A work order information extraction method, device, equipment and storage medium thereof
By performing word segmentation and vector probability score calculation on work order data, the problem of manual work order review being unable to quickly and accurately understand the core information of work orders has been solved, thus achieving high efficiency and accuracy in work order processing.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- GUANGZHOU BAIYUN DISTRICT GOVERNMENT SERVICES & DATA ADMINISTRATION BUREAU
- Filing Date
- 2026-02-13
- Publication Date
- 2026-05-29
AI Technical Summary
In existing technologies, manual review of 12345 hotline work orders makes it difficult to quickly and accurately understand the core information of the work orders, resulting in low processing efficiency and accuracy.
By acquiring work order data, an initial string is constructed and segmented. Information is extracted using the vector probability scores of the word sequence, including encoding, attention calculation, linear classification, and probability transformation. The label sequence is acquired and decoded, and post-processing is performed to obtain the information extraction results of the work order data.
It enables the rapid and accurate extraction of key information from work order data, assisting business personnel in accurately grasping the core content of work orders and improving the efficiency and accuracy of work order processing.
Smart Images
Figure CN122114852A_ABST
Abstract
Description
Technical Field
[0001] This application relates to the field of work order information extraction technology, specifically to a work order information extraction method, apparatus, equipment and its storage medium. Background Technology
[0002] With the rapid development of information technology, the digital transformation of government services has become an unstoppable trend. In this process, the efficiency and accuracy of the 12345 government service hotline's work order processing have increasingly become important standards for measuring the level of government services.
[0003] Order review is a crucial step in the 12345 hotline's work order processing, and this step relies on an accurate understanding and interpretation of the work order information. Currently, during order review, the 12345 hotline's work order information needs to be read word by word to understand the context before making a judgment. Given the massive amount of work order data and the lengthy work order texts, it is impossible to quickly and accurately understand the core information of the work order, thus affecting the efficiency and accuracy of work order processing. Summary of the Invention
[0004] In view of this, this application provides a method, apparatus, equipment and storage medium for extracting work order information, which solves the technical problems of difficulty in ensuring data accuracy and low work order processing efficiency due to manual review.
[0005] The first aspect of this application provides a method for extracting work order information, including: Respond to the work order information extraction request and obtain the work order data to be extracted; The work order data is combined with preset extraction instructions to form an initial string; The initial string is segmented to obtain a word sequence; Using the probability scores corresponding to the word sequence, information is extracted from the word sequence to obtain the information extraction results corresponding to the work order data.
[0006] Further, the step of using the probability scores corresponding to the word sequence to extract information from the word sequence and obtaining the information extraction result corresponding to the work order data specifically includes: The word sequence is encoded to obtain a vector representation of the word sequence; Attention is calculated on the vector representation to obtain the corresponding context vector; Perform linear classification on the context vector to obtain the original classification score; The original classification scores are subjected to probability transformation to obtain probability scores; Based on the probability score, information is extracted from the word sequence to obtain the information extraction result of the work order data.
[0007] Further, the step of extracting information from the word sequence based on the probability score to obtain the information extraction result of the work order data specifically includes: Obtain the tag sequence corresponding to the word sequence; The label sequences whose probability scores are greater than a preset threshold are taken as target label sequences, and the target label sequences are decoded to obtain text sequences; The text sequence is post-processed to obtain the information extraction results of the work order data.
[0008] Further, the step of using the label sequences whose probability scores are greater than a preset threshold as target label sequences and decoding the target label sequences to obtain text sequences specifically includes: Determine the starting label in the label sequence whose probability score is greater than a preset threshold; Determine the termination tags in the tag sequence whose probability scores are greater than a preset threshold; The tag sequence between the start tag and the end tag is taken as the target tag sequence; The target label sequence is decoded to obtain a text sequence.
[0009] Furthermore, when the information extraction result is the target extracted information, after post-processing the text sequence to obtain the information extraction result of the work order data, the process further includes: The target extraction information is then standardized.
[0010] A second aspect of this application provides a work order information extraction device, comprising: The work order data acquisition module is used to respond to work order information extraction requests and acquire the work order data to be extracted. A string construction module is used to construct an initial string by combining the work order data with a preset extraction instruction; The word segmentation module is used to segment the initial string to obtain a word sequence; The information extraction module is used to extract information from the word sequence using the probability scores corresponding to the word sequence, and obtain the information extraction result corresponding to the work order data.
[0011] Furthermore, the information extraction module includes: The vector representation submodule is used to encode the word sequence to obtain a vector representation of the word sequence; The attention calculation submodule is used to perform attention calculation on the vector representation to obtain the corresponding context vector. The linear transformation submodule is used to perform linear classification on the context vector to obtain the original classification score; The probability transformation submodule is used to perform probability transformation on the original classification scores to obtain probability scores; The information extraction submodule is used to extract information from the word sequence based on the probability score to obtain the information extraction result of the work order data.
[0012] Furthermore, the information extraction submodule includes: The tag sequence acquisition submodule is used to acquire the tag sequence corresponding to the word sequence; The tag sequence decoding submodule is used to take the tag sequences with probability scores greater than a preset threshold as target tag sequences, and decode the target tag sequences to obtain text sequences; The post-processing submodule is used to perform post-processing on the text sequence to obtain the information extraction results of the work order data.
[0013] A third aspect of this application provides a computer device, including: a processor and a memory; the processor is connected to the memory, wherein the memory is used to store a computer program, and the processor is used to invoke the computer program to cause the computer device to execute the work order information extraction method.
[0014] A fourth aspect of this application provides a computer-readable storage medium storing a computer program adapted to be loaded and executed by a processor, such that a computer device having the processor performs the work order information extraction method.
[0015] One of the above technical solutions has the following advantages and effects: One of the work order information extraction methods in the above scheme includes: Respond to the work order information extraction request and obtain the work order data to be extracted; The work order data is combined with preset extraction instructions to form an initial string; The initial string is segmented to obtain a word sequence; By using the vector probability scores corresponding to the word sequence, information is extracted from the word sequence to obtain the information extraction results corresponding to the work order data.
[0016] As can be seen from the above scheme, the above work order information extraction method obtains the work order data to be extracted, constructs an initial string by combining the work order data with a preset extraction instruction, and performs word segmentation on the initial string to obtain a word sequence including the work order data and the preset extraction instruction information. The information is extracted from the word sequence using the vector probability score corresponding to the word sequence to obtain the information extraction result corresponding to the work order data. This enables the extraction of key information from the work order data, assists business personnel in accurately grasping the core content of the work order, and improves the efficiency and accuracy of work order processing. Attached Figure Description
[0017] The accompanying drawings, which are included to provide a further understanding of this application and form part of this application, illustrate exemplary embodiments and are used to explain this application, but do not constitute an undue limitation of this application. In the drawings: Figure 1 This is a flowchart illustrating a work order information extraction method provided in this application; Figure 2 This is a flowchart illustrating step 104 of the work order information extraction method provided in this application; Figure 3 This is a flowchart illustrating step S52 of the work order information extraction method provided in this application; Figure 4 This is a schematic diagram of the structure of a work order information extraction device provided in this application; Figure 5 This is a structural schematic diagram of a work order information extraction device provided in this application. Detailed Implementation
[0018] The 12345 hotline's closing reply marks the end of the entire process for a work order, from "acceptance to dispatch to processing to feedback," signifying that the request has entered the final confirmation stage. The closing reply typically includes four core parts: the verification status of the issue, the handling measures, the results, and the next steps. Because the quality of the closing reply directly affects the public's evaluation of government services, it must meet standardized requirements such as positively responding to the request, listing the policy basis, and clearly explaining the handling process. Currently, 12345 closing replies are mainly written and responded to by staff of the handling unit (i.e., the government department or agency specifically responsible for handling the issue), and undergo internal process review and approval. Therefore, when faced with a massive volume of 12345 hotline work orders and lengthy work order texts, it becomes difficult to quickly and accurately understand the core information of the work orders, thus affecting the efficiency and accuracy of work order processing.
[0019] In view of this, embodiments of this application provide a method, apparatus, device and storage medium for extracting work order information, which solves the technical problem that the prior art cannot quickly and accurately obtain the core information of work orders, thereby affecting the processing efficiency and accuracy of work orders.
[0020] The technical solutions in the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this application, and not all embodiments. Based on the embodiments of this application, all other embodiments obtained by those skilled in the art without creative effort are within the protection scope of this application.
[0021] In the description of this application, it should be noted that the terms "center," "upper," "lower," "left," "right," "vertical," "horizontal," "inner," and "outer," etc., indicate the orientation or positional relationship based on the orientation or positional relationship shown in the accompanying drawings. They are used only for the convenience of describing this application and simplifying the description, and do not indicate or imply that the device or element referred to must have a specific orientation, or be constructed and operated in a specific orientation. Therefore, they should not be construed as limitations on this application. Furthermore, the terms "first," "second," and "third" are used for descriptive purposes only and should not be construed as indicating or implying relative importance.
[0022] In the description of this application, it should be noted that, unless otherwise expressly specified and limited, the terms "installation," "connection," and "linking" should be interpreted broadly. For example, they can refer to a fixed connection, a detachable connection, or an integral connection; they can refer to a mechanical connection or an electrical connection; they can refer to a direct connection or an indirect connection through an intermediate medium; and they can refer to the internal connection between two components. Those skilled in the art can understand the specific meaning of the above terms in this application based on the specific circumstances.
[0023] Please see Figure 1 , Figure 1 This is a schematic diagram of the structure of a work order information extraction method according to this application.
[0024] Depend on Figure 1 As can be seen, the work order information extraction method in this embodiment includes: 101. Respond to the work order information extraction request and obtain the work order data to be extracted.
[0025] It is understood that the work order data may come from the 12345 hotline information system, obtained through API interface or database export, or it may come from telephone, website, mini program, official account, APP, State Council, provincial network, letters and visits, etc.
[0026] 102. Construct an initial string by combining the work order data with the preset extraction instructions.
[0027] In this embodiment, the preset extraction instruction is mainly used to specify the target information to be extracted, such as the citizen's name and gender, phone number, ID number, involved party, time of incident, address of incident, processing time, and citizen's request. It can be understood that the target information to be extracted can be one or more.
[0028] It should be noted that work order data generally includes, but is not limited to: citizen information, core demands, matter information, completion response content, and feedback information. Citizen information includes the citizen's name, gender, and contact number; core demands are the problem description and the citizen's request; matter information includes the involved parties, the time of the incident, and the address involved; completion response content includes the processing procedure, the legal basis for the processing, the result of the processing, and the response opinion; and feedback information includes citizen satisfaction and follow-up records.
[0029] For example, the work order data is the completion reply content of the 12345 hotline work order. The completion reply content includes the work order number, the replying unit, the reply time, and an explanation of the handling status of the issues raised by citizens.
[0030] Preset extraction instructions: involved entity, processing unit, processing result.
[0031] The work order data and the preset extraction instructions are combined to construct an initial string, that is, the above-mentioned completion reply content and the preset extraction instructions are combined to construct an initial string containing the completion reply content and the preset extraction instructions.
[0032] 103. Perform word segmentation on the initial string to obtain a word sequence.
[0033] In this implementation, the initial string is segmented into words, that is, the initial string is split into characters to obtain a word sequence. Specifically, regular expressions can be used to segment the initial string.
[0034] 104. Using the vector probability scores corresponding to the word sequences, information is extracted from the word sequences to obtain the information extraction results corresponding to the work order data.
[0035] Please see Figure 2 The following are the specific steps of step 104 in the work order information extraction method of this application. In this embodiment, information is extracted from the word sequence using the vector probability score corresponding to the word sequence to obtain the information extraction result corresponding to the work order data. Specifically, the steps include: S1. Encode the word sequence to obtain a vector representation of the word sequence.
[0036] It is understandable that encoding the word sequence involves converting the word sequence into a digital form that a computer can understand and compute, such as a high-dimensional vector of 512*768.
[0037] S2. Perform attention calculation on the vector representation to obtain the corresponding context vector.
[0038] It should be noted that the context vector is the result of a weighted sum of the value vectors after the query vector interacts with all key vectors through an attention mechanism. The result is a high-dimensional, computable mathematical object rich in semantic, syntactic, and task information, transformed according to its context. The mathematical expression for the context vector X is: .
[0039] Where α is the attention weight and V is the value vector.
[0040] S3. Perform linear classification on the context vector to obtain the original classification score.
[0041] In this embodiment, the context vector is linearly classified to obtain the original classification score. Specifically, the context vector is mapped to a Logits vector using a linear classifier. The mathematical expression for the Logits vector s is: .
[0042] Where W is the weight matrix and b is the bias vector.
[0043] It should be noted that the linear classifier determines its weight matrix and bias vector based on a large amount of manually labeled work order data.
[0044] A Logits vector is a numerical vector in which each value represents the original classification score of a word in a word sequence, indicating whether it belongs to a particular category. Each element of the Logits vector corresponds to a possible category, which is the label sequence of the word sequence.
[0045] For example, the Logits vector might be a vector like [3.2, -1.5, 0.8,...], referred to as the "classification raw score".
[0046] S4. Perform probability transformation on the original classification scores to obtain probability scores.
[0047] In this embodiment, the original classification scores are converted into probability scores using the softmax function, i.e., the Logits vector is normalized by applying the softmax function. The formula for the softmax function is: .
[0048] Where zi is the i-th element in the Logits vector, K is the length of the Logits vector, and e is the natural constant (approximately 2.71828).
[0049] For example, the Logits vector [3.2, -1.5, 0.8,...] might become [0.95, 0.02, 0.03,...] after being normalized by the Softmax function, and is called the "probability score".
[0050] S5. Based on the probability score, extract information from the word sequence to obtain the information extraction results of the work order data.
[0051] Please see Figure 3 This refers to the specific steps of step S5 in this embodiment. In this embodiment, step S5 specifically includes: S51. Obtain the label sequence corresponding to the word sequence.
[0052] S52. Select the label sequences with probability scores greater than a preset threshold as target label sequences, and decode the target label sequences to obtain text sequences.
[0053] Preferably, the preset threshold is greater than or equal to 0.8. It is understood that in some embodiments, there may be cases where the probability scores are all less than the preset threshold, indicating that the target information to be extracted is not present in the work order data.
[0054] In this embodiment, step S52 specifically includes: S521. Determine the starting label in the label sequence whose probability score is greater than a preset threshold.
[0055] S522. Determine the termination label in the label sequence whose probability score is greater than a preset threshold.
[0056] S523. The label sequence between the start label and the end label is taken as the target label sequence.
[0057] S524. Decode the target label sequence to obtain the text sequence.
[0058] For example, if the work order's address is "gas convenience point" and the extraction instruction is "address in question," then after calculation, the probability scores of the tag sequences corresponding to "gas" and "point" are both greater than the preset threshold of 0.8. Therefore, the tag sequence corresponding to "gas convenience point" is the target tag sequence. Decoding the tag sequence corresponding to "gas convenience point" will yield "gas convenience point."
[0059] In this embodiment, decoding the target label sequence means decoding the target label sequence in a digital form that a computer can understand and calculate into a text form that a human can understand.
[0060] S53. Post-process the text sequence to obtain the information extraction results of the work order data.
[0061] It is understandable that post-processing of text sequences includes, but is not limited to: splitting, extracting, cleaning, and assembling.
[0062] The cleaning process mainly involves removing unnecessary symbols, spaces, etc.
[0063] It should be noted that, in other embodiments, when the information extraction result is the target extracted information, after post-processing the text sequence to obtain the information extraction result of the work order data, the method further includes: Standardize the target extraction information.
[0064] The standardization of extracted target information aims to eliminate ambiguity caused by abbreviations, acronyms, or non-standard expressions, thereby improving data consistency and usability. This standardization process includes unifying time information, addresses, and other data into standardized formats. For example, it converts different time expressions such as "May 20, 2024," "20-05-2024," and "May 20, 2024" from different sources into the "YYYY-MM-DD" format, where YYYY represents a 4-digit year, MM represents a 2-digit month, and DD represents a 2-digit day. It also ensures accurate identification and standardized writing of hierarchical information such as province, city, district, and street in addresses, eliminating data discrepancies caused by differences in address representation.
[0065] The work order information extraction method in this embodiment obtains the work order data to be extracted, constructs an initial string by combining the work order data with preset extraction instructions, and performs word segmentation on the initial string to obtain a word sequence including work order data and preset extraction instruction information. The method then uses the vector probability scores corresponding to the word sequence to extract information from the word sequence, thereby obtaining the information extraction result corresponding to the work order data. This enables the extraction of key information from the work order data, assists business personnel in accurately grasping the core content of the work order, and improves the efficiency and accuracy of work order processing.
[0066] The above are embodiments of a work order information extraction method provided in this application. The following are embodiments of a work order information extraction device provided in this application.
[0067] Please see Figure 4 In this embodiment, the work order information extraction device includes: A second aspect of this application provides a work order information extraction device, comprising: The data acquisition module 201 is used to respond to work order information extraction requests and obtain the work order data to be extracted.
[0068] The string construction module 202 is used to construct an initial string by combining work order data with preset extraction instructions.
[0069] The word segmentation module 203 is used to segment the initial string to obtain a word sequence.
[0070] The information extraction module 204 is used to extract information from the word sequence by using the vector probability scores corresponding to the word sequence, and obtain the information extraction results corresponding to the work order data.
[0071] In this embodiment, the information extraction module 204 includes: The vector representation submodule is used to encode word sequences to obtain vector representations of the word sequences.
[0072] The attention calculation submodule is used to perform attention calculations on the vector representation to obtain the corresponding context vector.
[0073] The linear transformation submodule is used to perform linear classification on the context vector to obtain the original classification score.
[0074] The probability transformation submodule is used to perform probability transformation on the original classification scores to obtain probability scores.
[0075] The information extraction submodule is used to extract information from the word sequence based on the probability score to obtain the information extraction result of the work order data.
[0076] The information extraction submodule includes: The tag sequence retrieval submodule is used to retrieve the tag sequence corresponding to the word sequence.
[0077] The label sequence decoding submodule is used to take the label sequences with probability scores greater than a preset threshold as target label sequences and decode the target label sequences to obtain text sequences.
[0078] The post-processing submodule is used to post-process the text sequence to obtain the information extraction results of the work order data.
[0079] The work order information extraction device in this embodiment acquires the work order data to be extracted through the data acquisition module 201, constructs an initial string by combining the work order data with preset extraction instructions using the string construction module 202, and performs word segmentation processing on the initial string through the word segmentation processing module 203 to obtain a word sequence including work order data and preset extraction instruction information. The information extraction module 204 extracts information from the word sequence based on the vector probability score corresponding to the word sequence to obtain the information extraction result corresponding to the work order data. This enables the extraction of key information from the work order data, assists business personnel in accurately grasping the core content of the work order, and improves the efficiency and accuracy of work order processing.
[0080] Please see Figure 5This application embodiment also provides a work order information extraction device, which includes a processor 301 and a memory 302.
[0081] The memory 302 is used to store program code and transfer program code to the processor 301.
[0082] The processor 301 is used to execute the steps of the work order information extraction method of the foregoing embodiments according to the instructions in the program code.
[0083] This application also provides a computer device, including a processor and a memory; the processor is connected to the memory, wherein the memory is used to store a computer program, and the processor is used to call the computer program so that the computer device executes the work order information extraction method.
[0084] Those skilled in the art will understand that embodiments of this application can be provided as methods, systems, or computer program products. Therefore, this application can take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, this application can take the form of a computer program product embodied on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.
[0085] This application is described with reference to flowchart illustrations and / or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of this application. It will be understood that each block of the flowchart illustrations and / or block diagrams, and combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, special-purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, generate instructions for implementing the flowchart... Figure 1 One or more processes and / or boxes Figure 1 A device that provides the functions specified in one or more boxes.
[0086] It should also be noted that the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such a process, method, article, or apparatus. Without further limitation, an element defined by the phrase "comprising one..." does not exclude the presence of other identical elements in the process, method, article, or apparatus that includes said element.
[0087] The above description is merely an embodiment of this application and is not intended to limit the scope of this application. Various modifications and variations can be made to this application by those skilled in the art. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of this application should be included within the scope of the claims of this application.
Claims
1. A method for extracting work order information, characterized in that, The method includes: Respond to the work order information extraction request and obtain the work order data to be extracted; The work order data is combined with preset extraction instructions to form an initial string; The initial string is segmented to obtain a word sequence; Using the probability scores corresponding to the word sequence, information is extracted from the word sequence to obtain the information extraction results corresponding to the work order data.
2. The work order information extraction method according to claim 1, characterized in that, The step of extracting information from the word sequence using the probability score corresponding to the word sequence to obtain the information extraction result corresponding to the work order data specifically includes: The word sequence is encoded to obtain a vector representation of the word sequence; Attention is calculated on the vector representation to obtain the corresponding context vector; Perform linear classification on the context vector to obtain the original classification score; The original classification scores are subjected to probability transformation to obtain probability scores; Based on the probability score, information is extracted from the word sequence to obtain the information extraction result of the work order data.
3. The work order information extraction method according to claim 2, characterized in that, The step of extracting information from the word sequence based on the probability score to obtain the information extraction result of the work order data specifically includes: Obtain the tag sequence corresponding to the word sequence; The label sequences whose probability scores are greater than a preset threshold are taken as target label sequences, and the target label sequences are decoded to obtain text sequences; The text sequence is post-processed to obtain the information extraction results of the work order data.
4. The work order information extraction method according to claim 3, characterized in that, The step of taking the tag sequences with probability scores greater than a preset threshold as target tag sequences and decoding the target tag sequences to obtain text sequences specifically includes: Determine the starting label in the label sequence whose probability score is greater than a preset threshold; Determine the termination tags in the tag sequence whose probability scores are greater than a preset threshold; The tag sequence between the start tag and the end tag is taken as the target tag sequence; The target label sequence is decoded to obtain a text sequence.
5. The work order information extraction method according to claim 3, characterized in that, When the information extraction result is the target extracted information, after post-processing the text sequence to obtain the information extraction result of the work order data, the process further includes: The target extraction information is then standardized.
6. A work order information extraction device, characterized in that, The device includes: The work order data acquisition module is used to respond to work order information extraction requests and acquire the work order data to be extracted. A string construction module is used to construct an initial string by combining the work order data with a preset extraction instruction; The word segmentation module is used to segment the initial string to obtain a word sequence; The information extraction module is used to extract information from the word sequence using the probability scores corresponding to the word sequence, and obtain the information extraction result corresponding to the work order data.
7. The work order information extraction device according to claim 6, characterized in that, The information extraction module includes: The vector representation submodule is used to encode the word sequence to obtain a vector representation of the word sequence; The attention calculation submodule is used to perform attention calculation on the vector representation to obtain the corresponding context vector. The linear transformation submodule is used to perform linear classification on the context vector to obtain the original classification score; The probability transformation submodule is used to perform probability transformation on the original classification scores to obtain probability scores; The information extraction submodule is used to extract information from the word sequence based on the probability score to obtain the information extraction result of the work order data.
8. The work order information extraction device according to claim 7, characterized in that, The information extraction submodule includes: The tag sequence acquisition submodule is used to acquire the tag sequence corresponding to the word sequence; The tag sequence decoding submodule is used to take the tag sequences with probability scores greater than a preset threshold as target tag sequences, and decode the target tag sequences to obtain text sequences; The post-processing submodule is used to perform post-processing on the text sequence to obtain the information extraction results of the work order data.
9. A computer device, characterized in that, include: Processor and memory; The processor is connected to a memory, wherein the memory is used to store a computer program, and the processor is used to call the computer program so that the computer device executes the work order information extraction method according to any one of claims 1-5.
10. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores a computer program adapted to be loaded and executed by a processor, such that a computer device having the processor performs the work order information extraction method according to any one of claims 1-5.