Intelligent work order submission method and system based on large model, and storage medium

Abstract

The invention provides an intelligent work order submission method and system based on a large model and a storage medium, and the method comprises the steps: obtaining work order event information, inputting the work order event information into a pre-trained intelligent work order submission model for semantic recognition, and obtaining event semantics; determining a work order template table according to the event semantics, and performing entity identification on the work order event information to obtain a work order entity; performing information filling in the work order template table according to the entity type of the work order entity to obtain a target work order table, and determining a work order approval flow according to the form identifier of the work order template table and the event semantics; and submitting the work order to the target work order table according to the work order approval flow. According to the embodiment of the invention, on the basis of the entity type of the work order entity, work order information filling can be automatically carried out on the work order template table according to the work order event information, a user does not need to manually fill the work order information, and the work order submission efficiency is improved.

Description

Technical Field

[0001] The present invention relates to the field of artificial intelligence technology, and in particular to a large-model-based intelligent work order submission method, system, and storage medium. Background Art

[0002] In a company's organization, the work order system is the core platform supporting daily transaction processing, service request response, and problem tracking. Whether it's IT operations and maintenance, customer service, or HR processes, work orders are essential for registering information, assigning tasks, tracking progress, and closing the process loop. Especially in multi-departmental collaboration and large-scale service scenarios, the work order process has become a crucial tool for standardized office work.

[0003] In the existing work order submission process, the work order content is generally filled in manually, which makes the user's work order filling operation cumbersome and reduces the efficiency of work order submission. Summary of the Invention

[0004] The purpose of the embodiments of the present invention is to provide an intelligent work order submission method, system and storage medium based on a large model to solve the problem of low efficiency of work order submission in the prior art.

[0005] The embodiment of the present invention is implemented as follows: a method for submitting intelligent work orders based on a large model, the method comprising:

[0006] Obtaining work order event information, and inputting the work order event information into a pre-trained intelligent work order submission model for semantic recognition to obtain event semantics;

[0007] Determine a work order template table according to the event semantics, and perform entity recognition on the work order event information to obtain a work order entity;

[0008] In the work order template table, information is filled in according to the entity type of the work order entity to obtain a target work order table, and the work order approval flow is determined according to the form identifier of the work order template table and the event semantics;

[0009] The work order is submitted to the target work order table according to the work order approval flow.

[0010] Preferably, before inputting the work order event information into the pre-trained intelligent work order submission model for semantic recognition, the method further includes:

[0011] Inputting the work order sample into the intelligent work order submission model for text encoding to obtain sample encoding features, and performing vector conversion on the sample encoding features to obtain a sample encoding vector;

[0012] Calculating the similarity between the sample coding vector and the preset coding vector to obtain a sample vector similarity, and determining the semantics corresponding to the maximum sample vector similarity as the sample semantics;

[0013] Matching the sample semantics with the historical work order database to obtain a sample work order table;

[0014] Segmenting the work order sample to obtain sample segmentations, and identifying entities on the sample segmentations to obtain sample entities;

[0015] Determining a sample urgency level according to the sample semantics, and determining a model loss according to the sample coding features, the sample work order table, and the sample entity;

[0016] The parameters of the intelligent work order submission model are updated according to the model loss until the intelligent work order submission model converges, thereby obtaining the pre-trained intelligent work order submission model.

[0017] Preferably, determining the work order approval flow according to the form identifier of the work order template table and the event semantics includes:

[0018] Determining a time vocabulary in the work order event information according to the event semantics, and determining an urgency level of the work order according to the time vocabulary;

[0019] Obtaining a user ID corresponding to the work order event information, and determining user permissions based on the user ID;

[0020] The form identifier of the work order template table, the user authority and the work order urgency level are combined to obtain level combination information, and the level combination information is matched with the approval flow query table to obtain the work order approval flow.

[0021] Preferably, information is filled in according to the entity type of the work order entity to obtain a target work order table, including:

[0022] Obtaining a fill-in identifier of a fill-in box in the work order template table, and matching the entity type of the work order entity with the fill-in identifier to obtain an entity fill-in position;

[0023] In the work order template table, information of the work order entity is filled in according to the entity filling position to obtain the target work order table.

[0024] Preferably, after filling in information of the work order entity according to the entity filling position to obtain the target work order table, the method further includes:

[0025] Determine a check box combination in the fill-in box according to the form identifier of the work order template table, and combine the information in the fill-in boxes in the check box combination to obtain check information;

[0026] Performing semantic recognition on the verification information to obtain verification semantics, and performing semantic contradiction detection on the verification semantics;

[0027] If the semantic contradiction detection fails, a semantic contradiction prompt is given to the check box combination.

[0028] Preferably, after filling in information of the work order entity according to the entity filling position to obtain the target work order table, the method further includes:

[0029] If no information is filled in the fill-in box, obtaining the box identifier of the fill-in box, and matching the box identifier with the necessary information table corresponding to the form identifier of the work order template table;

[0030] If the box identifier successfully matches the necessary information table, the fill-in box is determined to be a missing box, and the type of entity to be filled in is determined according to the box identifier;

[0031] Combining the entity type to be filled in with the event semantics to obtain semantic combination information;

[0032] Performing vector conversion on the semantic combination information to obtain a semantic combination vector, and performing similarity calculation between the combination vector and the historical work order vector to obtain the work order similarity;

[0033] Determine the historical work order table corresponding to the maximum work order similarity as the associated work order table, and determine the associated information box in the associated work order table according to the entity type to be filled in;

[0034] Fill the information in the associated information box into the corresponding missing box, and mark the missing box as missing.

[0035] Preferably, after calculating the similarity between the combined vector and the historical work order vector to obtain the work order similarity, the method further includes:

[0036] If the work order similarity is less than a similarity threshold, a missing prompt is generated according to the missing box, and a missing question is asked to the user corresponding to the work order event information according to the missing prompt;

[0037] If supplementary information regarding the missing question is received, the missing box is filled in according to the supplementary information.

[0038] Another object of an embodiment of the present invention is to provide an intelligent work order submission system based on a large model, the system comprising:

[0039] A semantic recognition module is used to obtain work order event information and input the work order event information into a pre-trained intelligent work order submission model for semantic recognition to obtain event semantics;

[0040] An entity recognition module is used to determine a work order template table according to the event semantics, and perform entity recognition on the work order event information to obtain a work order entity;

[0041] An information filling module is used to fill in information in the work order template table according to the entity type of the work order entity to obtain a target work order table, and determine the work order approval flow according to the form identifier of the work order template table and the event semantics;

[0042] The work order submission module is used to submit the work order to the target work order table according to the work order approval flow.

[0043] Preferably, the semantic recognition module is further used to:

[0044] Inputting the work order sample into the intelligent work order submission model for text encoding to obtain sample encoding features, and performing vector conversion on the sample encoding features to obtain a sample encoding vector;

[0045] Calculating the similarity between the sample coding vector and the preset coding vector to obtain a sample vector similarity, and determining the semantics corresponding to the maximum sample vector similarity as the sample semantics;

[0046] Matching the sample semantics with the historical work order database to obtain a sample work order table;

[0047] Segmenting the work order sample to obtain sample segmentations, and identifying entities on the sample segmentations to obtain sample entities;

[0048] Determining a sample urgency level according to the sample semantics, and determining a model loss according to the sample coding features, the sample work order table, and the sample entity;

[0049] The parameters of the intelligent work order submission model are updated according to the model loss until the intelligent work order submission model converges, thereby obtaining the pre-trained intelligent work order submission model.

[0050] The embodiment of the present invention can effectively identify the event semantics of the work order event information by inputting the work order event information into the pre-trained intelligent work order submission model for semantic recognition, and can automatically determine the work order template table based on the event semantics, without the need for the user to manually select the work order template table, thereby improving the efficiency of work order submission; by performing entity recognition on the work order event information, it can effectively extract the work order entity in the work order event information; based on the entity type of the work order entity, it can automatically fill in the work order information of the work order template table with the work order event information, without the need for the user to manually fill in the work order information, thereby further improving the efficiency of work order submission; through the form identification and event semantics of the work order template table, it can automatically determine the work order approval flow, and based on the determined work order approval flow, effectively guarantee the submission of the target work order table. BRIEF DESCRIPTION OF THE DRAWINGS

[0051] Figure 1 This is a flow chart of a method for submitting an intelligent work order based on a large model provided by the first embodiment of the present invention;

[0052] Figure 2 2 is a schematic diagram of the structure of an intelligent work order submission system based on a large model provided in a second embodiment of the present invention;

[0053] Figure 3 It is a structural diagram of a terminal device provided by the third embodiment of the present invention. DETAILED DESCRIPTION

[0054] In order to make the purpose, technical solutions and advantages of the present invention more clearly understood, the present invention will be further described in detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present invention and are not intended to limit the present invention.

[0055] In order to illustrate the technical solution of the present invention, specific embodiments are provided below.

[0056] Example 1

[0057] See also Figure 1 , is a flow chart of a large-model-based intelligent work order submission method provided by the first embodiment of the present invention. The large-model-based intelligent work order submission method can be applied to any device or system. The large-model-based intelligent work order submission method includes the following steps:

[0058] Step S10: Acquire work order event information, and input the work order event information into a pre-trained intelligent work order submission model for semantic recognition to obtain event semantics;

[0059] The intelligent work order submission model utilizes a combination of a Large Language Model (LLM) and Retrieval-Augmented Generation (RAG) technology. Incorporating the RAG technical framework, it builds a knowledge base covering the entire work order business domain, including work order scenarios (e.g., IT repair reporting, HR approval, etc.), historical work order data, departmental organizational structure, work order process rules, form field semantic definitions, and dependency logic. By structuring and vectorizing the content of salespeople in the work order business domain, the LLM model can dynamically retrieve relevant knowledge when processing work order tasks, forming a "long-term memory" and significantly improving contextual understanding and reasoning capabilities.

[0060] In this step, the work order event information can be obtained in a multimodal manner. For example, the work order event information can be in the format of text information, voice information or picture information. When the work order event information is text information input by the user, the work order event information is directly input into the pre-trained intelligent work order submission model for semantic recognition. When the work order event information is voice information input by the user, the work order event information is voice recognized to obtain the work order event text, and the work order event text is input into the pre-trained intelligent work order submission model for semantic recognition. When the work order event information is picture information, the text in the picture information is text located, and the work order event text is determined based on the text location result.

[0061] Optionally, before inputting the work order event information into the pre-trained intelligent work order submission model for semantic recognition, the method further includes:

[0062] Input the work order sample into the intelligent work order submission model for text encoding to obtain sample encoding features, and then perform vector conversion on the sample encoding features to obtain a sample encoding vector; wherein, historical work order data is collected and cleaned, classified and labeled to obtain work order samples, and business processes and organizational structures are sorted out to form a knowledge graph or index library;

[0063] Calculating the similarity between the sample coding vector and the preset coding vector to obtain a sample vector similarity, and determining the semantics corresponding to the maximum sample vector similarity as the sample semantics; wherein the preset coding vectors are stored in an index library, and the index library stores the correspondence between different preset coding vectors and corresponding semantics;

[0064] Matching the sample semantics with a historical work order database to obtain a sample work order table; wherein the historical work order database stores a correspondence between different sample semantics and corresponding sample work order tables;

[0065] Segmenting the work order sample to obtain sample segmentation, and identifying entities on the sample segmentation to obtain sample entities; wherein the sample segmentation is matched with an entity dictionary to obtain a sample entity, and the preset entities in the entity dictionary can be set according to requirements;

[0066] Determine the sample urgency level according to the sample semantics, and determine the model loss according to the sample coding features, the sample work order table and the sample entity; wherein the sample urgency level is used to indicate the urgency of the work order event corresponding to the work order sample, determine the standard coding features, the standard work order table and the standard entity based on the sample label of the work order sample, calculate the difference between the standard coding features and the sample coding features to obtain the feature difference, calculate the difference between the standard work order table and the sample work order table to obtain the work order table difference, calculate the difference between the standard entity and the sample entity to obtain the entity difference, determine the first loss based on the feature difference, determine the second loss based on the work order table difference, determine the third loss based on the entity difference, perform weighted operation on the first loss, the second loss and the third loss to obtain the model loss, and during the weighted operation, the weighting coefficients of the first loss, the second loss and the third loss can be set according to requirements;

[0067] The parameters of the intelligent work order submission model are updated according to the model loss until the intelligent work order submission model converges, thereby obtaining the pre-trained intelligent work order submission model; wherein, by establishing an evaluation mechanism, tracking indicators such as filling efficiency, accuracy, user satisfaction, etc., the model capability, knowledge base quality and interactive experience are continuously optimized based on feedback.

[0068] Step S20: determining a work order template table according to the event semantics, and performing entity recognition on the work order event information to obtain a work order entity;

[0069] Among them, the event semantics are matched with the historical work order database to obtain the work order template table, which eliminates the need for users to manually select the work order template table, thereby improving the efficiency of work order submission. In this step, the text corresponding to the work order event information is segmented to obtain the work order event segmentation, and the work order event segmentation is matched with the entity dictionary to obtain the work order entity.

[0070] In this embodiment, the pre-trained intelligent work order submission model automatically performs entity recognition, extracting key entities (such as device name, location, and time) from work order event information to obtain the work order entity. Optionally, it also supports reasoning about dependent fields. For example, if a user enters "laptop loses power and won't start," the "device model" can be completed as "laptop" and the "problem type" as "hardware failure" based on empirical rules. Furthermore, combining historical data with semantic understanding can determine whether there are logical conflicts or missing key fields, and perform system form verification, avoiding additional front-end reminders.

[0071] Step S30: Fill in information in the work order template table according to the entity type of the work order entity to obtain a target work order table, and determine the work order approval flow according to the form identifier of the work order template table and the event semantics;

[0072] Among them, by filling in information according to the entity type of the work order entity in the work order template table, the user does not need to manually fill in the work order information, which further improves the efficiency of work order submission. The form identification and event semantics of the work order template table can automatically determine the work order approval flow. Based on the determined work order approval flow, the submission of the target work order form is effectively guaranteed.

[0073] In this embodiment, the pre-trained intelligent work order submission model can automatically identify key features such as the work order type, urgency, and processing department corresponding to the work order event information. For example:

[0074] User input: "The projector in my conference room suddenly went black. I have a presentation tomorrow. Please arrange for technical assistance as soon as possible." will automatically output:

[0075] -Work Order Type: IT Equipment Failure

[0076] - Urgency: High

[0077] - Location: Meeting room (can be associated with user location)

[0078] - Initial responsible person: IT support team

[0079] Through semantic tag extraction and rule verification mechanism, the work that originally required filling out forms item by item can be completed automatically, improving submission efficiency.

[0080] Optionally, fill in information according to the entity type of the work order entity to obtain a target work order table, including:

[0081] Obtaining a fill identifier of a fill box in the work order template table, and matching the entity type of the work order entity with the fill identifier to obtain an entity fill position; wherein, obtaining a fill type corresponding to the fill identifier, performing a type match between the fill type and the entity type of the work order entity, and if the type match is successful, obtaining a position of the fill box in the work order template table to obtain an entity fill position;

[0082] In the work order template table, information of the work order entity is filled in according to the entity filling position to obtain the target work order table.

[0083] Furthermore, after filling in information for the work order entity according to the entity filling position to obtain the target work order table, the method further includes:

[0084] Determine a check box combination in the fill-in box according to the form identifier of the work order template table, and combine the information in the fill-in boxes in the check box combination to obtain verification information; wherein, the form identifier of the work order template table is matched with the verification query table to obtain the check box combination, the check box combination is used to represent different fill-in boxes in the work order template table that need to be semantically contradictory. The verification query table stores the corresponding relationship between different form identifiers and corresponding check box combinations;

[0085] Performing semantic recognition on the verification information to obtain verification semantics, and performing semantic contradiction detection on the verification semantics; wherein, semantic entities in the verification semantics are obtained, the semantic entities are combined in pairs to obtain entity combinations, and the entity combinations are matched with a semantic contradiction table, wherein the semantic contradiction table stores pre-set entities with semantic contradictions, and if the entity combination does not match the semantic contradiction table, it is determined that the semantic contradiction detection is qualified;

[0086] If the semantic contradiction detection fails, a semantic contradiction prompt is performed on the check box combination; wherein, if the entity combination matches the semantic contradiction table, the semantic contradiction detection is determined to be unqualified, and the fill-in boxes corresponding to the check box combination are marked with colors or logos to achieve a semantic contradiction prompt effect for the check box combination, so as to prompt the user that there is a semantic contradiction between the fill-in boxes marked with colors or logos.

[0087] Furthermore, after filling in information for the work order entity according to the entity filling position to obtain the target work order table, the method further includes:

[0088] If no information is filled in the fill-in box, the box identifier of the fill-in box is obtained, and the box identifier is matched with the necessary information table corresponding to the form identifier of the work order template table; wherein the necessary information table stores the fill-in boxes that must be filled in the work order template table;

[0089] If the box identifier successfully matches the necessary information table, the fill-in box is determined to be a missing box, and the type of the entity to be filled in is determined based on the box identifier; wherein the box identifier is matched with a type query table to obtain the type of the entity to be filled in, and the type query table stores the correspondence between different box identifiers and corresponding entity types to be filled in;

[0090] Combining the entity type to be filled in with the event semantics to obtain semantic combination information;

[0091] Perform vector conversion on the semantic combination information to obtain a semantic combination vector, and perform similarity calculation between the combination vector and the historical work order vector to obtain the work order similarity; wherein the historical work order vector is a required fill-in box in the historical work order, corresponding to the vector representation of the combination of the entity type to be filled in and the work order semantics;

[0092] Determining the historical work order table corresponding to the maximum work order similarity as the associated work order table, and determining the associated information box in the associated work order table according to the entity type to be filled in; wherein, matching the entity type to be filled in with the entity type in the fill-in box in the associated work order table, and determining the matched fill-in box in the associated work order table as the associated information box;

[0093] Fill the information in the associated information box into the corresponding missing box, and mark the missing box as missing.

[0094] In this embodiment, after calculating the similarity between the combined vector and the historical work order vector to obtain the work order similarity, the following steps are further included:

[0095] If the work order similarity is less than the similarity threshold, a missing prompt is generated according to the missing box, and a missing question is asked to the user corresponding to the work order event information according to the missing prompt; wherein the similarity threshold can be set according to needs. If the work order similarity is less than the similarity threshold, it is determined that there is no associated work order table corresponding to the work order template table in the historical work order data;

[0096] If supplementary information regarding the missing question is received, the missing box is filled in according to the supplementary information.

[0097] Preferably, determining the work order approval flow according to the form identifier of the work order template table and the event semantics includes:

[0098] Determining a time vocabulary in the work order event information based on the event semantics, and determining the work order urgency level based on the time vocabulary; wherein the time vocabulary is matched with a level query table to obtain the work order urgency level, and the level query table stores a correspondence between different time vocabulary and corresponding work order urgency levels;

[0099] Obtaining a user ID corresponding to the work order event information, and determining user permissions based on the user ID;

[0100] The form identifier of the work order template table, the user authority and the work order urgency level are combined to obtain level combination information, and the level combination information is matched with the approval flow query table to obtain the work order approval flow; wherein the approval flow query table stores the correspondence between different level combination information and the corresponding work order approval flows.

[0101] Step S40: submitting a work order to the target work order table according to the work order approval flow;

[0102] Among them, the target work order form can be automatically submitted through the application programming interface and trigger the subsequent work order approval process (such as approval process, notification reminder, etc.). At the same time, based on the intelligent work order submission model, a work order intelligent agent service can be provided to serve users to understand the work order processing status in real time.

[0103] In this embodiment, by inputting work order event information into a pre-trained intelligent work order submission model for semantic recognition, the event semantics of the work order event information can be effectively identified. Based on the event semantics, the work order template table can be automatically determined, eliminating the need for the user to manually select the work order template table, thereby improving the efficiency of work order submission. By performing entity recognition on the work order event information, the work order entity in the work order event information can be effectively extracted. Based on the entity type of the work order entity, the work order event information can be automatically filled into the work order template table, eliminating the need for the user to manually fill in the work order information, further improving the efficiency of work order submission. Based on the form identifier and event semantics of the work order template table, the work order approval flow can be automatically determined. Based on the determined work order approval flow, the submission of the target work order form is effectively guaranteed. In this embodiment, through natural language or voice description, structured parsing is performed based on the pre-trained intelligent work order submission model, achieving semantic analysis of work order submission. Based on intelligent verification and reasoning completion, work order submission is made simpler and more convenient, reducing the redundant submission logic of the traditional work order process.

[0104] Example 2

[0105] See also Figure 2 , is a schematic diagram of the structure of a large-model-based intelligent work order submission system 100 provided in a second embodiment of the present invention, including:

[0106] The semantic recognition module 10 is used to obtain work order event information and input the work order event information into a pre-trained intelligent work order submission model for semantic recognition to obtain event semantics.

[0107] Optionally, the semantic recognition module 10 is further configured to: input the work order sample into the intelligent work order submission model for text encoding to obtain sample encoding features, and perform vector conversion on the sample encoding features to obtain a sample encoding vector;

[0108] Calculating the similarity between the sample coding vector and the preset coding vector to obtain a sample vector similarity, and determining the semantics corresponding to the maximum sample vector similarity as the sample semantics;

[0109] Matching the sample semantics with the historical work order database to obtain a sample work order table;

[0110] Segmenting the work order sample to obtain sample segmentations, and identifying entities on the sample segmentations to obtain sample entities;

[0111] Determining a sample urgency level according to the sample semantics, and determining a model loss according to the sample coding features, the sample work order table, and the sample entity;

[0112] The parameters of the intelligent work order submission model are updated according to the model loss until the intelligent work order submission model converges, thereby obtaining the pre-trained intelligent work order submission model.

[0113] The entity recognition module 11 is used to determine the work order template table according to the event semantics, and perform entity recognition on the work order event information to obtain the work order entity.

[0114] The information filling module 12 is used to fill in information in the work order template table according to the entity type of the work order entity to obtain a target work order table, and determine the work order approval flow according to the form identifier of the work order template table and the event semantics.

[0115] Optionally, the information filling module 12 is further configured to: determine a time vocabulary in the work order event information according to the event semantics, and determine an urgency level of the work order according to the time vocabulary;

[0116] Obtaining a user ID corresponding to the work order event information, and determining user permissions based on the user ID;

[0117] The form identifier of the work order template table, the user authority and the work order urgency level are combined to obtain level combination information, and the level combination information is matched with the approval flow query table to obtain the work order approval flow.

[0118] Furthermore, the information filling module 12 is further configured to: obtain a filling identifier of a filling box in the work order template table, and match the entity type of the work order entity with the filling identifier to obtain an entity filling position;

[0119] In the work order template table, information of the work order entity is filled in according to the entity filling position to obtain the target work order table.

[0120] Furthermore, the information filling module 12 is further configured to: determine a check box combination in the filling box according to the form identifier of the work order template table, and combine the information in the filling boxes in the check box combination to obtain check information;

[0121] Performing semantic recognition on the verification information to obtain verification semantics, and performing semantic contradiction detection on the verification semantics;

[0122] If the semantic contradiction detection fails, a semantic contradiction prompt is given to the check box combination.

[0123] In this embodiment, the information filling module 12 is further configured to: if no information is filled in the filling box, obtain the box identifier of the filling box, and match the box identifier with the necessary information table corresponding to the form identifier of the work order template table;

[0124] If the box identifier successfully matches the necessary information table, the fill-in box is determined to be a missing box, and the type of entity to be filled in is determined according to the box identifier;

[0125] Combining the entity type to be filled in with the event semantics to obtain semantic combination information;

[0126] Performing vector conversion on the semantic combination information to obtain a semantic combination vector, and performing similarity calculation between the combination vector and the historical work order vector to obtain the work order similarity;

[0127] Determine the historical work order table corresponding to the maximum work order similarity as the associated work order table, and determine the associated information box in the associated work order table according to the entity type to be filled in;

[0128] Fill the information in the associated information box into the corresponding missing box, and mark the missing box as missing.

[0129] Preferably, the information filling module 12 is further configured to: if the work order similarity is less than a similarity threshold, generate a missing prompt according to the missing box, and ask a missing question to the user corresponding to the work order event information according to the missing prompt;

[0130] If supplementary information regarding the missing question is received, the missing box is filled in according to the supplementary information.

[0131] The work order submission module 13 is used to submit the work order to the target work order table according to the work order approval flow.

[0132] In this embodiment, by inputting the work order event information into the pre-trained intelligent work order submission model for semantic recognition, the event semantics of the work order event information can be effectively identified, and the work order template table can be automatically determined based on the event semantics, without the user having to manually select the work order template table, thereby improving the efficiency of work order submission. By performing entity recognition on the work order event information, the work order entity in the work order event information can be effectively extracted. Based on the entity type of the work order entity, the work order event information can be automatically filled in the work order template table with the work order information, without the user having to manually fill in the work order information, thereby further improving the efficiency of work order submission. Through the form identification and event semantics of the work order template table, the work order approval flow can be automatically determined. Based on the determined work order approval flow, the submission of the target work order table is effectively guaranteed.

[0133] Example 3

[0134] Figure 3 This is a block diagram of a terminal device 2 provided in the third embodiment of the present application. Figure 3 As shown, the terminal device 2 of this embodiment includes: a processor 20, a memory 21, and a computer program 22 stored in the memory 21 and executable on the processor 20, such as a program for the large-scale model-based intelligent work order submission method. When the processor 20 executes the computer program 22, the steps of each embodiment of the large-scale model-based intelligent work order submission method described above are implemented.

[0135] Exemplarily, the computer program 22 may be divided into one or more modules, which are stored in the memory 21 and executed by the processor 20 to implement the present application. The one or more modules may be a series of computer program instruction segments capable of performing specific functions, which are used to describe the execution process of the computer program 22 in the terminal device 2. The terminal device may include, but is not limited to, a processor 20 and a memory 21.

[0136] The processor 20 may be a central processing unit (CPU), other general-purpose processors, digital signal processors (DSP), application-specific integrated circuits (ASIC), field-programmable gate arrays (FPGA), other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. A general-purpose processor may be a microprocessor or any conventional processor.

[0137] The memory 21 may be an internal storage unit of the terminal device 2, such as a hard disk or memory of the terminal device 2. The memory 21 may also be an external storage device of the terminal device 2, such as a plug-in hard disk, a smart media card (SMC), a secure digital (SD) card, a flash card, etc. equipped on the terminal device 2. Furthermore, the memory 21 may include both an internal storage unit of the terminal device 2 and an external storage device. The memory 21 is used to store the computer program and other programs and data required by the terminal device. The memory 21 may also be used to temporarily store data that has been output or is about to be output.

[0138] In addition, the functional modules in the various embodiments of the present application may be integrated into a single processing unit, or each unit may exist physically separately, or two or more units may be integrated into a single unit. The aforementioned integrated units may be implemented in the form of hardware or software functional units.

[0139] If the integrated module is implemented in the form of a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Among them, the computer-readable storage medium can be non-volatile or volatile. Based on this understanding, the present application implements all or part of the process in the above-mentioned embodiment method, and can also be completed by instructing the relevant hardware through a computer program. The computer program can be stored in a computer-readable storage medium, and the computer program can implement the steps of the above-mentioned various method embodiments when executed by the processor. Among them, the computer program includes computer program code, and the computer program code can be in source code form, object code form, executable file or some intermediate form, etc. The computer-readable storage medium may include: any entity or device that can carry computer program code, recording medium, U disk, mobile hard disk, magnetic disk, optical disk, computer memory, read-only memory (ROM, Read-Only Memory), random access memory (RAM, Random Access Memory), electric carrier signal, telecommunication signal and software distribution medium, etc. It should be noted that the content contained in computer-readable storage media can be appropriately increased or decreased according to the requirements of legislation and patent practices in the jurisdiction. For example, in some jurisdictions, according to legislation and patent practices, computer-readable storage media do not include electrical carrier signals and telecommunications signals.

[0140] The above-described embodiments are only used to illustrate the technical solutions of the present application, rather than to limit them. Although the present application has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that they can still modify the technical solutions described in the aforementioned embodiments, or make equivalent replacements for some of the technical features therein. These modifications or replacements do not deviate the essence of the corresponding technical solutions from the spirit and scope of the technical solutions of the various embodiments of the present application, and should all be included in the scope of protection of the present application.

Claims

1. An intelligent work order submission method based on a large model, characterized in that: The method comprises: Obtaining work order event information, and inputting the work order event information into a pre-trained intelligent work order submission model for semantic recognition to obtain event semantics; Determine a work order template table according to the event semantics, and perform entity recognition on the work order event information to obtain a work order entity; In the work order template table, information is filled in according to the entity type of the work order entity to obtain a target work order table, and the work order approval flow is determined according to the form identifier of the work order template table and the event semantics; The work order is submitted to the target work order table according to the work order approval flow.

2. The intelligent work order submission method based on a large model according to claim 1, characterized in that: Before inputting the work order event information into the pre-trained intelligent work order submission model for semantic recognition, the method further includes: Inputting the work order sample into the intelligent work order submission model for text encoding to obtain sample encoding features, and performing vector conversion on the sample encoding features to obtain a sample encoding vector; Calculating the similarity between the sample coding vector and the preset coding vector to obtain a sample vector similarity, and determining the semantics corresponding to the maximum sample vector similarity as the sample semantics; Matching the sample semantics with the historical work order database to obtain a sample work order table; Segmenting the work order sample to obtain sample segmentations, and identifying entities on the sample segmentations to obtain sample entities; Determining a sample urgency level according to the sample semantics, and determining a model loss according to the sample coding features, the sample work order table, and the sample entity; The parameters of the intelligent work order submission model are updated according to the model loss until the intelligent work order submission model converges, thereby obtaining the pre-trained intelligent work order submission model.

3. The intelligent work order submission method based on a large model according to claim 1 is characterized in that: Determining a work order approval flow according to the form identifier of the work order template table and the event semantics includes: Determining a time vocabulary in the work order event information according to the event semantics, and determining an urgency level of the work order according to the time vocabulary; Obtaining a user ID corresponding to the work order event information, and determining user permissions based on the user ID; The form identifier of the work order template table, the user authority and the work order urgency level are combined to obtain level combination information, and the level combination information is matched with the approval flow query table to obtain the work order approval flow.

4. The intelligent work order submission method based on a large model according to claim 1, characterized in that: Fill in the information according to the entity type of the work order entity to obtain the target work order table, including: Obtaining a fill-in identifier of a fill-in box in the work order template table, and matching the entity type of the work order entity with the fill-in identifier to obtain an entity fill-in position; In the work order template table, information of the work order entity is filled in according to the entity filling position to obtain the target work order table.

5. The intelligent work order submission method based on a large model according to claim 4 is characterized in that: After filling in information for the work order entity according to the entity filling position to obtain the target work order table, the method further includes: Determine a check box combination in the fill-in box according to the form identifier of the work order template table, and combine the information in the fill-in boxes in the check box combination to obtain check information; Performing semantic recognition on the verification information to obtain verification semantics, and performing semantic contradiction detection on the verification semantics; If the semantic contradiction detection fails, a semantic contradiction prompt is given to the check box combination.

6. The intelligent work order submission method based on a large model according to claim 4 is characterized in that: After filling in information for the work order entity according to the entity filling position to obtain the target work order table, the method further includes: If no information is filled in the fill-in box, obtaining the box identifier of the fill-in box, and matching the box identifier with the necessary information table corresponding to the form identifier of the work order template table; If the box identifier successfully matches the necessary information table, the fill-in box is determined to be a missing box, and the type of entity to be filled in is determined according to the box identifier; Combining the entity type to be filled in with the event semantics to obtain semantic combination information; Performing vector conversion on the semantic combination information to obtain a semantic combination vector, and performing similarity calculation between the combination vector and the historical work order vector to obtain the work order similarity; Determine the historical work order table corresponding to the maximum work order similarity as the associated work order table, and determine the associated information box in the associated work order table according to the entity type to be filled in; Fill the information in the associated information box into the corresponding missing box, and mark the missing box as missing.

7. The intelligent work order submission method based on a large model according to claim 6, characterized in that: After calculating the similarity between the combined vector and the historical work order vector to obtain the work order similarity, the following steps are also included: If the work order similarity is less than a similarity threshold, a missing prompt is generated according to the missing box, and a missing question is asked to the user corresponding to the work order event information according to the missing prompt; If supplementary information regarding the missing question is received, the missing box is filled in according to the supplementary information.

8. An intelligent work order submission system based on a large model, characterized in that: The system comprises: A semantic recognition module is used to obtain work order event information and input the work order event information into a pre-trained intelligent work order submission model for semantic recognition to obtain event semantics; An entity recognition module is used to determine a work order template table according to the event semantics, and perform entity recognition on the work order event information to obtain a work order entity; An information filling module is used to fill in information in the work order template table according to the entity type of the work order entity to obtain a target work order table, and determine the work order approval flow according to the form identifier of the work order template table and the event semantics; The work order submission module is used to submit the work order to the target work order table according to the work order approval flow.

9. The intelligent work order submission system based on a large model according to claim 8, characterized in that: The semantic recognition module is also used for: Inputting the work order sample into the intelligent work order submission model for text encoding to obtain sample encoding features, and performing vector conversion on the sample encoding features to obtain a sample encoding vector; Calculating the similarity between the sample coding vector and the preset coding vector to obtain a sample vector similarity, and determining the semantics corresponding to the maximum sample vector similarity as the sample semantics; Matching the sample semantics with the historical work order database to obtain a sample work order table; Segmenting the work order sample to obtain sample segmentations, and identifying entities on the sample segmentations to obtain sample entities; Determining a sample urgency level according to the sample semantics, and determining a model loss according to the sample coding features, the sample work order table, and the sample entity; The parameters of the intelligent work order submission model are updated according to the model loss until the intelligent work order submission model converges, thereby obtaining the pre-trained intelligent work order submission model.

10. A computer-readable storage medium storing a computer program, characterized in that: When the computer program is executed by a processor, the steps of the method according to any one of claims 1 to 7 are implemented.

Images