Work order dispatching method and device, electronic equipment and storage medium

Abstract

This invention relates to the field of work order processing technology, providing a work order dispatch method, apparatus, electronic device, and storage medium. The work order dispatch method includes: acquiring user input data; extracting event information from the user input data; generating event description text based on the event information; determining the event classification result of the event description text; determining the processing node path based on the event classification result; determining processing basis data based on the event description text, event classification result, and processing node path; generating structured work order data based on the event description text, event classification result, processing node path, and processing basis data; and dispatching the structured work order data. This invention, by extracting event information to generate event description text, can accurately and completely reconstruct the details of the event scene, correlate multiple data sets, and further determine the processing basis data as the basis for dispatch and handling, greatly improving the intelligence and accuracy of work order dispatch.

Description

Technical Field

[0001] This invention relates to the field of work order processing technology, and in particular to a work order dispatch method, apparatus, electronic device, and storage medium. Background Technology

[0002] In modern urban governance, the work order workflow system is a key link in responding to user requests and improving service efficiency. The traditional work order workflow is usually as follows: users submit requests through an application or service hotline, staff receive the information and manually create a work order, then dispatchers assign the work order to the corresponding responsible parties for offline verification and processing based on experience and division of responsibilities, and finally the results are fed back to the user and archived in the system.

[0003] In the work order dispatch process, dispatching and handling rely heavily on human experience and lack sufficient automation. Although some systems can achieve keyword matching, in the complex dispatching decisions and specific handling stages, staff still need to rely on their personal experience to manually search and compare a large number of knowledge bases, policy and regulatory documents to find the basis for dispatching and handling methods. The whole process is not only inefficient but also prone to errors. Summary of the Invention

[0004] This invention provides a work order dispatch method, apparatus, electronic device, and storage medium to overcome the deficiencies in the prior art.

[0005] This invention provides a work order transfer method, comprising: Acquire user input data, extract event information from the user input data, and generate event description text based on the event information; Determine the event classification result of the event description text, and determine the processing node path based on the event classification result; Based on the event description text, the event classification result, and the processing node path, determine the processing basis data; Based on the event description text, the event classification result, the processing node path, and the processing basis data, structured work order data is generated, and the structured work order data is then dispatched.

[0006] According to a work order dispatch method provided by the present invention, the step of extracting event information from the user input data and generating event description text based on the event information includes: Determine the processing model corresponding to the user input data; wherein the processing model includes at least one of a visual large model, a language large model, a speech large model, and a multimodal large model; Based on the processing model, the event information is extracted from the user input data, and the event description text in a preset format is generated based on the event information; wherein, the event information includes time, location, problem type, and event phenomenon.

[0007] According to a work order dispatch method provided by the present invention, determining the event classification result of the event description text includes: Obtain an event management multi-level classification system; wherein, the event management multi-level classification system includes multiple primary classification items and multiple secondary classification items, each primary classification item is associated with at least one secondary classification item, and each secondary classification item has a corresponding processing time limit; In the event management multi-level classification system, target first-level classification items and target second-level classification items that match the event description text are determined; Business keywords are determined from the event description text, and the target primary category, the target secondary category, and the business keywords are used as the event classification result.

[0008] According to a work order transfer method provided by the present invention, the step of determining the processing node path based on the event classification result includes: If the target first-level category item in the event classification result is not another comprehensive category, obtain the processing node architecture tree; Based on the event description text and the event classification result, the processing node path is determined from the processing node architecture tree.

[0009] According to a work order transfer method provided by the present invention, the step of determining the processing node path based on the event classification result includes: If the target primary category in the event classification result is "other comprehensive", check whether the target processing node can be determined based on the event description text; If the target processing node can be determined based on the event description text, the path of the processing node is determined based on the target processing node. If the target processing node cannot be determined based on the event description text, the preset base path will be used as the processing node path.

[0010] According to a work order dispatch method provided by the present invention, the step of determining processing basis data based on the event description text, the event classification result, and the processing node path includes: Based on the event description text, the event classification results, and the processing node path, a structured query object is constructed; Based on the structured query object and the invoked knowledge retrieval engine, a retrieval operation is performed in the pre-set multi-source processing basis knowledge base to obtain the target processing basis; Based on the event description text, the event classification result, the processing node path, and the target processing basis, the processing basis data is generated.

[0011] According to a work order dispatch method provided by the present invention, the step of generating structured work order data based on the event description text, the event classification result, the processing node path, and the processing basis data includes: The event classification results and the processing basis data are aligned. According to the predefined data model, the event description text, the processing node path, the data-aligned event classification result, and the processing basis data are encapsulated to generate the work order structured data that conforms to the interface specifications of the work order processing system.

[0012] The present invention also provides a work order transfer device, comprising: The extraction module is configured to acquire user input data, extract event information from the user input data, and generate event description text based on the event information; The first determining module is configured to determine the event classification result of the event description text and determine the processing node path based on the event classification result; The second determining module is configured to determine the processing basis data based on the event description text, the event classification result, and the processing node path. The dispatch module is configured to generate structured work order data based on the event description text, the event classification result, the processing node path, and the processing basis data, and then dispatch the structured work order data.

[0013] The present invention also provides an electronic device, including a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the computer program to implement the work order dispatch method as described above.

[0014] The present invention also provides a non-transitory computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the work order dispatch method as described above.

[0015] The present invention also provides a computer program product, including a computer program that, when executed by a processor, implements the work order dispatch method as described above.

[0016] The work order dispatch method, apparatus, electronic device, and storage medium provided by this invention generate event description text by extracting event information, which can accurately and completely reconstruct the details of the event scene, improving work order recognition accuracy and handling efficiency. Furthermore, it intelligently determines the most relevant processing basis data as the basis for dispatch and handling by associating the event description text, event classification results, and processing node paths. The entire process is driven by data and preset logic, no longer relying on the experience level and retrieval ability of specific staff. When dispatching a work order, it comes with clear processing basis data, enabling the responsible party to quickly and accurately understand the work order and execution standards, greatly improving the intelligence and accuracy of work order dispatch, and providing clear guidance for subsequent handling stages, thereby shortening the overall processing time. Attached Figure Description

[0017] To more clearly illustrate the technical solutions in this invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are some embodiments of this invention. For those skilled in the art, other drawings can be obtained from these drawings without creative effort.

[0018] Figure 1 This is one of the flowcharts of the work order transfer method provided by the present invention.

[0019] Figure 2 This is a schematic diagram illustrating the generation of event description text in an embodiment provided by the present invention.

[0020] Figure 3 This is an example diagram of user input data and event description text in an embodiment provided by the present invention.

[0021] Figure 4 This is the second flowchart of the work order transfer method provided by the present invention.

[0022] Figure 5 This is an example diagram of event description text and work order structured data in the embodiments provided by the present invention.

[0023] Figure 6 This is a schematic diagram of the work order dispatch device provided by the present invention.

[0024] Figure 7 This is a schematic diagram of the structure of the electronic device provided by the present invention. Detailed Implementation

[0025] To make the objectives, technical solutions, and advantages of this invention clearer, the technical solutions of this invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some, not all, of the embodiments of this invention. All other embodiments obtained by those skilled in the art based on the embodiments of this invention without creative effort are within the scope of protection of this invention.

[0026] 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. Furthermore, the collection, use and processing of the relevant data must comply with the relevant laws, regulations and standards of the relevant countries and regions, and corresponding operation entry points are provided for users to choose to authorize or refuse.

[0027] Figure 1 This is a flowchart illustrating a work order dispatch method according to an exemplary embodiment. For example... Figure 1 As shown in an exemplary embodiment, the work order dispatch method includes steps 110 to 140, which are described in detail below.

[0028] Step 110: Obtain user input data, extract event information from the user input data, and generate event description text based on the event information.

[0029] In this embodiment of the invention, user input data is received, including but not limited to data uploaded through applications, mini-programs, or web pages. User input data includes at least an event description, reporting time, and address; it may also include attachments uploaded by the user.

[0030] Event information is extracted from user input data, including time, location, problem type, and event description. Event description text is then generated based on the extracted event information.

[0031] Step 120: Determine the event classification result of the event description text, and determine the processing node path based on the event classification result.

[0032] In this embodiment of the invention, an intelligent classification operation is performed on the generated event description text to obtain the corresponding event classification result. Based on the event classification result, a path matching operation is performed to determine the processing node path for reassigning the work order. The processing node path records the responsible party for processing the work order and the responsible party for reassigning the work order.

[0033] Step 130: Based on the event description text, the event classification result, and the processing node path, determine the processing basis data.

[0034] In this embodiment of the invention, processing basis data is determined based on the event description text, event classification results, and processing node path. The processing basis data records data such as the violation identification and handling basis of the event when the responsible party handles the work order.

[0035] Step 140: Based on the event description text, the event classification result, the processing node path, and the processing basis data, generate work order structured data, and reassign the work order structured data.

[0036] In this embodiment of the invention, based on the event description text, event classification results, processing node path, and processing basis data, a unified and standardized work order structured data is generated, and the work order structured data is transferred to the responsible entity for processing the work order.

[0037] In this embodiment of the invention, by extracting event information to generate event description text, the details of the event scene can be accurately and completely restored, improving the accuracy of work order identification and the efficiency of handling. Furthermore, the event description text, event classification results, and processing node paths are correlated to intelligently determine the most relevant processing basis data as the basis for dispatching and handling. The entire process is driven by data and preset logic, no longer relying on the experience level and retrieval ability of specific staff. When dispatching a work order, it comes with clear processing basis data, enabling the responsible party to quickly and accurately understand the work order and execution standards, greatly improving the intelligence and accuracy of work order dispatch, and providing clear guidance for subsequent handling stages, thereby shortening the overall processing time.

[0038] In an exemplary embodiment of the present invention, the step of extracting event information from the user input data and generating event description text based on the event information includes: Determine the processing model corresponding to the user input data; wherein the processing model includes at least one of a visual large model, a language large model, a speech large model, and a multimodal large model; Based on the processing model, the event information is extracted from the user input data, and the event description text in a preset format is generated based on the event information; wherein, the event information includes time, location, problem type, and event phenomenon.

[0039] In embodiments of the present invention, such as Figure 2 As shown, after receiving user input data, it is determined whether there are attachments in the user input data, and different processing paths will be executed based on whether there are attachments.

[0040] Specifically, if the user does not upload attachments, the language model is designated as the processing model, and it processes the submitted user input data. This process follows these instructions: the role is set as a work order event information optimization assistant, with the goal of generating a concise, accurate, and stylistically appropriate problem description based on the event details, time, and address provided by the user. It is required to integrate all available information and ensure that the final output event description text contains no additional tags or annotations. It is particularly important to note that the language model must not fabricate details when the provided information is insufficient.

[0041] When a user uploads at least one attachment, the appropriate large model is invoked for joint analysis and processing based on the attachment type. Specifically: If the attachment type is an image, the visual big model is used as the processing model. The visual big model analyzes the uploaded image and associated text information, including the event description filled in by the user, as well as the automatically obtained reporting time, reporting address, etc.

[0042] If the attachment type is audio, the audio big data model is used as the processing model. The audio big data model converts the audio content into text and analyzes and processes it in conjunction with the event description and other information.

[0043] If the attachment type is video, the multimodal large model is used as the processing model. The multimodal large model extracts key visual and audio information from the video and processes it in conjunction with the event description.

[0044] The main model corresponding to each attachment is configured to execute the following processing logic: Using the role of a work order event information optimization assistant, based on the received event description, reporting address, and all reported attachments (images, audio, or video), it generates a structurally complete, semantically accurate, and linguistically standardized problem description text as the event description text. The processing logic includes steps such as information fusion, content parsing, language standardization, and output format control.

[0045] In the information fusion step, the event content in the text description, the acquired reporting time and geographical location information, are semantically aligned and integrated with the visual or audio information reflected in the reported attachments. Specifically, for image attachments, visual features are combined with textual information; for audio attachments, the speech-to-text description is fused with other data; and for video attachments, image and audio content are extracted from the video for multimodal information fusion. Through this semantic alignment and element integration, the final output is ensured to contain complete event information, including time, location, issue type, and event phenomenon.

[0046] In the content parsing step, for image attachments, key visual features are extracted and concise, objective image content descriptions are generated to supplement details that are not clear or require corroboration in the text description; for audio attachments, key audio information is identified to form concise and clear audio content descriptions; for video attachments, audiovisual information is analyzed, key content related to the event is extracted, and converted into structured text descriptions.

[0047] In the process of language standardization, the language should be organized in an objective, formal, and concise writing style, and subjective judgments, emotional wording, or speculative content should be prohibited.

[0048] In the output format control step, only a single paragraph of plain text description is output, without titles, numbers, explanatory statements, markers, or other redundant information.

[0049] In an exemplary embodiment of the present invention, the generated event description text is as follows: Figure 3 As shown below, the following specific examples illustrate this. Specifically, based on the attachment type and its content, the reference format for the generated event description text is as follows: When the user input data contains only text descriptions, the generated event description text format is as follows: On XX / XX / XXXX, XX problem was discovered in XX location.

[0050] When the user input data contains image attachments, the visual big data model is invoked to analyze the image content. A concise and objective description is generated based on each uploaded image, and combined with the text description to form a complete response. An example of the format of the generated event description text is: On XX / XX / XXXX, XX problem was discovered in XX location. Figure 1 Displays the status of XX. Figure 2 To reveal the phenomenon of XX.

[0051] When user input data includes voice attachments, the voice model is invoked to convert the voice content into text and analyze it. Key event points involved in the voice are automatically extracted and integrated into the event description. An example of the format of the generated event description text is: On XX / XX / XXXX, XX problem was discovered in XX location; Based on the voice content, the user described the XX phenomenon.

[0052] When the user input data contains video attachments, a multimodal large model is invoked to perform audiovisual information analysis on the video content. Based on the image and audio content in the video, key event information is extracted and a structured response is generated. An example of the format of the generated event description text is: On XX / XX / XXXX, XX problem was discovered in XX location; Video 1 shows XX scene, and Video 2 reveals XX dynamics.

[0053] When a user-reported event lacks a specific description, to avoid fabricating details, the generated event description text can be output as follows: On XX / XX / XXXX, a report was received in XX location, but a specific event description is missing.

[0054] In this embodiment of the invention, a multi-model is used to automatically perform semantic parsing on unstructured data such as images, voice, and video uploaded by citizens, generating concise and specific standardized event supplementary descriptions, which are then used as supplementary information to the work order text. After the standardized event supplementary descriptions are integrated with the original work order content, standardized event description text is generated, which can more accurately and completely restore the details of the event scene and improve the accuracy of work order recognition and processing efficiency.

[0055] Based on existing work order dispatch logic, this invention performs semantic parsing and structured processing on the original work orders reported by citizens, automatically identifies and extracts key event information such as time, location, problem type, and event phenomenon, and generates objective, formal, and concise standardized work order text according to preset templates. This effectively eliminates subjective evaluations, emotional language, and ambiguous expressions, while completely preserving the original event information, thereby improving the accuracy and efficiency of work order dispatch.

[0056] In an exemplary embodiment of the present invention, determining the event classification result of the event description text includes: Obtain an event management multi-level classification system; wherein, the event management multi-level classification system includes multiple primary classification items and multiple secondary classification items, each primary classification item is associated with at least one secondary classification item, and each secondary classification item has a corresponding processing time limit; In the event management multi-level classification system, target first-level classification items and target second-level classification items that match the event description text are determined; Business keywords are determined from the event description text, and the target primary category, the target secondary category, and the business keywords are used as the event classification result.

[0057] In embodiments of the present invention, such as Figure 4 As shown, the event description text serves as the input data for this embodiment. A pre-defined multi-level event management classification system is used. This system is a structured data model, comprising at least two levels: primary classification and secondary classification. Primary classifications include various primary category items, such as urban appearance and order, environmental sanitation, construction waste, municipal facilities, waste disposal, heating services, property services, water supply services, drainage services, and other comprehensive primary category items. Each primary category item is associated with one or more secondary category items, and each secondary category item is bound to a corresponding processing time limit, measured in days, which serves as the basis for subsequent work order assignment and processing time control.

[0058] Invoke the large language model and perform the following parallel processing based on the event description text: The primary classification determination involves identifying the first-level and second-level categories that best match the event description text based on the multi-level classification system. These identified categories are then designated as target first-level and target second-level categories. If multiple second-level categories match the event description text, the category with the shorter processing time or more urgent nature of the problem is prioritized as the target second-level category. Business keyword extraction involves automatically identifying all business keywords from the event description text and generating a set of business keywords that truly correspond to the event content. Business keywords originate from the names of secondary category items in the multi-level event management classification system or common terms within the domain. There is no limit to the number of business keywords, but content not explicitly stated in the event description text must not be fabricated. When the event description information in the event description text is vague, lacks key elements, or cannot match other secondary classification items in the event management multi-level classification system other than other comprehensive categories, its target primary classification item and target secondary classification item will be automatically classified into other comprehensive categories.

[0059] The primary target category, secondary target category, and business keywords are used as the event classification results. The primary target category, secondary target category, and business keyword set are serialized according to a predefined JSON (JavaScript Object Notation) schema, outputting a structured data object containing the following fields: ""event_class":{ "level1": "The target first-level category to be matched"; "level2": "The matched target secondary category item"; "processing_time": "Processing time limit for the corresponding target secondary classification item, in the format of "x days"; "tag": ["List of all relevant business keywords identified in the event"].

[0060] }".

[0061] The output structured data object does not contain any explanatory text, format escape characters, newlines, or non-JSON content, ensuring that it can be directly parsed and used by the downstream work order dispatch module.

[0062] In this embodiment of the invention, for clearly defined event description text, the matching event classification result is automatically output. For event description text with vague descriptions, insufficient information, or that cannot match the existing classification system, a unified classification to other comprehensive components is adopted, and business keywords in the event description text are extracted simultaneously. These business keywords can effectively assist subsequent event processing and analysis, improving work efficiency and accuracy. This invention proposes a comprehensive event processing mechanism for vague or incomplete information, enhancing the flexibility and intelligence of classification. Through automated business keyword extraction, manual intervention is reduced, improving the adaptability and processing capacity of the solution.

[0063] In an exemplary embodiment of the present invention, determining the processing node path based on the event classification result includes: If the target first-level category item in the event classification result is not another comprehensive category, obtain the processing node architecture tree; Based on the event description text and the event classification result, the processing node path is determined from the processing node architecture tree.

[0064] In this embodiment of the invention, it is determined whether the target primary category item in the event classification result belongs to other comprehensive categories. If the target primary category item in the event classification result is not other comprehensive categories, that is, it has clear primary and secondary category items, the processing node path is obtained by combining the event description text and the event classification result and performing multi-dimensional matching from the processing node architecture tree.

[0065] The processing node architecture tree includes multiple processing nodes, each representing the responsible entity for handling work orders. Within the tree, processing nodes with hierarchical relationships are connected. The processing node structure tree records multiple complete hierarchical paths.

[0066] A predefined category-node mapping rule base is established, which sets the priority mapping relationship between secondary category items and processing nodes. Based on the event description text, event classification results, and the category-node mapping rule base, the processing node path is determined from the processing node architecture tree. The processing node path represents the complete dispatch path of the work order.

[0067] When generating processing node paths, the following principles must be followed: Prioritize specific nodes; that is, if the event description text and event classification results are sufficient to uniquely identify the leaf processing node in the processing node architecture tree, then the output processing node path includes the identified leaf processing node. If a specific leaf processing node cannot be matched in the processing node architecture tree, then other processing nodes besides the leaf processing nodes are determined to form the processing node path.

[0068] Direct path connection means that when no processing node can be matched in the processing node architecture tree, if other responsible entities that do not belong to the processing node architecture tree can be identified based on the event description text and event classification results, the path corresponding to the responsible entity is directly output as the processing node path. Path validity verification means that the output processing node path must strictly exist in the preset organizational structure, be sequentially continuous, and cannot be skipped, abbreviated, or fabricated.

[0069] The final output processing node path is a single-line string, a complete processing node path separated by commas, without any explanations, punctuation, spaces or structured tags, for direct parsing and use by downstream work order processing systems.

[0070] In an exemplary embodiment of the present invention, determining the processing node path based on the event classification result includes: If the target primary category in the event classification result is "other comprehensive", check whether the target processing node can be determined based on the event description text; If the target processing node can be determined based on the event description text, the path of the processing node is determined based on the target processing node. If the target processing node cannot be determined based on the event description text, the preset base path will be used as the processing node path.

[0071] In this embodiment of the invention, if the target primary category is "Other Comprehensive", then the event processing branch for comprehensive categories is entered. When classified as "Other Comprehensive", the language big model is invoked, and the following operations are performed based on the event description text: If the event description in the event description text is clear, the location is well-defined, and the boundaries of responsibility are delineable, then the target processing node is determined based on the event description text; If the event description text contains unclear descriptions, ambiguous locations, overlapping responsibilities, or overlapping areas, making it impossible to clearly assign responsibility to a single entity, then the basic path corresponding to the pre-defined responsible entity will be used as the processing node path to generate structured response data. This data will strictly satisfy the following constraints: The event_class field retains the original event classification results of the input completely, without any parsing, rewriting or format conversion; The transfer_department field is fixed to the base path; The reassignment_basis field performs factual analysis based on the event description text, clearly indicating missing information, overlapping responsibilities, or the responsible party and basis for transfer, and prohibits the use of hypothetical or templated expressions; The remaining fields (illegal_act, legal_basis, violation_details, penalty_standard) are filled with default values ​​according to preset rules, indicating that there is currently no basis for determining illegality; The final output is a JSON object that conforms to standard syntax, without any extra text, escape characters, or formatting modifiers.

[0072] This invention also intelligently matches and assigns tasks to the corresponding responsible parties by combining event classification results and event description text. For other comprehensive work orders, if the corresponding responsible party can be clearly identified based on its standardized event text and topic tags, it is automatically transferred to that responsible party; otherwise, it is transferred to the last-resort responsible party, and an explanation of why accurate assignment is not possible is generated simultaneously, effectively avoiding duplicate event transfers. In an exemplary embodiment of this invention, the step of determining the processing basis data based on the event description text, the event classification results, and the processing node path includes: Based on the event description text, the event classification results, and the processing node path, a structured query object is constructed; Based on the structured query object and the invoked knowledge retrieval engine, a retrieval operation is performed in the pre-set multi-source processing basis knowledge base to obtain the target processing basis; Based on the event description text, the event classification result, the processing node path, and the target processing basis, the processing basis data is generated.

[0073] In this embodiment of the invention, based on the various types of data generated in the preceding process, a legal norms knowledge retrieval operation is performed to support the generation of subsequent basis for violation identification and handling. The legal norms knowledge retrieval operation includes two stages: query condition mapping and knowledge base retrieval.

[0074] During the query condition mapping phase, the event description text, event classification results, and processing node paths are used as input parameters. Through pre-defined data fusion logic, a unified structured query object is constructed. The structured query object contains the following fields: normal_text: The normalized event description text; cate_result: A complete structured representation of the event classification results; dep_path: The path string of the matched processing node.

[0075] This structured query object is used for semantic understanding and contextual association in the subsequent knowledge base retrieval stage, ensuring that the retrieval process fully combines the three dimensions of event semantics, business category, and responsible entity.

[0076] During the knowledge base retrieval phase, the knowledge retrieval engine is invoked to perform retrieval operations within a pre-configured multi-source processing knowledge base. This multi-source processing knowledge base is a dynamically configurable collection of heterogeneous documents, covering various normative bases required for event handling, including: national laws and administrative regulations, local regulations, industry technical standards, and management plans, statutory responsibility lists, and internal work procedures of relevant responsible entities in the target city.

[0077] Multi-source processing, based on a knowledge base, supports importing various common electronic document formats, including but not limited to: plain text files (such as .txt, .md), markup language files (such as .html, .htm, .json, .csv), portable document formats (.pdf), image files (such as .jpg, .jpeg, .png), office documents (such as .doc, .docx, .ppt, .pptx, .xlsx), and email files (.eml). Upon document import, it automatically performs content extraction and structured parsing, uniformly converting unstructured or semi-structured data into searchable text corpora and building indexes to support efficient queries.

[0078] The knowledge retrieval engine supports multiple retrieval modes, including: Vector semantic retrieval is based on a pre-trained language model to embed structured query objects into document content in a multi-source processing knowledge base, and then calculates the semantic similarity between the two to achieve context-aware matching. Keyword full-text search, which is to perform precise or fuzzy Boolean matching based on extracted business keywords; The hybrid retrieval strategy combines semantic similarity scores and keyword matching weights, and generates a comprehensive relevance score through a configurable weighted algorithm to improve the recall accuracy and ranking accuracy of high-value legal provisions.

[0079] In this embodiment of the invention, maintenance personnel can dynamically adjust retrieval parameters, including the maximum number of documents to be recalled, semantic similarity threshold, keyword matching strength coefficient, etc., so as to adapt to the differentiated needs of different event types for the breadth and depth of coverage of processing criteria.

[0080] Finally, the knowledge retrieval engine outputs a list of processing bases arranged in descending order of comprehensive relevance. Each result includes the original text of the article, the source (law name, article number), an explanation of applicability, and the source document identifier, which can be called by the subsequent violation determination and handling base generation module.

[0081] In one embodiment of the present invention, based on the pre-generated structured event information and the retrieved list of processing criteria, a language big model is invoked to generate standardized processing criteria and handling suggestions. Specifically, the event description text, event classification results, processing node paths, and target processing criteria are transmitted to the language big model in a structured form to ensure contextual integrity and traceability of decision-making criteria. The language big model executes the following parallel tasks according to preset inference rules to obtain structured work order data: The classification results are preserved in their original form, meaning that the original event classification results (event_class field) are completely retained in the output without any parsing, rewriting, or format conversion. The basis for assignment is generated by combining the basis for target processing, explaining the jurisdictional responsibilities of the target responsible entity (i.e., the entity ultimately responsible for executing the work order) for this type of event, and clarifying the logic of responsibility matching; if there is no clear basis for authorization, it is marked: According to the division of functions, this responsible entity is temporarily taking the lead in handling the matter, and further verification is recommended; The description of the violation should use standard legal terminology to objectively state the suspected violation in the event, and subjective evaluation or vague description is prohibited. Legal citation refers to extracting applicable laws, regulations, or rules from the knowledge base and sorting them according to their level of validity and priority of application. The format is: Article X, Paragraph X of the full name of the regulation: Specific content xxxx; Event level determination, that is, based on event-related information, such as duration, scope of impact, whether actual consequences have occurred, whether it has been tried multiple times, etc., to determine the corresponding event level and explain the reasons for the determination; The penalty standard output specifies the applicable penalty measures and their corresponding ranges based on the penalty discretion benchmark.

[0082] The model output is enforced to be a single valid JSON object containing the following fields: event_class: Original event classification result (strictly accurate); transfer_department: Directly references the input processing node path; reassignment_basis: Description of responsibilities; illegal_act: Description of the violation; legal_basis: A list of legal provisions; violation_details: includes the event level and the criteria for judgment; penalty_standard: The penalty measures and the scope of discretion.

[0083] Perform format validation on the output to ensure: no extra text, explanations, newlines, or escape characters; all field content originates from the input, and fictitious clauses or arbitrary additions are prohibited; in cases of overlapping responsibilities, clearly identify the main responsible party and briefly explain the collaboration mechanism.

[0084] In this embodiment of the invention, event description text, event classification results, and processing node paths are mapped to structured query objects for automatically retrieving laws, regulations, local administrative regulations, departmental rules, and industry standards from a locally deployed multi-source processing basis knowledge base. This multi-source processing basis knowledge base, based on vectorized representation and deep semantic understanding, supports semantic-level retrieval of unstructured documents such as text, PDFs (Portable Document Format), and web pages, accurately recalling semantically related clauses with different wording. It employs a hybrid retrieval mechanism that integrates full-text retrieval and vector retrieval, and improves result relevance through re-ranking. The multi-source processing basis knowledge base is only invoked on demand during the retrieval phase, does not participate in model training, and possesses access control and data isolation capabilities, meeting the security and compliance requirements of highly sensitive scenarios.

[0085] This invention proposes an automatic retrieval scheme based on a combination of structured query and deep semantic understanding, achieving high-precision semantic retrieval across document formats and significantly improving the practicality and intelligence of the knowledge base. The introduction of a hybrid retrieval mechanism and re-ranking technology enhances the accuracy and relevance of search results, resolving common problems of low precision and information overload in traditional retrieval schemes. An access control and data isolation mechanism tailored to highly sensitive scenarios is designed to ensure both efficient and secure use of the knowledge base, meeting the stringent compliance requirements of government and other fields.

[0086] In an exemplary embodiment of the present invention, generating work order structured data based on the event description text, the event classification result, the processing node path, and the processing basis data includes: The event classification results and the processing basis data are aligned. According to the predefined data model, the event description text, the processing node path, the data-aligned event classification result, and the processing basis data are encapsulated to generate the work order structured data that conforms to the interface specifications of the work order processing system.

[0087] In this embodiment of the invention, the event classification results and processing basis data are aligned and merged at the field level to ensure that the data source is traceable and the content is conflict-free. The aggregated data, along with the event description text and processing node path, are encapsulated according to a predefined unified data model to generate a structured output object that conforms to the interface specifications of the work order processing system, i.e., work order structured data. The data model includes the following core fields: event_class: A nested object containing level1 (first-level category), level2 (second-level category), processing_time (processing time limit, in days), and tag (a list of business keywords). transfer_department: The complete processing node path, represented by a comma-separated hierarchical string; reassignment_basis: Explanation of the reasons for the transfer based on statutory duties or division of responsibilities; illegal_act: An objective and standardized description of the violation; legal_basis: A list of legal, regulatory or rule provisions sorted by applicable priority; violation_details: A nested object containing severity (event level) and determination_basis (factual basis for determining the severity). penalty_standard: The enforceable penalty measures and their severity determined based on the discretionary standards.

[0088] like Figure 5 As shown, the final generated work order structured data is output in a valid JSON format, without any additional text, comments, newlines, or escape characters. It can be directly parsed and called by downstream work order scheduling systems, mobile processing terminals, or work order service platforms, realizing a fully automated closed loop from intelligent identification to accurate assignment and lawful handling of events.

[0089] In this embodiment of the invention, based on the event description text, event classification results, processing node paths, and processing basis data, a standardized JSON format output is automatically generated. For events with clear descriptions and clearly matched responsible parties, a complete structured result is output, including event classification, transfer path, transfer basis, description of violations, applicable legal provisions, event level, and corresponding penalty standards. For events with unclear descriptions, or those that cannot be clearly classified or transferred, the underlying responsible party and the reasons for the inability to accurately transfer the event are output. This achieves automated, standardized, and evidence-based event handling, significantly reducing the burden of manual assignment and improving the accuracy of assignments and law enforcement compliance.

[0090] This invention proposes a solution for automatically generating structured work order data based on multi-dimensional information. Legal basis and penalty standards are explicitly embedded in the standardized output, ensuring the transparency and legality of the processing. For events that are poorly described or cannot be clearly categorized, a fallback mechanism is designed to improve fault tolerance and flexibility, preventing case omissions or duplicate transfers. This significantly improves the accuracy of event dispatch, reduces the error rate of manual intervention and judgment, and enhances overall efficiency and compliance.

[0091] In this embodiment of the invention, it is able to intelligently identify and process different types of multimodal attachments (images, audio, video), and generate accurate event descriptions by calling different large models for analysis and fusion.

[0092] This invention provides a parameterized, configuration-driven architecture design that can quickly adapt to the management needs of different regions and has strong cross-regional migration capabilities. Regional configuration is possible: regionally specific elements such as organizational structure, event classification system, and local management regulations are decoupled into independent configuration files or rule bases. Different regions only need to replace the localized configuration files without modifying the core code or rebuilding the model, enabling efficient migration and deployment.

[0093] By combining a multi-level classification system with business theme keywords, it can intelligently identify event categories and accurately assign tasks based on detailed event information and organizational structure. Especially in multi-regional and multi-responsible entity scenarios, it determines whether to directly assign to a specific processing node or fall back to a higher-level processing node based on a pre-set classification-node mapping rule library, ensuring the accuracy and legality of the path.

[0094] By integrating event description text, event classification results, and processing node paths, a unified structured query object is constructed. This structured query object design ensures that the semantics, classification, and responsible party of the event are closely integrated, enabling more accurate matching in subsequent searches.

[0095] This invention supports a multi-source processing knowledge base, combining various retrieval modes (such as semantic retrieval, keyword retrieval, and hybrid strategies) to ensure efficient and comprehensive retrieval of the processing basis for events. The knowledge base is extensive, covering national regulations, local regulations, and industry standards, supports multiple document formats (PDF, txt, docx, etc.), and allows operations and maintenance personnel to dynamically adjust retrieval parameters to adapt to the needs of different types of events.

[0096] In other embodiments of the present invention, it can be implemented through a workflow engine-driven approach or a configuration file-based code implementation approach, which have different technical characteristics in terms of system architecture and operational flexibility.

[0097] In one embodiment, a workflow engine approach is employed. This approach encapsulates event information extraction and work order dispatch into independent workflow nodes. This node decoupling effectively avoids processing delays caused by multiple images in a single report, improving overall throughput efficiency. Furthermore, this workflow architecture supports dynamic replacement of large model interfaces at runtime and allows online updates to knowledge base files (such as event classification standards and departmental responsibility mapping tables). When organizational structure, event classification system, or input data structure changes, only the logic or connection relationships of the corresponding workflow nodes need to be adjusted; recompiling or deploying core code is unnecessary, significantly improving system maintainability and scalability.

[0098] In another embodiment, a configuration file-based code implementation is adopted. This approach uses predefined configuration files (such as YAML or JSON format) to declare the access address, calling parameters, and processing logic order of the large model used in each processing stage. The configuration file is loaded, and event information extraction and work order dispatch operations are performed according to the process defined therein. This implementation method has a compact structure and clear dependencies, and is suitable for scenarios with high requirements for deployment environment stability and relatively fixed business rules.

[0099] The two implementation methods described above can be selected or combined according to actual deployment needs, operation and maintenance capabilities, and evolution strategies, and both can effectively support the work order transfer method described in this invention.

[0100] The work order transfer device provided by the present invention will be described below. The work order transfer device described below can be referred to in correspondence with the work order transfer method described above. It should be noted that the device provided in the following embodiments and the method provided in the above embodiments belong to the same concept, and the specific way in which each module and unit performs operations has been described in detail in the method embodiments, and will not be repeated here.

[0101] In one exemplary embodiment of the present invention, please refer to Figure 6 , Figure 6 This is an exemplary embodiment of a work order dispatching device, which includes the following modules.

[0102] Extraction module 610 is configured to acquire user input data, extract event information from the user input data, and generate event description text based on the event information; The first determining module 620 is configured to determine the event classification result of the event description text and determine the processing node path based on the event classification result; The second determining module 630 is configured to determine processing basis data based on the event description text, the event classification result, and the processing node path. The dispatch module 640 is configured to generate work order structured data based on the event description text, the event classification result, the processing node path, and the processing basis data, and to dispatch the work order structured data.

[0103] In an exemplary embodiment of the present invention, the extraction module 610 includes: The first determining submodule is configured to determine the processing model corresponding to the user input data; wherein the processing model includes at least one of a visual large model, a language large model, a speech large model, and a multimodal large model; The extraction submodule is configured to extract the event information from the user input data based on the processing model, and generate the event description text in a preset format based on the event information; wherein, the event information includes time, location, problem type, and event phenomenon.

[0104] In an exemplary embodiment of the present invention, the first determining module 620 includes: The first acquisition submodule is configured to acquire an event management multi-level classification system; wherein, the event management multi-level classification system includes multiple primary classification items and multiple secondary classification items, each primary classification item is associated with at least one secondary classification item, and each secondary classification item has a corresponding processing time limit; The second determination submodule is configured to determine, in the event management multi-level classification system, the target first-level classification item and the target second-level classification item that match the event description text; The third determination submodule is configured to determine business keywords from the event description text and use the target primary category, the target secondary category, and the business keywords as the event classification result.

[0105] In an exemplary embodiment of the present invention, the first determining module 620 includes: The second acquisition submodule is configured to acquire the processing node architecture tree if the target first-level classification item in the event classification result is not another comprehensive. The fourth determination submodule is configured to determine the processing node path from the processing node architecture tree based on the event description text and the event classification result.

[0106] In an exemplary embodiment of the present invention, the first determining module 620 includes: The detection submodule is configured to detect whether the target processing node can be determined based on the event description text if the target primary classification item in the event classification result is "other comprehensive". The fifth determination submodule is configured to determine the processing node path based on the target processing node if the target processing node can be determined based on the event description text; As a submodule, it is configured to use a preset base path as the processing node path if the target processing node cannot be determined based on the event description text.

[0107] In an exemplary embodiment of the present invention, the second determining module 630 includes: The submodule is configured to construct a structured query object based on the event description text, the event classification result, and the processing node path. The retrieval submodule is configured to perform retrieval operations in a pre-set multi-source processing basis knowledge base based on the structured query object and the invoked knowledge retrieval engine to obtain the target processing basis; The generation submodule is configured to generate the processing basis data based on the event description text, the event classification result, the processing node path, and the target processing basis.

[0108] In an exemplary embodiment of the present invention, the transfer module 640 includes: The data alignment submodule is configured to perform data alignment between the event classification results and the processing basis data; The encapsulation submodule is configured to encapsulate the event description text, the processing node path, the data-aligned event classification result, and the processing basis data according to a predefined data model, thereby generating the work order structured data that conforms to the interface specifications of the work order processing system.

[0109] Figure 7 An example is a schematic diagram of the physical structure of an electronic device, such as... Figure 7 As shown, the electronic device may include: a processor 710, a communications interface 720, a memory 730, and a communication bus 740, wherein the processor 710, the communications interface 720, and the memory 730 communicate with each other through the communication bus 740. The processor 710 can call logical instructions in the memory 730 to execute a work order dispatch method, which includes: acquiring user input data, extracting event information from the user input data, and generating event description text based on the event information; Determine the event classification result of the event description text, and determine the processing node path based on the event classification result; Based on the event description text, the event classification result, and the processing node path, determine the processing basis data; Based on the event description text, the event classification result, the processing node path, and the processing basis data, structured work order data is generated, and the structured work order data is then dispatched.

[0110] Furthermore, the logical instructions in the aforementioned memory 730 can be implemented as software functional units and, when sold or used as independent products, can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present invention, essentially, or the part that contributes to the prior art, or a part of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods described in the various embodiments of the present invention. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.

[0111] On the other hand, the present invention also provides a computer program product, the computer program product including a computer program, the computer program being able to be stored on a non-transitory computer-readable storage medium, the computer program being executed by a processor, the computer being able to execute the work order dispatch method provided by the above methods, the method including: acquiring user input data, extracting event information from the user input data, and generating event description text based on the event information; Determine the event classification result of the event description text, and determine the processing node path based on the event classification result; Based on the event description text, the event classification result, and the processing node path, determine the processing basis data; Based on the event description text, the event classification result, the processing node path, and the processing basis data, structured work order data is generated, and the structured work order data is then dispatched.

[0112] In another aspect, the present invention also provides a non-transitory computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the work order dispatch method provided by the above methods, the method comprising: acquiring user input data, extracting event information from the user input data, and generating event description text based on the event information; Determine the event classification result of the event description text, and determine the processing node path based on the event classification result; Based on the event description text, the event classification result, and the processing node path, determine the processing basis data; Based on the event description text, the event classification result, the processing node path, and the processing basis data, structured work order data is generated, and the structured work order data is then dispatched.

[0113] The device embodiments described above are merely illustrative. The units described as separate components may or may not be physically separate. The components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the modules can be selected to achieve the purpose of this embodiment according to actual needs. Those skilled in the art can understand and implement this without any creative effort.

[0114] Through the above description of the embodiments, those skilled in the art can clearly understand that each embodiment can be implemented by means of software plus necessary general-purpose hardware platforms, and of course, it can also be implemented by hardware. Based on this understanding, the above technical solutions, in essence or the part that contributes to the prior art, can be embodied in the form of a software product. This computer software product can be stored in a computer-readable storage medium, such as ROM / RAM, magnetic disk, optical disk, etc., and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute the methods described in the various embodiments or some parts of the embodiments.

[0115] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, and not to limit them; although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features; and these modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. A work order transfer method, characterized in that, include: Acquire user input data, extract event information from the user input data, and generate event description text based on the event information; Determine the event classification result of the event description text, and determine the processing node path based on the event classification result; Based on the event description text, the event classification result, and the processing node path, determine the processing basis data; Based on the event description text, the event classification result, the processing node path, and the processing basis data, structured work order data is generated, and the structured work order data is then dispatched.

2. The work order transfer method according to claim 1, characterized in that, The step of extracting event information from the user input data and generating event description text based on the event information includes: Determine the processing model corresponding to the user input data; wherein the processing model includes at least one of a visual large model, a language large model, a speech large model, and a multimodal large model; Based on the processing model, the event information is extracted from the user input data, and the event description text in a preset format is generated based on the event information; wherein, the event information includes time, location, problem type, and event phenomenon.

3. The work order transfer method according to claim 1, characterized in that, The determination of the event classification result of the event description text includes: Obtain an event management multi-level classification system; wherein, the event management multi-level classification system includes multiple primary classification items and multiple secondary classification items, each primary classification item is associated with at least one secondary classification item, and each secondary classification item has a corresponding processing time limit; In the multi-level classification system for event management, target primary classification items and target secondary classification items that match the event description text are determined. Business keywords are determined from the event description text, and the target primary category, the target secondary category, and the business keywords are used as the event classification result.

4. The work order transfer method according to claim 3, characterized in that, The process of determining the processing node path based on the event classification result includes: If the target first-level category item in the event classification result is not another comprehensive category, obtain the processing node architecture tree; Based on the event description text and the event classification results, the processing node path is determined from the processing node architecture tree.

5. The work order transfer method according to claim 3, characterized in that, The process of determining the processing node path based on the event classification result includes: If the target primary classification item in the event classification result is "other comprehensive", check whether the target processing node can be determined based on the event description text; If the target processing node can be determined based on the event description text, the path of the processing node is determined based on the target processing node. If the target processing node cannot be determined based on the event description text, the preset base path will be used as the processing node path.

6. The work order transfer method according to claim 1, characterized in that, The process of determining the processing basis data based on the event description text, the event classification result, and the processing node path includes: Based on the event description text, the event classification results, and the processing node path, a structured query object is constructed; Based on the structured query object and the invoked knowledge retrieval engine, a retrieval operation is performed in the pre-set multi-source processing basis knowledge base to obtain the target processing basis; Based on the event description text, the event classification result, the processing node path, and the target processing basis, the processing basis data is generated.

7. The work order transfer method according to any one of claims 1 to 6, characterized in that, The process of generating structured work order data based on the event description text, the event classification result, the processing node path, and the processing basis data includes: The event classification results and the processing basis data are aligned. According to the predefined data model, the event description text, the processing node path, the data-aligned event classification result, and the processing basis data are encapsulated to generate the work order structured data that conforms to the interface specifications of the work order processing system.

8. A work order transfer and dispatch device, characterized in that, include: The extraction module is configured to acquire user input data, extract event information from the user input data, and generate event description text based on the event information; The first determining module is configured to determine the event classification result of the event description text and determine the processing node path based on the event classification result; The second determining module is configured to determine the processing basis data based on the event description text, the event classification result, and the processing node path. The dispatch module is configured to generate structured work order data based on the event description text, the event classification result, the processing node path, and the processing basis data, and then dispatch the structured work order data.

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

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

Images