Data processing method and device, computer equipment and storage medium
By extracting features and vectorizing user queries, and combining them with deep learning models to generate business responses, the system solves the problems of low efficiency and misjudgment in traditional intelligent parking management systems, achieving efficient and accurate business processing and system scalability.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- SHENZHEN MIRACLE WISDOM NETWORK CO LTD
- Filing Date
- 2025-12-13
- Publication Date
- 2026-04-17
AI Technical Summary
Traditional intelligent parking management systems are inefficient, costly, and unable to accurately understand user intent when handling user inquiries. Their rule engines cannot cope with diverse natural language expressions, leading to misjudgments and poor system scalability.
By extracting features and vectorizing user queries, generating business responses using a similarity matching and processing strategy library, and combining deep learning models for deep semantic understanding, efficient and accurate business processing is achieved.
It improves user experience and the level of automation in business processes, enhances the scalability and maintainability of the system, and is able to handle complex and diverse user queries.
Smart Images
Figure CN121880987A_ABST
Abstract
Description
Technical Field
[0001] This application relates to the field of artificial intelligence technology, and in particular to a data processing method, apparatus, computer equipment, storage medium, and computer program product. Background Technology
[0002] With the development of smart city technologies, intelligent parking management systems need to handle increasingly complex user inquiries. Traditional methods primarily fall into two categories: one relies on human customer service, which is costly and inefficient; the other uses rule engines based on keyword matching, a rigid approach that fails to accurately understand user intent. However, these traditional methods have significant drawbacks. Human customer service is difficult to manage at scale, resulting in slow response times and high costs. Rule engines, on the other hand, cannot handle diverse natural language expressions, and are prone to misinterpretations of queries with similar semantics but different business intents, leading to irrelevant answers. Furthermore, the system has poor scalability; each new business requirement necessitates rewriting rules, making maintenance difficult. Therefore, existing technologies fail to achieve a balance between efficiency, accuracy, and flexibility. Summary of the Invention
[0003] Therefore, it is necessary to provide a data processing method, apparatus, computer equipment, computer-readable storage medium, and computer program product to address the aforementioned technical problems.
[0004] Firstly, this application provides a data processing method. The method includes:
[0005] Obtain the target question; wherein, the target question includes user queries related to parking management;
[0006] Feature extraction is performed on the target problem to obtain the feature vector of the target problem;
[0007] The feature vector is matched with a plurality of preset candidate feature identifiers for similarity to determine the target feature identifier; wherein, the candidate feature identifiers are used to match the parking service category of the target task;
[0008] Based on the target feature identifier, a processing strategy corresponding to the target feature identifier is matched from a preset set of processing strategies;
[0009] The processing strategy and the target problem are input into a preset processing model, and the processing result is output; wherein, the processing model is used to generate a corresponding business response based on the processing strategy and the target problem.
[0010] In one embodiment, matching the processing strategy corresponding to the target feature identifier from a preset set of processing strategies based on the target feature identifier includes:
[0011] From the preset set of processing strategies, match the business rule text corresponding to the target feature identifier;
[0012] The background knowledge text corresponding to the target feature identifier is obtained from the target database;
[0013] The business rule text and background technology text are determined as the processing strategy.
[0014] In one embodiment, the target problem includes text data and image data; obtaining the target problem includes:
[0015] The image data is subjected to optical character recognition processing to extract text information;
[0016] The target question is obtained by merging the text information with the text data.
[0017] In one embodiment, after the output has been processed, it further includes:
[0018] The processing results are subjected to structured parsing, and it is verified whether they conform to the preset business data specifications.
[0019] When the parsing verification fails, training sample data containing the question input and the model output is generated;
[0020] The training sample data is sent to the sample management service; wherein, the training sample data is used to optimize the training of the processing model.
[0021] In one embodiment, the method further includes:
[0022] Based on the preset time period, the number of times each parking service category is executed and the execution results are counted.
[0023] Based on the number of executions and execution results of each parking service category, the risk coefficient for each service category is determined;
[0024] The set of processing strategies is adjusted based on the number of executions, execution results, and risk coefficients of each parking service category.
[0025] In one embodiment, the step of performing similarity matching between the feature vector and a plurality of preset candidate feature identifiers to determine the target feature identifier includes:
[0026] Determine the similarity between the feature vector and each candidate feature identifier;
[0027] If the similarity between the feature vector and each candidate feature identifier does not meet the preset requirements, an early warning message is generated.
[0028] Secondly, this application also provides a data processing apparatus. The apparatus includes:
[0029] The data acquisition module is used to acquire target questions; wherein, the target questions include user queries related to parking management;
[0030] The feature extraction module is used to extract features from the target problem to obtain the feature vector of the target problem;
[0031] The feature matching module is used to perform similarity matching between the feature vector and a plurality of preset candidate feature identifiers to determine the target feature identifier; wherein, the candidate feature identifiers are used to match the parking service category of the target task;
[0032] The strategy matching module is used to match the processing strategy corresponding to the target feature identifier from a preset set of processing strategies based on the target feature identifier.
[0033] The data processing module is used to input the processing strategy and the target problem into a preset processing model and output the processing result; wherein, the processing model is used to generate a corresponding business response based on the processing strategy and the target problem.
[0034] Thirdly, this application also provides a computer device. The computer device includes a memory and a processor, the memory storing a computer program, and the processor executing the computer program to implement the data processing method as described in any one of the embodiments of this disclosure.
[0035] Fourthly, this application also provides a computer-readable storage medium. The computer-readable storage medium stores a computer program thereon, which, when executed by a processor, implements the data processing method as described in any one of the embodiments of this disclosure.
[0036] Fifthly, this application also provides a computer program product. The computer program product includes a computer program that, when executed by a processor, implements the data processing method as described in any of the embodiments of this disclosure.
[0037] The aforementioned data processing methods, devices, computer equipment, storage media, and computer program products extract and vectorize features from user-submitted parking management-related questions, transforming unstructured natural language queries into machine-understandable feature vectors. The system accurately matches target business feature identifiers by calculating the similarity between the feature vectors and preset parking business category identifiers. This process, through feature engineering and similarity matching algorithms, significantly improves the accuracy and efficiency of question classification, avoiding the rigidity and high maintenance costs of traditional rule-based matching. Based on the matched feature identifiers, the system flexibly calls corresponding processing strategies from a predefined strategy library. This decoupling design between strategies and business identifiers greatly enhances the system's scalability and maintainability. When the business scope expands, only the strategy library or identifier library needs to be updated, without modifying the core architecture. The system inputs the question and strategy into a processing model to generate the final business response. This model can perform deep semantic understanding and content generation by combining specific contexts with preset strategies, thereby ensuring the accuracy and practicality of the response and effectively improving user experience and the level of automation in business processing. Overall, this solution achieves efficient, accurate, and scalable operation of parking management question answering through layered processing and intelligent matching. Attached Figure Description
[0038] Figure 1 This is a diagram illustrating the application environment of a data processing method in one embodiment.
[0039] Figure 2 This is a flowchart illustrating a data processing method in one embodiment;
[0040] Figure 3 This is a structural block diagram of a data processing device in one embodiment;
[0041] Figure 4 This is an internal structural diagram of a computer device in one embodiment. Detailed Implementation
[0042] To make the objectives, technical solutions, and advantages of this application clearer, the following detailed description is provided in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative and not intended to limit the scope of this application.
[0043] The data processing method provided in this application embodiment can be applied to, for example... Figure 1In the application environment shown, terminal 102 communicates with server 104 via a network. A data storage system can store the data that server 104 needs to process. The data storage system can be integrated onto server 104 or placed on a cloud or other network server. In this application environment, the user inputs a user query containing a target question through terminal 102; terminal 102 sends the query to server 104; server 104 executes the data processing method of this embodiment, such as feature extraction of the target question, similarity matching to determine target feature identifiers, matching processing strategies, and finally generating a business response through a processing model; subsequently, server 104 returns the business response to terminal 102 to complete the intelligent response to the user query. Terminal 102 can be, but is not limited to, various personal computers, laptops, smartphones, tablets, IoT devices, and portable wearable devices. IoT devices can be smart speakers, smart TVs, smart air conditioners, smart in-vehicle devices, etc. Portable wearable devices can be smartwatches, smart bracelets, head-mounted devices, etc. Server 104 can be implemented using a standalone server or a server cluster composed of multiple servers.
[0044] In one embodiment, such as Figure 2 As shown, a data processing method is provided, including the following steps:
[0045] Step S200: Obtain the target question; wherein, the target question includes user queries related to parking management.
[0046] The server can receive user queries from terminal devices and extract target questions related to parking management. This process may involve preliminary parsing of the original query to filter irrelevant information and focus on the core question content. The target question can take various forms, including explicit text descriptions as well as multimodal data containing images or audio. Preprocessing these inputs ensures that subsequent steps are based on clear and accurate questions. The key to this step is ensuring the comprehensiveness and accuracy of the question acquisition, thus laying the foundation for subsequent feature extraction and processing strategy matching.
[0047] In one exemplary embodiment, obtaining the target question may further include analyzing historical user interaction data to identify potential related information. For example, by mining past user query records or operational behaviors, contextual information that may be missing in the current question can be supplemented. Furthermore, the system can combine real-time environmental data, such as the current status of the parking lot, weather conditions, or traffic conditions, to further enrich the background information of the target question. This multi-dimensional information integration helps improve the accuracy of subsequent processing, ensuring that the generated business response better meets user needs. Simultaneously, an anomaly detection mechanism can be introduced during the acquisition of the target question to identify and process noise or erroneous information in the input data, thereby ensuring the reliability of subsequent processes.
[0048] Step S202: Extract features from the target problem to obtain the feature vector of the target problem.
[0049] The feature extraction process aims to transform the target problem into a numerical representation that can be processed and understood by a computer. The server can use natural language processing techniques such as word segmentation, part-of-speech tagging, and named entity recognition to extract key semantic information and map it into feature vectors in a high-dimensional space. This process may involve using pre-trained language models or deep learning algorithms to capture semantic relationships and contextual information within the target problem. For example, for target problems containing text data, word embedding techniques can be used to generate vector representations for each word, which can then be combined using weighted averaging or other methods to form the feature vector for the entire problem. For multimodal data, image feature extraction methods (such as convolutional neural networks) and text feature extraction methods need to be combined to generate a comprehensive feature vector that integrates multiple types of information. The core of this step is to ensure that the feature vector accurately reflects the essential attributes of the target problem, thus providing a reliable basis for subsequent similarity matching.
[0050] In one exemplary embodiment, feature extraction can also be optimized using domain knowledge. For example, in a parking management scenario, the system can predefine a set of keywords or semantic templates related to parking operations and assign higher weights to these keywords during feature extraction. Furthermore, an attention mechanism can be introduced to dynamically adjust the importance of different features, enabling the model to focus more on the key content of the target problem. This domain knowledge-based optimization strategy helps improve the discriminative power of feature vectors, thereby increasing the accuracy of subsequent processing. Simultaneously, the system can design multiple feature extraction schemes and automatically select the most suitable scheme based on the characteristics of the input data to achieve more flexible and efficient processing.
[0051] Step S204: Perform similarity matching between the feature vector and a plurality of preset candidate feature identifiers to determine the target feature identifier; wherein, the candidate feature identifiers are used to match the parking service category of the target task.
[0052] The server can calculate the similarity between the feature vector and each candidate feature identifier to find the business category that best matches the target problem. This process can employ various similarity calculation methods, such as cosine similarity, Euclidean distance, or deep learning-based metrics. The system will determine whether the matching results meet the requirements based on a preset threshold and select the optimal target feature identifier. The selection of the target feature identifier directly affects the matching effect of subsequent processing strategies, so it is necessary to ensure the accuracy and robustness of the matching process. In addition, the system can introduce a multi-level matching mechanism, first performing coarse-grained screening and then gradually refining the matching range, thereby improving matching efficiency and accuracy. This staged processing approach not only reduces computational overhead but also effectively avoids mismatches caused by data noise.
[0053] In one exemplary embodiment, the system can trigger an early warning mechanism when the similarity between the feature vector and all candidate feature identifiers fails to meet a preset requirement. The early warning information can be used to alert the user that their input may be ambiguous or abnormal, and can also serve as feedback data to optimize the candidate feature identifier library. For example, by analyzing these early warning messages, the system can dynamically update the candidate feature identifiers to adapt to constantly changing user needs and business scenarios. Furthermore, the system can combine manual intervention to review and correct early warning situations, thereby further improving the reliability of the matching. This intelligent early warning and update mechanism helps enhance the system's adaptability, enabling it to maintain a high matching success rate even when faced with diverse inputs.
[0054] Step S206: Based on the target feature identifier, match the processing strategy corresponding to the target feature identifier from the preset processing strategy set.
[0055] The processing strategy set can contain specific operational plans for different parking service categories. The server can retrieve this set to find the processing strategy corresponding to the target feature identifier. This process can be based on rule matching, machine learning models, or a combination of both. For example, for a specific parking service category, the system may have preset strategies such as priority ranking, resource allocation rules, or exception handling procedures. The matched processing strategy will directly determine how subsequent business responses are generated. The system can also index and optimize the processing strategy set or manage it hierarchically to ensure that appropriate strategies can be quickly located even in large-scale data scenarios. Furthermore, the system supports dynamic adjustment of the processing strategy set, continuously optimizing the strategy content based on actual operating results and user feedback, thereby better adapting to complex parking management needs. This flexible and efficient matching mechanism lays a solid foundation for generating accurate business responses.
[0056] In one exemplary embodiment, the system can also introduce strategy evaluation to verify and optimize the matched processing strategies in real time. The feasibility and effectiveness of the strategies can be evaluated through simulated execution or historical data backtesting, and strategy parameters can be dynamically adjusted or alternative solutions recommended based on the evaluation results. This mechanism can effectively cope with complex and ever-changing real-world scenarios, ensuring that the selected strategy is always in an optimal state. Furthermore, the system can record process data for each strategy matching and execution for subsequent analysis and model training, thereby continuously improving the intelligence level of strategy matching. Simultaneously, it can also have anomaly detection capabilities, identifying potential risks during strategy execution and triggering contingency plans in a timely manner to avoid service interruptions or user experience degradation due to strategy deviations.
[0057] Step S208: Input the processing strategy and the target problem into a preset processing model and output the processing result; wherein, the processing model is used to generate a corresponding business response based on the processing strategy and the target problem.
[0058] The processing model integrates the feature information of the strategy and the target question to generate a business response that meets the user's needs. This model can be built on a deep learning architecture, such as a sequence-to-sequence (Seq2Seq) model or a Transformer model, to capture and reason about complex semantic relationships. After inputting the processing strategy, the model combines the contextual information of the target question to progressively derive the optimal response. This process may involve multiple rounds of iterative optimization to ensure that the generated business response not only meets the core intent of the user's query but also takes into account the constraints of the actual scenario.
[0059] In one exemplary implementation, the processing model can also incorporate reinforcement learning mechanisms to continuously adjust its output strategy through interaction with the environment. For example, in a parking management scenario, the model can dynamically adjust recommended solutions, such as parking space allocation and navigation route planning, based on the real-time status of the parking lot, historical user behavior, and external environmental variables. This adaptive capability enables the system to maintain high efficiency even under high concurrency or resource constraints. Furthermore, the processing model can support multimodal input and output, such as fusing text, image, or voice information to provide users with a more intuitive and convenient service experience. The processing model can also integrate an anomaly detection module to identify potential problems in input data or intermediate results. For example, when a significant deviation is detected between the target problem and the matched processing strategy, the system triggers a fallback mechanism to re-evaluate the results of feature extraction or strategy matching steps. This multi-layered verification mechanism effectively reduces the probability of erroneous responses while improving the overall reliability of the system.
[0060] In the aforementioned data processing method, features are extracted and vectorized from parking management-related questions raised by users, transforming unstructured natural language queries into machine-understandable feature vectors. The system accurately matches target business feature identifiers by calculating the similarity between these feature vectors and preset parking business category identifiers. This process, through feature engineering and similarity matching algorithms, significantly improves the accuracy and efficiency of question classification, avoiding the rigidity and high maintenance costs of traditional rule-based matching. Based on the matched feature identifiers, the system flexibly calls corresponding processing strategies from a predefined strategy library. This decoupling design between strategies and business identifiers greatly enhances the system's scalability and maintainability. When the business scope expands, only the strategy library or identifier library needs to be updated, without modifying the core architecture. The system inputs the question and strategy into the processing model to generate the final business response. This model can perform deep semantic understanding and content generation by combining specific contexts with preset strategies, thereby ensuring the accuracy and practicality of the response and effectively improving user experience and the level of automation in business processing. Overall, this solution achieves efficient, accurate, and scalable operation of parking management question answering through layered processing and intelligent matching.
[0061] In one embodiment, matching the processing strategy corresponding to the target feature identifier from a preset set of processing strategies based on the target feature identifier includes:
[0062] From the preset set of processing strategies, match the business rule text corresponding to the target feature identifier.
[0063] The background knowledge text corresponding to the target feature identifier is obtained from the target database.
[0064] The business rule text and background technology text are determined as the processing strategy.
[0065] The system utilizes natural language processing (NLP) technology to parse and structure business rule text and background knowledge text. For example, it extracts key information using named entity recognition (NAME) and determines logical relationships through semantic analysis. This structured data can be integrated into a unified processing strategy representation, facilitating the use of subsequent processing models. Furthermore, the system can dynamically adjust the text content based on contextual information to ensure that the generated processing strategy better adapts to the current target problem.
[0066] In one exemplary embodiment, the system can introduce a multi-source information fusion mechanism. For example, it can incorporate real-time environmental data, user historical interaction records, and domain expert knowledge to enrich the content dimensions of the processing strategy. This comprehensive information integration approach not only improves the accuracy of strategy matching but also enhances the system's decision-making capabilities in complex scenarios. Simultaneously, the system supports continuous optimization of processing strategies by collecting user feedback and operational data to continuously update and improve the strategy library, adapting to ever-changing business needs and the technological environment.
[0067] In this embodiment, by combining a multi-source information processing strategy generation method, the system's adaptability and decision-making efficiency in complex scenarios can be significantly improved. Specifically, the system can use machine learning algorithms to weightedly fuse multi-source data, ensuring that various types of information are rationally allocated and effectively utilized during the strategy generation process.
[0068] In one embodiment, the target problem includes text data and image data; obtaining the target problem includes:
[0069] The image data is processed by optical character recognition to extract text information.
[0070] The target question is obtained by merging the text information with the text data.
[0071] The image data can include scene images related to parking management, such as parking lot signs, parking space numbers, or vehicle location information. By performing Optical Character Recognition (OCR) processing on this image data, the system can extract key text information, such as license plate numbers, parking space numbers, or directional text. This process not only expands the information dimension of the target problem but also provides richer contextual support for subsequent feature extraction and policy matching. When merging text information, the system performs deduplication and semantic integration on the extracted content to ensure that the final generated target problem is coherent and free of redundancy. In addition, the merged data can be subjected to grammatical and semantic verification to further improve its accuracy and usability.
[0072] In one exemplary implementation, image data processing can also incorporate computer vision techniques to identify non-textual information within the image. For example, object detection algorithms can be used to locate vehicles, vacant parking spaces, or obstacles in the image, and these visual features can be transformed into structured descriptions. These descriptions can be combined with text information extracted by OCR to form a multimodal representation of the target question. This multimodal fusion approach enables the system to more comprehensively understand the background information of the user's query, thereby providing more accurate support for subsequent steps. Simultaneously, the system can also be designed with a dynamic weight allocation mechanism to automatically adjust the importance ratio of image data and text data in the target question based on their quality and relevance, achieving more flexible data processing.
[0073] In this embodiment, by combining multimodal data processing methods, the system can significantly improve its ability to understand complex scenarios.
[0074] In one embodiment, after the output has been processed, it further includes:
[0075] The processing results are then subjected to structured parsing, and their conformity with preset business data specifications is verified.
[0076] When the parsing verification fails, training sample data containing the question input and the model output is generated.
[0077] The training sample data is sent to the sample management service; wherein, the sample data is used to optimize the training of the processing model.
[0078] The system can decompose the processing results into multiple independently evaluable fields or modules through structured parsing, thereby more accurately locating potential problems. For example, in a parking management scenario, the parsing process can extract key information such as recommended parking space numbers, navigation route information, or cost estimates, and verify whether each item conforms to business rules and data format requirements. If logical errors or non-compliance with specifications are found in certain fields, the system will automatically generate a detailed error report to assist in subsequent manual review or automated correction processes.
[0079] In one exemplary embodiment, when parsing verification fails, the generated training sample data can include not only the original input and model output, but also key node information from intermediate computation processes. This additional information helps analyze at which specific stage the model deviated, thus providing a clearer direction for optimization. Simultaneously, the system supports classifying and labeling different types of verification failure cases to specifically strengthen weak links in subsequent training. The sample management service can also incorporate active learning strategies to prioritize the labeling and use of high-value training samples. For example, for frequently occurring but not yet fully covered anomalous scenarios, the system dynamically increases their sampling weights to ensure the model can better grasp the relevant patterns in the next iteration. This closed-loop feedback mechanism not only accelerates model convergence but also effectively enhances the system's ability to handle complex boundary conditions.
[0080] In this embodiment, by introducing a closed-loop feedback mechanism, the system can continuously optimize the performance of the processing model. In actual operation, whenever parsing and verification fail, the system automatically triggers a sample generation process and transfers this sample data to the sample management service. The sample management service preprocesses and classifies this data to ensure it can be efficiently utilized for subsequent model training. Simultaneously, the system also supports periodic backtracking analysis of historical sample data, identifying key areas for improvement by comparing the performance of different model versions.
[0081] In one embodiment, the method further includes:
[0082] Based on the preset time period, the number of times each parking service category is executed and the execution results are counted.
[0083] Based on the number of executions and the execution results of each parking service category, the risk coefficient for each service category is determined.
[0084] The set of processing strategies is adjusted based on the number of executions, execution results, and risk coefficients of each parking service category.
[0085] The system can identify performance differences among various parking service categories by analyzing historical data. For example, some service categories may have higher failure rates due to complex rules or numerous external dependencies. Based on this statistical data, the system can calculate the risk coefficient for each service category and dynamically adjust the set of processing strategies accordingly. The adjustment process may include optimizing the strategy priority of high-risk categories, adding exception handling branches, or introducing additional verification steps. Furthermore, the system can combine machine learning algorithms to deeply analyze the statistical results, predict potential risk trends, and take preventative measures in advance. This data-driven strategy optimization approach not only improves the system's stability and reliability but also significantly reduces operating costs caused by business fluctuations. Simultaneously, by continuously monitoring the number of executions and results, the system can identify hidden efficiency bottlenecks and make targeted improvements to relevant processes, thereby continuously enhancing overall performance.
[0086] In one exemplary embodiment, the system can also introduce multi-dimensional evaluation indicators to comprehensively analyze the performance of parking service categories. For example, in addition to the number of executions and results, factors such as user satisfaction, response time, and resource consumption can be considered to form a more comprehensive risk assessment model. These indicators can be integrated into the risk coefficient calculation process through weighted calculation to ensure that the evaluation results more accurately reflect the actual operating conditions. Furthermore, the system can be designed with a dynamic weight adjustment mechanism to flexibly adjust the weight ratio of each indicator according to changes in business priorities or the external environment at different times. For example, during peak periods, more attention may be paid to response time, while during off-peak periods, more emphasis may be placed on resource utilization. This flexible evaluation approach helps the system make optimal strategy adjustments in different scenarios.
[0087] In one exemplary embodiment, the system can also combine real-time monitoring and early warning functions to focus on tracking high-risk business categories. When the risk coefficient of a certain category exceeds a preset threshold, the system automatically triggers an early warning mechanism and generates a detailed analysis report for management personnel to refer to. The report content may include information such as historical trends, abnormal distributions, and potential influencing factors, helping management personnel quickly locate the root cause of the problem and formulate countermeasures. Simultaneously, the system supports linking early warning information with processing strategies, using automated processes to temporarily adjust relevant strategies to alleviate short-term risk pressure. This real-time response capability not only enhances the system's resilience but also provides strong support for the smooth operation of the business. The system can also utilize big data analytics to uncover the correlations between parking business categories. For example, the failure of certain business categories may have implicit connections to the abnormal performance of other categories. By constructing a correlation network model, the system can identify these potential dependencies and consider them during strategy optimization. For example, when the risk coefficient of a certain category increases, the system can simultaneously adjust the strategy configuration of other categories with high correlation to prevent risk spread. This global perspective optimization approach helps improve the overall business coordination efficiency while reducing the possibility of systemic risks caused by localized problems.
[0088] In this embodiment, by introducing a multi-dimensional evaluation and real-time monitoring mechanism, the system can achieve comprehensive optimization of parking service categories. In practical applications, the system can not only identify potential problems through in-depth analysis of historical data, but also respond quickly by combining real-time dynamic information, ensuring the timeliness and effectiveness of strategy adjustments.
[0089] In one embodiment, the step of performing similarity matching between the feature vector and a plurality of preset candidate feature identifiers to determine the target feature identifier includes:
[0090] Determine the similarity between the feature vector and each candidate feature identifier.
[0091] If the similarity between the feature vector and each candidate feature identifier does not meet the preset requirements, an early warning message is generated.
[0092] The system can evaluate the matching results based on a preset similarity threshold. If the similarity between the feature vector and all candidate feature identifiers fails to meet the requirements, it indicates that the current input data may exceed the coverage of existing feature identifiers or that there are data quality issues. In this case, the generated warning information can include specific similarity values, a list of candidate feature identifiers, and possible cause analysis, providing a basis for subsequent manual intervention or automated processing.
[0093] In one exemplary implementation, the warning information can also trigger the system's self-adjustment mechanism. For example, when a large number of cases where similarity does not meet the requirements are detected, the system can automatically initiate a feature identifier expansion process, optimizing the candidate feature identifier set by introducing new training data or dynamically adjusting the feature extraction algorithm. Furthermore, the system can log these anomalies and generate improvement suggestions based on contextual information for administrator reference. This proactive response strategy effectively reduces the risk of business interruption due to feature matching failures. The system can be designed with a multi-level warning mechanism, taking differentiated response measures based on the degree of similarity deviation. For example, for minor deviations, the problem can be solved by prompting the user to provide more information or adjust the input format; while for severe deviations, a higher-level intervention process, such as manual review or model retraining, needs to be triggered. This tiered processing approach not only improves the system's flexibility but also minimizes the impact on user experience.
[0094] In this embodiment, by introducing a multi-level early warning mechanism, the system can more accurately respond to anomalies in feature matching. In actual operation, when similarity matching fails, the system will automatically select an appropriate response strategy based on the degree of deviation.
[0095] It should be understood that although the steps in the flowcharts of the embodiments described above are shown sequentially according to the arrows, these steps are not necessarily executed in the order indicated by the arrows. Unless explicitly stated herein, there is no strict order restriction on the execution of these steps, and they can be executed in other orders. Moreover, at least some steps in the flowcharts of the embodiments described above may include multiple steps or multiple stages. These steps or stages are not necessarily completed at the same time, but can be executed at different times. The execution order of these steps or stages is not necessarily sequential, but can be performed alternately or in turn with other steps or at least some of the steps or stages of other steps.
[0096] Based on the same inventive concept, this application also provides a data processing apparatus for implementing the data processing method described above. The solution provided by this apparatus is similar to the implementation scheme described in the above method; therefore, the specific limitations in one or more data processing apparatus embodiments provided below can be found in the limitations of the data processing method described above, and will not be repeated here.
[0097] In one embodiment, such as Figure 3 As shown, a data processing device 300 is provided, including: a data acquisition module 301, a feature extraction module 303, a feature matching module 305, a strategy matching module 307, and a data processing module 309, wherein:
[0098] The data acquisition module is used to acquire target questions; wherein, the target questions include user queries related to parking management;
[0099] The feature extraction module is used to extract features from the target problem to obtain the feature vector of the target problem;
[0100] The feature matching module is used to perform similarity matching between the feature vector and a plurality of preset candidate feature identifiers to determine the target feature identifier; wherein, the candidate feature identifiers are used to match the parking service category of the target task;
[0101] The strategy matching module is used to match the processing strategy corresponding to the target feature identifier from a preset set of processing strategies based on the target feature identifier.
[0102] The data processing module is used to input the processing strategy and the target problem into a preset processing model and output the processing result; wherein, the processing model is used to generate a corresponding business response based on the processing strategy and the target problem.
[0103] Each module in the aforementioned data processing device can be implemented entirely or partially through software, hardware, or a combination thereof. These modules can be embedded in or independent of the processor in a computer device, or stored in the memory of a computer device as software, so that the processor can call and execute the operations corresponding to each module.
[0104] In one embodiment, a computer device is provided, which may be a terminal, and its internal structure diagram may be as shown in Figure Y. The computer device includes a processor, memory, input / output interface, communication interface, display unit, and input device. The processor, memory, and input / output interface are connected via a system bus, and the communication interface, display unit, and input device are connected to the system bus via the input / output interface. The processor of the computer device provides computing and control capabilities. The memory of the computer device includes a non-volatile storage medium and internal memory. The non-volatile storage medium stores an operating system and computer programs. The internal memory provides an environment for the operation of the operating system and computer programs in the non-volatile storage medium. The input / output interface of the computer device is used for exchanging information between the processor and external devices. The communication interface of the computer device is used for wired or wireless communication with external terminals; wireless communication can be achieved through Wi-Fi, mobile cellular networks, NFC (Near Field Communication), or other technologies. When the computer program is executed by the processor, it implements a data processing method. The display unit of the computer device is used to form a visually visible image and may be a display screen, a projection device, or a virtual reality imaging device. The display screen can be an LCD screen or an e-ink screen. The input device of the computer device can be a touch layer covering the display screen, or buttons, trackballs, or touchpads set on the casing of the computer device, or external keyboards, touchpads, or mice, etc.
[0105] Those skilled in the art will understand that Figure 4 The structure shown is merely a block diagram of a portion of the structure related to the present application and does not constitute a limitation on the computer device to which the present application is applied. Specific computer devices may include more or fewer components than those shown in the figure, or combine certain components, or have different component arrangements.
[0106] It should be noted that the user information (including but not limited to user device information, user personal information, etc.) and data (including but not limited to data used for analysis, data stored, data displayed, etc.) involved in this application are all information and data authorized by the user or fully authorized by all parties, and the collection, use and processing of related data must comply with the relevant laws, regulations and standards of the relevant countries and regions.
[0107] Those skilled in the art will understand that all or part of the processes in the above embodiments can be implemented by a computer program instructing related hardware. The computer program can be stored in a non-volatile computer-readable storage medium. When executed, the computer program can include the processes of the embodiments described above. Any references to memory, databases, or other media used in the embodiments provided in this application can include at least one of non-volatile and volatile memory. Non-volatile memory can include read-only memory (ROM), magnetic tape, floppy disk, flash memory, optical memory, high-density embedded non-volatile memory, resistive random access memory (ReRAM), magnetic random access memory (MRAM), ferroelectric random access memory (FRAM), phase change memory (PCM), graphene memory, etc. Volatile memory can include random access memory (RAM) or external cache memory, etc. By way of illustration and not limitation, RAM can take many forms, such as Static Random Access Memory (SRAM) or Dynamic Random Access Memory (DRAM). The databases involved in the embodiments provided in this application may include at least one type of relational database and non-relational database. Non-relational databases may include, but are not limited to, blockchain-based distributed databases. The processors involved in the embodiments provided in this application may be general-purpose processors, central processing units, graphics processing units, digital signal processors, programmable logic devices, quantum computing-based data processing logic devices, etc., and are not limited to these.
[0108] The technical features of the above embodiments can be combined in any way. For the sake of brevity, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.
[0109] The embodiments described above are merely illustrative of several implementation methods of this application, and while the descriptions are specific and detailed, they should not be construed as limiting the scope of this patent application. It should be noted that those skilled in the art can make various modifications and improvements without departing from the concept of this application, and these all fall within the protection scope of this application. Therefore, the protection scope of this application should be determined by the appended claims.
Claims
1. A data processing method, characterized in that, The method includes: Obtain the target question; wherein, the target question includes user queries related to parking management; Feature extraction is performed on the target problem to obtain the feature vector of the target problem; The feature vector is matched with a plurality of preset candidate feature identifiers for similarity to determine the target feature identifier; wherein, the candidate feature identifiers are used to match the parking service category of the target task; Based on the target feature identifier, a processing strategy corresponding to the target feature identifier is matched from a preset set of processing strategies; The processing strategy and the target problem are input into a preset processing model, and the processing result is output; wherein, the processing model is used to generate a corresponding business response based on the processing strategy and the target problem.
2. The method according to claim 1, characterized in that, The step of matching the processing strategy corresponding to the target feature identifier from a preset set of processing strategies based on the target feature identifier includes: From the preset set of processing strategies, match the business rule text corresponding to the target feature identifier; The background knowledge text corresponding to the target feature identifier is obtained from the target database; The business rule text and background technology text are determined as the processing strategy.
3. The method according to claim 1, characterized in that, The target problem includes text data and image data; the acquisition of the target problem includes: The image data is subjected to optical character recognition processing to extract text information; The target question is obtained by merging the text information with the text data.
4. The method according to claim 1, characterized in that, After the output has been processed, it also includes: The processing results are subjected to structured parsing, and it is verified whether they conform to the preset business data specifications. When the parsing verification fails, training sample data containing the question input and the model output is generated; The training sample data is sent to the sample management service; wherein, the training sample data is used to optimize the training of the processing model.
5. The method according to claim 1, characterized in that, The method further includes: Based on the preset time period, the number of times each parking service category is executed and the execution results are counted. Based on the number of executions and execution results of each parking service category, the risk coefficient for each service category is determined; The set of processing strategies is adjusted based on the number of executions, execution results, and risk coefficients of each parking service category.
6. The method according to claim 1, characterized in that, The step of performing similarity matching between the feature vector and a plurality of preset candidate feature identifiers to determine the target feature identifier includes: Determine the similarity between the feature vector and each candidate feature identifier; If the similarity between the feature vector and each candidate feature identifier does not meet the preset requirements, an early warning message is generated.
7. A data processing apparatus, characterized in that, The device includes: The data acquisition module is used to acquire target questions; wherein, the target questions include user queries related to parking management; The feature extraction module is used to extract features from the target problem to obtain the feature vector of the target problem; The feature matching module is used to perform similarity matching between the feature vector and a plurality of preset candidate feature identifiers to determine the target feature identifier; wherein, the candidate feature identifiers are used to match the parking service category of the target task; The strategy matching module is used to match the processing strategy corresponding to the target feature identifier from a preset set of processing strategies based on the target feature identifier. The data processing module is used to input the processing strategy and the target problem into a preset processing model and output the processing result; wherein, the processing model is used to generate a corresponding business response based on the processing strategy and the target problem.
8. A computer device comprising a memory and a processor, wherein the memory stores a computer program, characterized in that, When the processor executes the computer program, it implements the steps of the method according to any one of claims 1 to 6.
9. A computer-readable storage medium having a computer program stored thereon, characterized in that, When the computer program is executed by a processor, it implements the steps of the method according to any one of claims 1 to 6.
10. A computer program product, comprising a computer program, characterized in that, When the computer program is executed by a processor, it implements the steps of the method according to any one of claims 1 to 6.