data extracted from different systems based on their generation or update timestamps to eliminate time inconsistencies caused by system clock differences or data synchronization delays. Business logic calibration includes performing consistency checks and corrections on the extracted data according to the actual logic and rules of the electricity marketing business. For example, ensuring that the status flow of the same business in different systems conforms to the preset business process, or standardizing data fields in terms of format, units, etc., thereby obtaining first business data that is consistent in both time and business logic.

[0034] Furthermore, semantic standardization processing is performed on the first business data to obtain second business data with a unified semantic standard, including: The first business data, after being timestamped and calibrated according to business logic, is transformed and mapped into second business data with unified semantic standards, based on a predefined unified data model and terminology specification for the power marketing field.

[0035] In this embodiment, continuous extraction ensures data comprehensiveness, while timestamp alignment and business logic calibration resolve cross-system data consistency issues. Semantic standardization processes unify data into standardized field names and formats, eliminating semantic differences between heterogeneous data sources and facilitating subsequent conflict resolution and analysis. Conflict resolution is performed by combining power business priority rules and data freshness assessment, ensuring that when differences exist among multiple data sources, the most reliable and relevant target business data can be intelligently selected.

[0036] As an optional implementation, the step of resolving conflicts in the second business data and obtaining the target business data based on preset power business priority rules and data freshness assessment includes: Based on the business type and timestamp in the second business data, dynamically match the currently effective power policy version and obtain the matching power policy version; Based on the matching power policy version, load the corresponding business scenario adjudication rule set to obtain the adapted business scenario adjudication rule set; Based on the matching power policy version and the appropriate set of business scenario adjudication rules, conflicting data in the second business data is adjudicated to obtain the target business data.

[0037] It should be noted that the electricity business priority rules represent data priority strategies, set according to factors such as the importance, scope of impact, or type of electricity business. For example, billing data may have a higher priority than marketing campaign data. Data freshness assessment refers to considering the timeliness of data; in this embodiment, newer data is more valuable. When conflicting data regarding the same key information exists in the second business data, the system can intelligently query and match the current and local effective version of the electricity policy, ensuring that the policy basis for the ruling is the latest and most accurate. By comprehensively utilizing these priority rules and freshness assessment mechanisms, the system intelligently adjudicates the conflicting data, thereby obtaining highly accurate target business data that meets actual business needs. Target business data includes users' historical electricity consumption, applicable electricity price standards, and relevant policy documents.

[0038] Furthermore, different electricity policies may have different provisions regarding the handling of overdue payments, electricity price calculation, and subsidy disbursement. Therefore, once a matching electricity policy version is determined, the system will load a predefined set of adjudication rules based on the business logic and scenarios stipulated in that policy version. In other words, it will load adjudication rules corresponding to the current policy version and specific business scenarios, thereby providing accurate and detailed judgment criteria for conflict adjudication and avoiding the use of outdated or inapplicable rules.

[0039] In this embodiment, by introducing dynamic matching of electricity policy versions and loading of business scenario adjudication rule sets, the problems of insufficient accuracy and poor policy adaptability that may result from traditional fixed-rule adjudication in complex and ever-changing electricity marketing environments are solved. Because the adjudication process can dynamically adapt to constantly changing electricity policies and complex business scenarios, it avoids adjudication errors caused by untimely policy updates or rigid rules. This not only ensures that the data output by the intelligent question-and-answer system is highly consistent with actual business rules, but also enhances users' trust in the question-and-answer results.

[0040] As an optional implementation, the step of embedding a traceability link pointing to the target business data in the generated response based on a traceability mechanism, and performing data traceability based on the traceability link, includes: Intercept click requests for the source tracing link and parse the verification information in the request; Obtain target business data that matches the information to be verified; Load the business rules and calculation logic that generate the information to be verified; Based on the loaded business rules and calculation logic, the calculation process from the target business data to the information to be verified is replayed. The reenacted computation process is presented to the user in a step-by-step manner, enabling traceability of intelligent question answering.

[0041] Understandably, in practical applications, after obtaining the intelligent response result, users may only see the final business data or conclusion, without understanding how the result evolved from the original business data through a series of complex business rules and calculation logics. This lack of transparency may cause users to doubt the accuracy and credibility of the intelligent question-and-answer results, especially in key electricity marketing business scenarios involving monetary calculations and policy interpretations, where users often need to verify the result generation process. Therefore, this embodiment embeds a traceability link pointing to the target business data through a traceability mechanism, re-enacts the calculation process, and responds to users in a step-by-step explanation manner, thereby achieving transparent traceability of the intelligent question-and-answer results.

[0042] In some embodiments, when a user clicks a link for tracing in the smart response interface, the system immediately intercepts the click request. The request typically contains specific information that the user wants to verify, such as a specific value, business status, or conclusion given in the smart response. This information is the information to be verified. By parsing the request, the system can accurately identify the information that the user is interested in verifying, providing a clear target for subsequent tracing operations.

[0043] Specifically, in the electricity marketing back-end business system, the system retrieves the original target business data from the corresponding back-end system based on the type and context of the information to be verified, and loads the business rules and calculation logic upon which the aforementioned information to be verified is based. Here, business rules refer to various policies, calculation formulas, and business processes in the electricity marketing field, such as tiered pricing rules, peak-valley pricing rules, and preferential subsidy policies. Calculation logic refers to the specific algorithms and steps for applying these business rules to the target business data to derive the information to be verified. These rules and logic are typically stored in a configurable form in the rule engine or business logic library and can be dynamically loaded based on the business type and timestamp of the information to be verified, ensuring that the loaded rules and logic are completely consistent with the generation time and business scenario of the information to be verified.

[0044] As an optional implementation, the reenacted computation process is responded to by the user in a step-by-step interpretation, including: Extract the business type and time range of the information to be verified and determine the initial display granularity from the preset display strategy; The replayed calculation process is aggregated according to the initial display granularity, and an interactive operation is adopted, allowing users to drill down or roll up to view data and corresponding calculation details at different granularities as needed; When a user drills down to view, the system dynamically loads and displays the target business data, business rules, and calculation steps corresponding to the fine-grained levels, and highlights or marks key business rules and data points in the current calculation path.

[0045] As an optional implementation, displaying key business rules and data points in the current calculation path through highlighting or specific markers includes: Based on the user's current role type, business scenario, and timestamp of the information to be verified, obtain the power business rule influence factor configuration that matches the role type, business scenario, and timestamp. Based on the power business rule influence factor configuration, assess the influence weight of each business rule and data point in the current calculation path on the final result; The intensity and type of the highlighted display or specific mark are adjusted based on the influence weight, and the adjusted highlighted display or specific mark is displayed.

[0046] In this embodiment, after receiving a user's click request for the traceability link and parsing out the information to be verified, the system first analyzes the business category associated with the information and the time period involved. Based on the business category and time period, it determines an initial display granularity most suitable for the current scenario from a pre-set display strategy library, taking into account factors such as business complexity and user habits. For example, for monthly electricity bill inquiries, the initial granularity might be set as "monthly overview" or "itemized costs".

[0047] Furthermore, the raw data, intermediate calculation results, and applied business rules from the reenacted calculation process are integrated and summarized according to the initial display granularity determined above. Simultaneously, the system supports interactive operation, allowing users to understand their needs and, through clicking, dragging, or other methods, select "drill down" to view more detailed data and calculation steps, or "roll up" to return to a more macroscopic view.

[0048] Specifically, when a user drills down to view, the system dynamically loads and displays the target business data, business rules, and calculation steps corresponding to the fine-grained levels. It also highlights key business rules and data points that have a significant impact on the final result in the current calculation path by using highlighting or specific marking.

[0049] In this embodiment, by extracting the business type and time range of the information to be verified and determining the initial display granularity, the system can intelligently provide an appropriate starting view based on the context of the user's query, preventing the user from getting lost in a large amount of information. Furthermore, the replayed calculation process is aggregated, and users can drill down or scroll up to view data and calculation details at different granularities, allowing users to flexibly explore the calculation path from macro to micro or vice versa according to their needs, improving the efficiency and relevance of information acquisition. In addition, when the user drills down, fine-grained data is dynamically loaded and key business rules and data points are highlighted, intuitively guiding the user to focus on the factors that have the greatest impact on the results, thereby helping the user quickly locate the root cause of the problem or understand the core logic, further improving the accuracy of tracing and the user experience.

[0050] It should be noted that, compared to providing only a linear step-by-step explanation, this embodiment allows users to freely switch between data and calculation details of different granularities in an interactive manner according to their own needs. This greatly improves the efficiency of users in understanding complex business logic. By dynamically loading fine-grained data and highlighting key business rules and data points, users can quickly identify the core elements that affect the final result, thereby more effectively diagnosing problems, verifying the accuracy of calculations, or understanding the impact of policies. This can improve the transparency and user satisfaction of the intelligent Q&A service for electricity marketing.

[0051] As an optional implementation, the step of evaluating the influence weights of each business rule and data point in the current calculation path on the final result according to the power business rule influence factor configuration includes: The current computation path is parsed to identify the business rules and data points contained in the computation path, and a business rule dependency graph is constructed. The business rule dependency graph represents the logical relationship and computation order between each business rule and data item. Based on the business rule dependency graph, multiple business rules and data points that serve as inputs to other business rules are identified, wherein the multiple business rules include business rules with multiple inputs or multiple outputs. The independent impact of the multiple business rules and the data points that serve as inputs to other business rules on the current calculation path is quantified separately. The independent influence, combined with the power business rule influence factor configuration, serves as the influence weight of each business rule and the data point on the final result.

[0052] In some embodiments, parsing the current computation path refers to performing a structured analysis of the replayed computation process to identify all business rules and data points involved. This can be achieved through static code analysis, runtime log analysis, or predefined business process models. Multiple business rules refer to those with complex logic whose output may be affected by multiple inputs, or whose output may simultaneously serve as input for multiple subsequent business rules. Data points that serve as input for other business rules refer to data items that play a fundamental role in the computation path, whose changes in value directly or indirectly affect the execution results of multiple subsequent business rules. For example, a user's electricity consumption data may simultaneously affect electricity bill calculation rules, tiered electricity pricing rules, and subsidy rules.

[0053] In some embodiments, quantifying the independent impact of multiple business rules and data points that are inputs to other business rules in the current computation path includes: Sensitivity analysis can be used to quantify the impact by fine-tuning the values ​​of business rules or data points and observing the magnitude of changes in the final results; or, graph analysis can be used to calculate the centrality or propagation path length of business rules or data points in the dependency graph to reflect their influence.

[0054] Furthermore, the configuration of the impact factor for power business rules can be preset. For example, some core business rules (such as electricity price calculation) may be assigned a higher impact factor, while some auxiliary rules (such as notification sending) have a lower impact factor. By weighting or multiplying the quantified independent impact with the preset impact factor, a comprehensive impact weight can be obtained. This weight can then more comprehensively reflect the importance of business rules and data points in specific business scenarios.

[0055] As an optional implementation, the step of configuring the influence factors based on the independent influences and the power business rules as the influence weights of each business rule and the data points on the final result includes: Establish a configuration version library for power business rule impact factors, which stores power business rule impact factor configurations that take effect in different time periods; Based on the business type and timestamp in the information to be verified, retrieve the power business rule impact factor configuration version that matches the business type and timestamp from the configuration version library, and determine the target configuration version according to the preset configuration version priority rules; The quantified independent impacts are integrated with the target configuration version and used as the impact weights of each business rule and the data point on the final result.

[0056] It is understandable that in actual power marketing operations, the configuration of power business rule influence factors may be updated or iterated over time due to policy changes or the evolution of business scenarios. If the configuration version that matches the timestamp and business type of the information to be verified is not accurately identified and applied, the assessed influence weights may be inaccurate, thus affecting the accuracy and reliability of the tracing interpretation. Therefore, this embodiment uses configuration version management and priority adjudication mechanisms to ensure that the most appropriate power business rule influence factor configuration is always obtained and applied in a dynamically changing business environment, thereby improving the accuracy of the tracing results. Determining the target configuration version according to preset configuration version priority rules means that when multiple potentially matching configuration versions are retrieved, the system needs to adjudicate according to a set of pre-set priority rules to select the most accurate and authoritative single configuration version as the final target configuration version. The priority rules can comprehensively consider multiple factors such as the release time of the configuration version, policy authority, accuracy of the applicable scope, and reliability of the data source.

[0057] Specifically, the power business rule impact factor configuration version repository refers to a database or storage system that centrally stores and manages the power business rule impact factor configurations that take effect in different historical periods and business scenarios. It is used to record the effective date, expiration date, applicable business type, version number and other metadata of each configuration version to ensure that all configuration changes are traceable. The configuration version repository can be implemented in various forms such as relational database, NoSQL database or distributed file system.

[0058] Furthermore, when a user requests the source of a certain intelligent question-and-answer result, the specific business type of the information to be verified is first parsed from the request, such as electricity bill calculation, marketing activities, customer service, etc., as well as the timestamp of the occurrence. Based on this information as query conditions, a precise match is performed in the power business rule influence factor configuration version library to obtain all possible effective configuration versions within the specified business type and timestamp range.

[0059] Furthermore, the quantified independent impact is integrated with the target configuration version. After determining a unique target configuration version, the quantified independent impact values ​​of each business rule and data point are combined with the various impact factors defined in the target configuration version, thereby combining the independent impact with the policy orientation and business importance under specific time and specific business scenarios.

[0060] As an optional implementation, determining the target configuration version according to a preset configuration version priority rule includes: When several power business rule impact factor configuration versions are retrieved in the configuration version library, the timestamps of the conflicting configuration versions are obtained, and the effective date and expiration date of the power business policy are identified. Construct a policy lifecycle map based on the effective and expiration dates of electricity business policies, and mark the policy transition periods; Determine whether the timestamp of the information to be verified falls within the policy transition period. If the timestamp of the information to be verified falls within the policy transition period, select a configuration version that matches the policy transition rules or the exception clauses for specific business scenarios. Identify the business scenario tags in the information to be verified, and adjust the priority of the configuration version according to the business scenario tags; Add the authority level of the data source associated with the conflicting configuration version to the configuration version priority rule, so as to select the configuration version with the highest priority as the target configuration version based on the authority weight of the data source.

[0061] In this embodiment, constructing a policy lifecycle map based on the effective date and expiration date of the power business policy is a way to clearly show the complete timeline of each power business policy from its effective date to its expiration date through visualization or data structuring, and to clearly mark the policy transition period on this map; the policy transition period represents a specific time period during which new and old policies alternate and there may be overlapping or conflicting rules.

[0062] Specifically, when the timestamp of the information to be verified does fall within the policy transition period, priority is given to selecting configuration versions that match the preset policy transition rules or exceptions for specific business scenarios.

[0063] In this embodiment, by managing the policy lifecycle, dynamically adapting to business scenarios, and considering the authority of the data source, deviations in traceability results caused by improper configuration version selection are effectively avoided. This ensures that the question-and-answer system can more accurately determine the target configuration version, thereby providing users with more reliable intelligent question-and-answer traceability services. At the same time, it enhances users' trust in the traceability results and ensures the accuracy and compliance of power marketing business decisions.

[0064] Example 2, as Figure 2 As shown in the figure, this invention provides a method for comparing and tracing the source of intelligent question-and-answer results in electricity marketing. The system includes: The consultation receiving module is used to generate multiple intelligent responses based on the consultation content. The information recognition module is used to identify key information related to electricity marketing business based on the intelligent response; The data query module is used to query the power marketing back-end business system in real time based on the key information and to make conflict resolution, thereby obtaining target business data directly related to the key information. The data tracing module is used to embed a tracing link pointing to the target business data in the generated response based on the tracing mechanism, and to perform data tracing based on the tracing link.

[0065] In this embodiment, by generating multiple intelligent responses, the reliance on a single language processing unit is broken. Users can intuitively evaluate the merits of different responses and choose the most reliable answer. By identifying and comparing key information in electricity marketing business, interference from invalid information is reduced, providing precise information targets for subsequent integration with the back-end system and conflict resolution, thus improving data query efficiency. Conflict resolution addresses the issue of unclear data sources, ensuring that response data is based on real business systems and complies with current effective electricity policies. This avoids unreliable results caused by contradictions in multi-source data, improving the professionalism, accuracy, and reliability of responses. Data traceability enhances the credibility and transparency of question-and-answer results, breaking the "black box" model. Users can trace back to specific business data and reenact calculation steps, resolving the issue of questionable result credibility and further improving data reliability.

[0066] Preferably, the step of querying the power marketing back-end business system in real time based on the key information completed by comparison to obtain target business data directly related to the key information includes: continuously extracting business data related to the key information from multiple power marketing back-end business systems; performing timestamp alignment and business logic calibration on the extracted business data to obtain first business data; performing semantic standardization processing on the first business data to obtain second business data with unified semantic standards; and adjudicating conflicting data in the second business data according to preset power business priority rules and data freshness assessment to obtain target business data.

[0067] The above-described embodiments are preferred embodiments of this application and are not intended to limit the specific scope of this application. The scope of this application includes but is not limited to the specific embodiments described above. All equivalent changes made in accordance with the shape, structure, and method of this application are within the protection scope of this application.

Claims

1. A method for comparing and tracing the results of intelligent question-and-answer sessions in electricity marketing, characterized in that, Includes the following steps: S1. Generate multiple intelligent responses based on the consultation content; S2. Based on the intelligent response, identify key information related to electricity marketing business and perform information comparison; S3. Based on the key information completed by the comparison, query the power marketing back-end business system in real time and make conflict resolution to obtain the target business data directly related to the key information. S4. In the generated response, a traceability link pointing to the target business data is embedded based on the traceability mechanism, and data traceability is performed based on the traceability link.

2. The intelligent question-and-answer result comparison and tracing method for electricity marketing according to claim 1, characterized in that, The step of querying the power marketing back-end business system in real time based on the key information obtained from the comparison to obtain target business data directly related to the key information includes: Continuously extract business data related to the key information from multiple power marketing back-end business systems; The extracted business data is timestamped and the business logic is calibrated to obtain the first business data. Perform semantic standardization processing on the first business data to obtain second business data with a unified semantic standard; Based on preset power business priority rules and data freshness assessment, conflicting data in the second business data are adjudicated to obtain the target business data.

3. The intelligent question-and-answer result comparison and tracing method for electricity marketing according to claim 2, characterized in that, The step of resolving conflicts in the second business data based on preset power business priority rules and data freshness assessment, and obtaining target business data, includes: Based on the business type and timestamp in the second business data, dynamically match the currently effective power policy version and obtain the matching power policy version; Based on the matching power policy version, load the corresponding business scenario adjudication rule set to obtain the adapted business scenario adjudication rule set; Based on the matching power policy version and the appropriate set of business scenario adjudication rules, conflicting data in the second business data is adjudicated to obtain the target business data.

4. The intelligent question-and-answer result comparison and tracing method for electricity marketing according to claim 1, characterized in that, The generated response embeds a traceability link pointing to the target business data based on a traceability mechanism, and performs data traceability based on the traceability link, including: Intercept click requests for the source tracing link and parse the verification information in the request; Obtain target business data that matches the information to be verified; Load the business rules and calculation logic that generate the information to be verified; Based on the loaded business rules and calculation logic, the calculation process from the target business data to the information to be verified is replayed. The reenacted computation process is presented to the user in a step-by-step manner, enabling traceability of intelligent question answering.

5. The intelligent question-and-answer result comparison and tracing method for electricity marketing according to claim 4, characterized in that, The reenactment of the computation process is presented to the user in a step-by-step interpretive manner, including: Extract the business type and time range of the information to be verified and determine the initial display granularity from the preset display strategy; The replayed calculation process is aggregated according to the initial display granularity, and an interactive operation is adopted, allowing users to drill down or roll up to view data and corresponding calculation details at different granularities as needed; When a user drills down to view, the system dynamically loads and displays the target business data, business rules, and calculation steps corresponding to the fine-grained levels, and highlights or marks key business rules and data points in the current calculation path.

6. The intelligent question-and-answer result comparison and tracing method for electricity marketing according to claim 5, characterized in that, The method of displaying key business rules and data points in the current calculation path through highlighting or specific marking includes: Based on the user's current role type, business scenario, and timestamp of the information to be verified, obtain the power business rule influence factor configuration that matches the role type, business scenario, and timestamp. Based on the power business rule influence factor configuration, assess the influence weight of each business rule and data point in the current calculation path on the final result; The intensity and type of the highlighted display or specific mark are adjusted based on the influence weight, and the adjusted highlighted display or specific mark is displayed.

7. The intelligent question-and-answer result comparison and tracing method for electricity marketing according to claim 6, characterized in that, The step of evaluating the impact weights of each business rule and data point in the current calculation path on the final result based on the power business rule impact factor configuration includes: The current computation path is parsed to identify the business rules and data points contained in the computation path, and a business rule dependency graph is constructed. The business rule dependency graph represents the logical relationship and computation order between each business rule and data item. Based on the business rule dependency graph, multiple business rules and data points that serve as inputs to other business rules are identified, wherein the multiple business rules include business rules with multiple inputs or multiple outputs. The independent impact of the multiple business rules and the data points that serve as inputs to other business rules on the current calculation path is quantified separately. The independent influence, combined with the power business rule influence factor configuration, serves as the influence weight of each business rule and the data point on the final result.

8. The intelligent question-and-answer result comparison and tracing method for electricity marketing according to claim 7, characterized in that, The configuration of the influence factors based on the independent influence and the power business rules, as the influence weights of each business rule and the data point on the final result, includes: Establish a configuration version library for power business rule impact factors, which stores power business rule impact factor configurations that take effect in different time periods; Based on the business type and timestamp in the information to be verified, retrieve the power business rule impact factor configuration version that matches the business type and timestamp from the configuration version library, and determine the target configuration version according to the preset configuration version priority rules; The quantified independent impacts are integrated with the target configuration version and used as the impact weights of each business rule and the data point on the final result.

9. A method for comparing and tracing intelligent question-and-answer results in electricity marketing according to claim 8, characterized in that, The step of determining the target configuration version according to the preset configuration version priority rules includes: When several power business rule impact factor configuration versions are retrieved in the configuration version library, the timestamps of the conflicting configuration versions are obtained, and the effective date and expiration date of the power business policy are identified. Construct a policy lifecycle map based on the effective and expiration dates of electricity business policies, and mark the policy transition periods; Determine whether the timestamp of the information to be verified falls within the policy transition period. If the timestamp of the information to be verified falls within the policy transition period, select a configuration version that matches the policy transition rules or the exception clauses for specific business scenarios. Identify the business scenario tags in the information to be verified, and adjust the priority of the configuration version according to the business scenario tags; Add the authority level of the data source associated with the conflicting configuration version to the configuration version priority rule, so as to select the configuration version with the highest priority as the target configuration version based on the authority weight of the data source.

10. A smart question-and-answer result comparison and traceability system for electricity marketing, characterized in that, The system, applicable to the intelligent question-and-answer result comparison and tracing method for electricity marketing as described in any one of claims 1-9, comprises: The consultation receiving module is used to generate multiple intelligent responses based on the consultation content. The information recognition module is used to identify key information related to electricity marketing business based on the intelligent response; The data query module is used to query the power marketing back-end business system in real time based on the key information and to make conflict resolution, thereby obtaining target business data directly related to the key information. The data tracing module is used to embed a tracing link pointing to the target business data in the generated response based on the tracing mechanism, and to perform data tracing based on the tracing link.

Images