Key work order identification method and device, equipment and storage medium
By combining a dual recognition mechanism with regular expressions and deep learning models, the accuracy problem of recognizing complex semantic work orders in traditional methods has been solved, enabling efficient automatic detection and risk control of bank work orders.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2025-11-27
- Publication Date
- 2026-04-14
AI Technical Summary
In the digital operation and maintenance and customer service system of banks, existing technologies and traditional methods are unable to efficiently identify key work orders with complex semantics, resulting in delayed service response and potential economic losses.
A dual recognition mechanism is adopted. First, key keywords are identified through regular expression matching function. Then, deep learning is performed using the Bert-BiLSTM-IDCNN-CRF model. Combined with user terminal review and model training optimization, the automatic detection of key work orders is realized.
It improved the accuracy and response speed of key work orders, enhanced risk control capabilities, and ensured the accurate classification and timely handling of work orders with complex semantics.
Smart Images
Figure CN121859074A_ABST
Abstract
Description
Technical Field
[0001] This application relates to the field of information technology, and to, but is not limited to, a method, apparatus, device, and storage medium for identifying key work orders. Background Technology
[0002] In the current digital operation and maintenance and customer service system of banks, the efficient management of massive business work orders has become a core link in ensuring business continuity and improving service quality. With the continuous expansion of business scale, the number of work orders has exploded, and the semantics of work order content are diverse and the expression forms are complex. Traditional methods relying on manual screening or static rule matching are no longer sufficient to meet the actual needs for timely and accurate identification of key work orders. This limitation not only leads to delayed service response but also triggers systemic risks and potential economic losses. In recent years, the rapid development of deep learning technology in the field of natural language processing has provided an effective path for the semantic understanding and classification of complex text, significantly improving the model's ability to identify and analyze unstructured data. However, current solutions using deep learning technology to assist in work order processing have shortcomings such as limited semantic understanding capabilities and reliance on fixed rules and keyword configurations. Summary of the Invention
[0003] In view of this, the key work order identification method, apparatus, equipment, and storage medium provided in the embodiments of this application can construct a key work order detection mechanism with high accuracy, fully understand the deep semantics of business work orders, and realize automatic detection of key work orders.
[0004] The key work order identification method, apparatus, equipment, and storage medium provided in this application embodiment are implemented as follows: One aspect of this application provides a method for identifying key work orders, the method comprising: The work order to be identified is subjected to key keyword identification to obtain the first identification result; If the first identification result indicates that the work order is of priority, mark the work order to be identified as a priority work order; If the first identification result is non-priority, the work order to be identified is input into the priority work order identification model to obtain the second identification result; If the second identification result indicates that the work order to be identified is a priority, then mark the work order to be identified as a priority work order. If the second identification result is non-priority, the work order to be identified will be marked as a regular work order.
[0005] In one possible implementation, key keywords are identified in the work order to be identified to obtain the first identification result, including: The work order to be identified is preprocessed to obtain the work order text information; The first recognition result is obtained by examining the work order text information through at least one preset regular expression matching function.
[0006] In one possible implementation, each regular expression function has a corresponding entity category. The work order text information is examined using at least one pre-defined regular expression matching function to obtain a first recognition result, including: Call at least one regular expression matching function to check whether there are key keywords representing the corresponding entity category in the work order text information; If the work order text information includes at least one key keyword, the first identification result is determined to be the key keyword. If the work order text does not contain key keywords, the first identification result is determined to be non-keywords.
[0007] In one possible implementation, the work order to be identified is input into the key work order identification model to obtain a second identification result, including: The work order to be identified is preprocessed to obtain the work order text information; The work order text information is segmented using the Bert word segmenter to obtain a keyword sequence; Input the keyword sequence into the key work order recognition model to obtain the entity label sequence; The entity label sequence is examined by at least one preset regular expression matching function to obtain a second recognition result.
[0008] In one possible implementation, the preprocessing includes the removal of special characters and text normalization.
[0009] In one possible implementation, the method also includes: The marked work order to be identified will be sent to the user terminal for review; If the work order to be identified is marked based on the first identification result, and the review result reported by the user terminal is that the marking is incorrect, the regular expression matching function is adjusted based on the work order to be identified.
[0010] In one possible implementation, the method also includes: The marked work order to be identified will be sent to the user terminal for review; If the work order to be identified is marked based on the second identification result, and the review result reported by the user terminal is that the marking is incorrect, the work order to be identified will be added to the training set of the key work order identification model. The work order recognition model was retrained using the updated training set.
[0011] Another aspect of the embodiments of this application provides a key work order identification device, the device comprising: The first identification module is used to identify key keywords in the work order to be identified and obtain the first identification result. The first marking module is used to mark the work order to be identified as a key work order if the first identification result indicates that it is a key work order. The second identification module is used to input the work order to be identified into the key work order identification model when the first identification result is non-key, and obtain the second identification result. The second marking module is used to mark the work order to be identified as a key work order when the second identification result indicates that it is a key work order. The third marking module is used to mark the work order to be identified as a regular work order if the second identification result is non-priority.
[0012] In one possible implementation, the first identification module is further used for: The work order to be identified is preprocessed to obtain the work order text information; The first recognition result is obtained by examining the work order text information through at least one preset regular expression matching function.
[0013] In one possible implementation, each regular expression function has a corresponding entity category, and the first identification module is further used for: Call at least one regular expression matching function to check whether there are key keywords representing the corresponding entity category in the work order text information; If the work order text information includes at least one key keyword, the first identification result is determined to be the key keyword. If the work order text does not contain key keywords, the first identification result is determined to be non-keywords.
[0014] In one possible implementation, the second identification module is further used for: The work order to be identified is preprocessed to obtain the work order text information; The work order text information is segmented using the Bert word segmenter to obtain a keyword sequence; Input the keyword sequence into the key work order recognition model to obtain the entity label sequence; The entity label sequence is examined by at least one preset regular expression matching function to obtain a second recognition result.
[0015] In one possible implementation, the preprocessing includes the removal of special characters and text normalization.
[0016] In one possible implementation, the device further includes: The first review module is used to send the marked work orders to the user terminal for review. The first adjustment module is used to adjust the regular expression matching function based on the work order to be identified when the work order to be identified is marked based on the first identification result and the review result fed back by the user terminal is that the marking is incorrect.
[0017] In one possible implementation, the device further includes: The second review module is used to send the marked work orders to the user terminal for review. The training set update module is used to add the work order to be identified to the training set of the key work order identification model when the work order to be identified is marked based on the second identification result and the review result fed back by the user terminal is that the marking is incorrect. The second adjustment module is used to retrain the work order recognition model using the updated training set.
[0018] The electronic device provided in this application includes a memory and a processor. The memory stores a computer program that can run on the processor. When the processor executes the program, it implements the method described in this application.
[0019] The computer-readable storage medium provided in this application embodiment stores a computer program thereon, which, when executed by a processor, implements the method provided in this application embodiment.
[0020] The key work order identification method, apparatus, equipment, and storage medium provided in this application embodiment identify key keywords in the work order to be identified, obtaining a first identification result. If the first identification result indicates the work order is key, it is marked as a key work order. If the first identification result indicates the work order is not key, it is input into a key work order identification model to obtain a second identification result. If the second identification result indicates the work order is key, it is marked as a key work order. If the second identification result indicates the work order is not key, it is marked as a regular work order. This application embodiment combines two identification methods to jointly identify work order types, constructing a highly accurate key work order detection mechanism. It can fully understand the deep semantics of business work orders, achieve automatic detection of key work orders, assist in the timely discovery and handling of key work orders, and improve risk prevention and control capabilities. Attached Figure Description
[0021] To more clearly illustrate the technical solutions in the embodiments of this application, the drawings used in the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of this application. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.
[0022] Figure 1 A flowchart illustrating a method for identifying key work orders according to an embodiment of this application is shown. Figure 2 A schematic diagram illustrating a key work order identification process according to an embodiment of this application is shown; Figure 3 This diagram illustrates a key work order identification system according to an embodiment of this application. Figure 4 A schematic diagram of a key work order identification device according to an embodiment of this application is shown; Figure 5 A schematic diagram of an electronic device according to an embodiment of this application is shown. Detailed Implementation
[0023] To make the objectives, technical solutions, and advantages of the embodiments of this application clearer, the specific technical solutions of this application will be further described in detail below with reference to the accompanying drawings of the embodiments of this application. The following embodiments are used to illustrate this application, but are not intended to limit the scope of this application.
[0024] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this application belongs. The terminology used herein is for the purpose of describing embodiments of this application only and is not intended to limit this application.
[0025] In the following description, references are made to “some embodiments,” which describe a subset of all possible embodiments. However, it is understood that “some embodiments” may be the same subset or different subsets of all possible embodiments and may be combined with each other without conflict.
[0026] It should be noted that the terms "first, second, third" used in the embodiments of this application are used to distinguish similar or different objects and do not represent a specific order of objects. It can be understood that "first, second, third" can be interchanged in a specific order or sequence where permitted, so that the embodiments of this application described herein can be implemented in an order other than that illustrated or described herein.
[0027] The key work order identification method of this application embodiment can be executed by any electronic device capable of controlling the hot-rolled strip steel system. This electronic device may include, but is not limited to, mobile phones, wearable devices (such as smartwatches, smart bracelets, smart glasses, etc.), tablet computers, laptops, vehicle terminals, PCs (Personal Computers), etc. The functions implemented by this method can be achieved by a processor in the electronic device calling program code. Of course, the program code can be stored in a computer storage medium. Therefore, the electronic device includes at least a processor and a storage medium.
[0028] The key work order identification method of this application can be used in application scenarios for identifying key work orders of any type. For example, this application embodiment can be applied to the application scenario of identifying key work orders in IT service management work orders. Alternatively, it can also be applied to the application scenario of identifying key work orders in banking work orders.
[0029] The related technologies that use computer algorithms to assist in processing work orders typically have the following drawbacks: Some methods for assisting in processing work orders rely on fixed rules and keyword configurations. They are less adaptable to the diversity of expressions, semantic variations, and synonyms in work order texts, and cannot accurately identify the same meaning implied in different expressions. This results in a high rate of missed detections and false detections in practical applications, affecting the stability and reliability of the recognition effect.
[0030] Some methods for assisting in processing work orders have limited semantic understanding capabilities, and most rely on traditional deep learning models to extract features and classify work order texts. When faced with highly repetitive, short, and non-standardized expressions in business work orders, these models lack deep semantic modeling capabilities, are prone to semantic confusion, and struggle to accurately distinguish similar categories or identify implicit intentions, resulting in limited classification accuracy.
[0031] Some methods for processing work orders are weak in handling unstructured text. Work order texts are often semantically redundant, loosely structured, and contain a large amount of colloquial or vague expressions. Existing deep learning methods typically rely on shallow semantic matching or static representations, which struggle to fully extract key elements and subjective intent in complex, non-standardized scenarios, leading to decreased recognition performance.
[0032] Some methods for assisting in work order processing only focus on work order classification and lack the ability to identify key work orders. They mainly limit work order identification to category determination (such as difficult work orders, faulty work orders, etc.) or work order allocation, lacking the ability to identify and prioritize key work orders based on business value or potential risks. The system cannot combine multi-dimensional information such as work order urgency, customer importance, and scope of event impact to determine whether a work order is a critical work order that needs to be prioritized.
[0033] Therefore, the technical problem solved by the embodiments of this application is how to automatically identify key work orders and ensure the reliability of the identification results.
[0034] The following describes in detail the key work order identification scheme of this application embodiment with reference to the accompanying drawings.
[0035] Figure 1 A flowchart illustrating a method for identifying key work orders according to an embodiment of this application is shown. Figure 1 As shown, the key work order identification method in this application embodiment may include the following steps S10-S50.
[0036] For ease of description, the key work order identification method of this application embodiment is described using an electronic device as the execution subject. It should be understood that the execution subject of this application embodiment can also be a processor or chip in an electronic device, and this application embodiment does not impose any limitations.
[0037] Step S10: The electronic device performs key keyword recognition on the work order to be identified and obtains the first recognition result.
[0038] In one possible implementation, the electronic device first acquires the work order to be identified, and then performs preliminary identification on the work order to obtain a first identification result. The work order to be identified can be determined through human-computer interaction between the user and a user terminal, and transmitted to the electronic device via the user terminal. The preliminary identification method of the electronic device for the work order to be identified can be key keyword recognition, and the first identification result can include whether it is a key or non-key item. A key first identification result indicates that the work order to be identified is a key work order, and a non-key first identification result indicates that the work order to be identified is not a key work order.
[0039] Optionally, in this embodiment, the electronic device can preprocess the work order to be identified after acquiring it, removing special symbols and redundant spaces, and performing text standardization to obtain the work order text information. The removed special symbols may include "XX", "@", or a newline character, etc. Keyword recognition can be achieved using at least one preset regular expression matching function; that is, after acquiring the work order text information, the electronic device can check the work order text information using at least one preset regular expression matching function to obtain a first recognition result.
[0040] In some embodiments, each regular expression function determined by the electronic device has a corresponding entity category, used to accurately match whether the work order text contains keywords representing that entity category. For example, the regular expression for the entity category "emergency fault" can be designed as: r"emergency(fault|error|abnormal)|outbreak(fault|problem)". The electronic device can sequentially call each regular expression matching function to check whether the work order text information contains key keywords representing the corresponding entity category. If the work order text information includes at least one key keyword, the first identification result is determined to be key. If the work order text information does not contain key keywords, the first identification result is determined to be non-key.
[0041] For example, in this embodiment, the electronic device can determine the regular expression function based on historical work orders. Specifically, a large number (e.g., 5000) of historical business work orders in the current domain can be collected, covering any scenario such as system failures, business risks, and emergency handling. Then, each historical business work order can be manually annotated through human-computer interaction to filter out core features with key work order attributes and determine multiple (e.g., 24) key entity categories. For example, these could include the following multiple entity categories: Urgency level categories: "Emergency Fault", "Immediate Response", "Extremely Urgent", "Customer Awaiting", "Unexpected Problem"; Key business categories: "VIP clients", "major clients", "private banking clients", "strategic clients", "important clients", and "public opinion". Risk level classification: "High-risk business", "Abnormal funds", "Compliance warning"; System impact categories: "Processing delayed", "Transaction interrupted", "Abnormal alarm", "Service stopped", "Transaction time taken"; The categories involving monetary amounts include: "duplicate billing", "advance payment", "duplicate payment", "duplicate deduction", and "duplicate transfer".
[0042] Furthermore, after obtaining multi-category body labels, the electronic device can use Python's `re` library to write regular expression matching functions for multiple entity categories. Each function corresponds to an entity category and is used to accurately match whether the work order text contains that category of feature expressions. For example, the regular expression for "emergency fault" can be designed as: `r"emergency(fault|error|abnormal)|sudden(fault|problem)"`. Optionally, after writing the regular expression matching functions for multiple entity categories, the electronic device can integrate these functions into a structured rule library, stored in JSON format. Each rule contains "entity category" and "matching function," facilitating the electronic device's use for key keyword identification and subsequent updates.
[0043] For example, when an electronic device receives a work order to be identified, it first performs text standardization processing, removing special symbols (such as "xx", "@", "newline", etc.) and redundant spaces to obtain the work order text information. Then, it sequentially calls regular expression matching functions for multiple entity categories in the rule base to check whether the work order text information contains a corresponding entity category. If any entity category matches (i.e., the work order text information contains at least one key keyword corresponding to that entity category), the first identification result is determined to be important. Optionally, the electronic device can also record the matched entity categories. If all matching functions fail (i.e., the work order text information does not contain any key keyword corresponding to the category label), the first identification result is determined to be unimportant.
[0044] Step S20: If the first identification result indicates that the work order to be identified is a key work order, mark the work order to be identified as a key work order.
[0045] In one possible implementation, after the electronic device performs keyword recognition on the work order to be identified in step S10, it obtains a first recognition result. If the first recognition result indicates the work order is a key work order, it proves that the work order to be identified already contains keywords that characterize it as a key work order, and the electronic device can directly mark the work order to be identified as a key work order without further recognition.
[0046] Step S30: If the first identification result is non-key, input the work order to be identified into the key work order identification model to obtain the second identification result.
[0047] In one possible implementation, after the electronic device performs keyword recognition on the work order to be identified in step S10, it obtains a first recognition result. If the first recognition result indicates the work order is not a key work order, it proves that the work order does not contain keywords that characterize it as a key work order. However, since the text content may contain other content not covered by the regular expression matching function but still characterizing the importance of the work order, the electronic device can further perform a second recognition on the work order to be identified using a key work order recognition model, obtaining a second recognition result.
[0048] Optionally, the electronic device may perform a second recognition of the work order to be identified, which may include preprocessing the work order to obtain work order text information. Then, the work order text information is segmented using a Bert word segmenter to obtain a keyword sequence. The keyword sequence is then input into a key work order recognition model to obtain an entity label sequence. Finally, the entity label sequence is checked using at least one preset regular expression matching function to obtain a second recognition result. The preprocessing method described above is similar to step S10 and will not be repeated here. The key work order recognition model can be a trained Bert-BiLSTM-IDCNN-CRF model, used to output entity labels representing the entity category corresponding to each keyword (such as "B-funds abnormal", "O", etc.). The electronic device checks the entity label sequence using at least one preset regular expression matching function to obtain the second recognition result by sequentially determining whether each entity label in the entity label sequence represents the entity category corresponding to the regular expression matching function. If at least one entity label can represent the entity category corresponding to at least one regular expression matching function, the electronic device determines that the second recognition result is key. If no entity label can characterize the entity category corresponding to at least one regular expression matching function, the electronic device determines that the second recognition result is non-key.
[0049] In some embodiments, the electronic device can train a key work order identification model. The training set for training this model can be determined from historical work orders acquired over time. Specifically, the electronic device can collect a large number (e.g., 5000) of historical business work orders from the current domain, covering any scenario such as system failures, business risks, and emergency handling. Then, each historical business work order can be manually labeled (i.e., labeled as "key work order" and "ordinary work order") through human-computer interaction. Simultaneously, key entity tags in the text are labeled using the BIO (Begin-Inside-Outside) annotation method. Here, "B-Entity Category" indicates the starting position of an entity (e.g., "B-High-Risk Business"). "I-Entity Category" indicates the middle position of an entity (e.g., "I-High-Risk Business"). "O" indicates the position of a non-key entity.
[0050] For example, the work order text "A VIP customer reported an anomaly in funds during a transfer, requiring urgent verification" is annotated as follows: "A": O, "VIP": B-Important Customer, "Customer": I-Important Customer, "Customer": I-Important Customer, "Report": O, "Feedback": O, "Transfer": O, "Account": O, "Time": O, "Prompt": O, "Show": O, "Funds": B-Important Customer, "Funds": I-Important Customer, "Abnormal": I-Important Customer, "Normal": I-Important Customer, ",": O, "Need": O, "Urgent": B-Important Customer, "Urgent": I-Important Customer, "Verify": I-Important Customer, "Check": I-Important Customer.
[0051] For example, after data annotation is completed, the electronic device can divide the annotated data into a training set (4000 records), a validation set (500 records), and a test set (500 records) according to a preset ratio (e.g., 5000 data points divided in an 8:1:1 ratio) for model training, parameter tuning, and performance evaluation. The electronic device can use the PyTorch framework to build a key work order recognition model that integrates multi-level feature extraction. The specific network structure may include: Bert pre-trained layer: The Chinese Bert-base pre-trained model is used as the underlying encoder to convert the pre-processed work order text (word sequence after word segmentation) into word embedding vectors (768 dimensions) containing contextual semantics, thus solving the problems of polysemy and semantic ambiguity.
[0052] BiLSTM layer: After the BERT output, a bidirectional LSTM (BiLSTM) network (128 hidden layers, 2 layers) is connected. The forward LSTM captures the left-to-right contextual information of the text, and the backward LSTM captures the right-to-left contextual information. The fused output contains a feature vector containing global semantics, which enhances the model's understanding of long-distance dependencies (such as the association between "needs to be processed within 2 hours" and "urgent").
[0053] IDCNN layer: Introducing the dilated convolution (IDCNN) layer (kernel size 3, dilation rate 2), the receptive field is expanded while maintaining computational efficiency through multiple dilated convolutions, extracting richer features, capturing local text features (such as the combination of "funds" and "abnormality"), making up for the shortcomings of BiLSTM in local semantic extraction, and improving the sensitivity to phrase-level features.
[0054] CRF layer: After the IDCNN output, a Conditional Random Field (CRF) layer is connected. By learning the transition probabilities between labels (e.g., "B-Immediate Processing" is more likely to be followed by "I-Immediate Processing" rather than "O"), the sequence dependency problem of entity labels is solved and boundary recognition errors are reduced (e.g., avoiding the mistaken splitting of "Immediate Processing" into two entities, "Immediate" and "Processing").
[0055] Furthermore, during the training of the above model, the electronic device can set the following training parameters: the optimizer uses AdamW, with an initial learning rate of 5e-5 and weight decay of 1e-3. The loss function is the negative log-likelihood loss (NLLLoss) inherent in the CRF layer, used to optimize the label sequence prediction accuracy. The training epochs are 30, employing an early stopping strategy (training stops if the F1 score on the validation set does not improve for 5 consecutive epochs). The batch size is 16, using gradient accumulation to address the instability problem of small sample training. After completing the model training, the model can be evaluated using entity recognition precision, recall, and F1 score on the test set as evaluation metrics. The goal of the model evaluation is to ensure that the F1 score of the entity labels is ≥0.85, guaranteeing the model's ability to recognize latent features.
[0056] Step S40: If the second identification result indicates that the work order to be identified is a key work order, mark the work order to be identified as a key work order.
[0057] In one possible implementation, if the second identification result indicates a priority, the electronic device can determine that the work order to be identified is a priority work order and mark it as such. Simultaneously, it can also mark the entity tag corresponding to the work order to be identified.
[0058] Step S50: If the second identification result is non-priority, mark the work order to be identified as a regular work order.
[0059] In one possible implementation, if the second recognition result is not important, the electronic device determines that the work order to be identified has not yet had its key entity tag identified after the dual recognition process, and therefore is not a work order that requires attention. Thus, the work order to be identified can be directly marked as a regular work order.
[0060] In some embodiments, to gradually improve the accuracy of the identification results during the use of the key work order identification system of this application, the electronic device can also periodically optimize the key work order identification system based on the marked work orders to be identified. That is, the electronic device can periodically send the marked work orders to be identified to the user terminal for review. The user terminal determines whether the marking is correct or incorrect based on the content of the work order to be identified and the marking result, obtains the review result, and feeds back the review result to the electronic device.
[0061] Optionally, if the work order to be identified is marked based on the first identification result, and the user terminal reports an incorrect labeling result, the electronic device can adjust the regular expression matching function based on the work order to be identified. If the work order to be identified is marked based on the second identification result, and the user terminal reports an incorrect labeling result, the electronic device can add the work order to be identified to the training set of the key work order identification model. The work order identification model is then retrained using the updated training set.
[0062] After receiving the review results, the electronic device can record misjudgment cases (such as "key work orders were misjudged as ordinary work orders" or "ordinary work orders were misjudged as key work orders"). Further analysis determines which identification process caused the misjudgment. If the misjudgment stems from a keyword matching omission during the key keyword identification process (e.g., adding the feature "emergency response to regulatory spot checks"), the key work order feature rule base is updated, adding the corresponding entity category and regular expression matching function. If the misjudgment originates from insufficient identification by the key work order identification model (e.g., missing detection of "hidden risk" entities), the misjudged work order is added to the model training set, re-labeled, and the model fine-tuned to improve its adaptability to new features.
[0063] Figure 2 This diagram illustrates a key work order identification process according to an embodiment of this application. Figure 2 As shown, users can submit work orders to be identified to the electronic device's key work order identification system through human-computer interaction with the user terminal. After receiving the work order transmitted from the user terminal, the electronic device performs key keyword identification on the work order to obtain a first identification result. It then further determines whether the first identification result indicates that the work order is a key work order. If the first identification result indicates it is a key work order, the work order is marked as a key work order. If the first identification result indicates it is not a key work order, the work order is input into the key work order identification model to obtain a second identification result. If the second identification result indicates it is a key work order, the work order is marked as a key work order. If the second identification result indicates it is not a key work order, the work order is marked as a regular work order.
[0064] Optionally, in this embodiment of the application, the electronic device may further include a front-end interface for displaying the marking results of the work orders to be identified to the user. Specifically, the electronic device may highlight work orders marked as "priority work orders" on the front end and process them with priority, while work orders marked as "ordinary work orders" will be entered into the processing queue according to the regular process.
[0065] Furthermore, the marked work orders to be identified can also be used as a historical business work order database to optimize the key keyword identification module and the key work order identification model.
[0066] Figure 3A schematic diagram of a key work order identification system according to an embodiment of this application is shown. Figure 3 As shown, in some embodiments, the electronic device in this application can construct a training set using more than 5,000 stored historical work orders, and then construct a key work order feature rule library including multiple regular expression matching functions using manual feedback rules and historical work orders. Simultaneously, a key work order recognition model (Bert-BiLSTM-IDCNN-CRF model) is trained based on the training set using manual feedback rules and annotation tools.
[0067] Optionally, after a user submits a work order, the electronic device treats the work order as a work order to be identified. Based on multiple regular expression matching functions in the key work order feature rule base, the device identifies key keywords in the work order to be identified, obtaining a first identification result. The device further determines whether the first identification result indicates that the work order to be identified is a key work order. If the first identification result indicates it is a key work order, the work order to be identified is marked as a key work order. If the first identification result indicates it is not a key work order, the work order to be identified is input into the key work order identification model for model identification, obtaining a second identification result. If the second identification result indicates it is a key work order, the work order to be identified is marked as a key work order. If the second identification result indicates it is not a key work order, the work order to be identified is marked as a regular work order. The marking results of the work order to be identified are then displayed through the front-end interface.
[0068] Based on the above technical features, this application embodiment combines the rapid screening of keyword recognition with the accurate recognition of deep learning model recognition to achieve "dual detection" of key work orders. This ensures real-time response to work orders with obvious features and improves the recognition accuracy of work orders with hidden features, providing an automated and intelligent solution for risk prevention and control of business work orders.
[0069] It should be understood that although the steps in the above flowcharts are shown sequentially according to the arrows, these steps are not necessarily executed in the order indicated by the arrows. Unless explicitly stated herein, there is no strict order restriction on the execution of these steps, and they can be executed in other orders. Moreover, at least some steps in the above flowcharts may include multiple sub-steps or multiple stages. These sub-steps or stages are not necessarily completed at the same time, but can be executed at different times. The execution order of these sub-steps or stages is not necessarily sequential, but can be performed alternately or in turn with other steps or at least some of the sub-steps or stages of other steps.
[0070] Based on the foregoing embodiments, this application provides a key work order identification device, which includes various modules and units included in each module, and can be implemented by a processor; of course, it can also be implemented by specific logic circuits; in the implementation process, the processor can be a central processing unit (CPU), microprocessor (MPU), digital signal processor (DSP) or field programmable gate array (FPGA), etc.
[0071] Figure 4 A schematic diagram of a key work order identification device according to an embodiment of this application is shown. Figure 4 As shown, the key work order identification device in this application embodiment includes: The first identification module 40 is used to identify key keywords in the work order to be identified and obtain the first identification result. The first marking module 41 is used to mark the work order to be identified as a key work order when the first identification result is key. The second identification module 42 is used to input the work order to be identified into the key work order identification model when the first identification result is non-key, and obtain the second identification result. The second marking module 43 is used to mark the work order to be identified as a key work order when the second identification result is key. The third marking module 44 is used to mark the work order to be identified as a regular work order when the second identification result is non-priority.
[0072] In one possible implementation, the first identification module 40 is further configured to: The work order to be identified is preprocessed to obtain the work order text information; The first recognition result is obtained by examining the work order text information through at least one preset regular expression matching function.
[0073] In one possible implementation, each regular expression function has a corresponding entity category, and the first identification module 40 is further used for: Call at least one regular expression matching function to check whether there are key keywords representing the corresponding entity category in the work order text information; If the work order text information includes at least one key keyword, the first identification result is determined to be the key keyword. If the work order text does not contain key keywords, the first identification result is determined to be non-keywords.
[0074] In one possible implementation, the second identification module 42 is further used for: The work order to be identified is preprocessed to obtain the work order text information; The work order text information is segmented using the Bert word segmenter to obtain a keyword sequence; Input the keyword sequence into the key work order recognition model to obtain the entity label sequence; The entity label sequence is examined by at least one preset regular expression matching function to obtain a second recognition result.
[0075] In one possible implementation, the preprocessing includes the removal of special characters and text normalization.
[0076] In one possible implementation, the device further includes: The first review module is used to send the marked work orders to the user terminal for review. The first adjustment module is used to adjust the regular expression matching function based on the work order to be identified when the work order to be identified is marked based on the first identification result and the review result fed back by the user terminal is that the marking is incorrect.
[0077] In one possible implementation, the device further includes: The second review module is used to send the marked work orders to the user terminal for review. The training set update module is used to add the work order to be identified to the training set of the key work order identification model when the work order to be identified is marked based on the second identification result and the review result fed back by the user terminal is that the marking is incorrect. The second adjustment module is used to retrain the work order recognition model using the updated training set.
[0078] The descriptions of the above device embodiments are similar to those of the above method embodiments, and have similar beneficial effects. For technical details not disclosed in the device embodiments of this application, please refer to the descriptions of the method embodiments of this application for understanding.
[0079] It should be noted that, in the embodiments of this application... Figure 4 The module division of the key work order identification device shown is illustrative and represents only one logical functional division; in actual implementation, other division methods may be used. Furthermore, the functional units in the various embodiments of this application can be integrated into one processing unit, exist as separate physical units, or be integrated into one unit with two or more units. The integrated units can be implemented in hardware, as software functional units, or a combination of both.
[0080] It should be noted that, in the embodiments of this application, if the above-described methods are implemented as software functional modules and sold or used as independent products, they can also be stored in a computer-readable storage medium. Based on this understanding, the technical solutions of the embodiments of this application, or the parts that contribute to related technologies, 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 an electronic device to execute all or part of the methods described in the various embodiments of this application. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, read-only memory (ROM), magnetic disks, or optical disks. Thus, the embodiments of this application are not limited to any specific hardware and software combination.
[0081] Figure 5 A schematic diagram of an electronic device according to an embodiment of this application is shown. For example... Figure 5 As shown in the figure, this application provides an electronic device, which can be a server, and its internal structure diagram can be as follows. Figure 5 As shown, the electronic device includes a processor 520, a memory, and a transceiver 540 connected via a system bus 510. The processor 520 provides computing and control capabilities. The memory includes a non-volatile storage medium 531 and internal memory 532. The non-volatile storage medium 531 stores an operating system, computer programs, and a database. The internal memory 532 provides an environment for the operation of the operating system and computer programs in the non-volatile storage medium 531. The database stores data. The transceiver 540 communicates with external terminals via a network connection. When the computer program is executed by the processor 520, it implements the methods described above.
[0082] This application provides a computer-readable storage medium storing a computer program thereon, which, when executed by a processor 520, implements the steps of the method provided in the above embodiments.
[0083] This application provides a computer program product containing instructions that, when run on a computer, cause the computer to perform the steps in the method provided in the above-described method embodiments.
[0084] Those skilled in the art will understand that Figure 5 The structure shown is merely a block diagram of a portion of the structure related to the present application and does not constitute a limitation on the electronic device to which the present application is applied. The specific electronic device may include more or fewer components than shown in the figure, or combine certain components, or have different component arrangements.
[0085] In one possible implementation, the shooting prompting device provided in this application can be implemented as a computer program, which can be configured as follows: Figure 5 The device operates on the electronic device shown. The memory of the electronic device can store various program modules that make up the above-described apparatus. The computer program composed of the various program modules causes the processor 520 to execute the steps of the methods in the various embodiments of this application described in this specification.
[0086] It should be noted that the descriptions of the storage medium and device embodiments above are similar to the descriptions of the method embodiments above, and have similar beneficial effects. For technical details not disclosed in the storage medium, storage medium, and device embodiments of this application, please refer to the descriptions of the method embodiments of this application for understanding.
[0087] It should be understood that the phrases "one embodiment," "an embodiment," or "some embodiments" mentioned throughout the specification mean that a specific feature, structure, or characteristic related to an embodiment is included in at least one embodiment of this application. Therefore, phrases such as "in one possible implementation," "in one embodiment," or "in some embodiments" appearing throughout the specification do not necessarily refer to the same embodiment. Furthermore, these specific features, structures, or characteristics can be combined in any suitable manner in one or more embodiments. It should be understood that in the various embodiments of this application, the sequence numbers of the above-described processes do not imply a sequential order of execution; the execution order of each process should be determined by its function and internal logic, and should not constitute any limitation on the implementation process of the embodiments of this application. The sequence numbers of the above-described embodiments are merely for descriptive purposes and do not represent the superiority or inferiority of the embodiments. The descriptions of the various embodiments above tend to emphasize the differences between the various embodiments; their similarities or commonalities can be referred to mutually, and for the sake of brevity, they will not be repeated here.
[0088] In this article, the term "and / or" is merely a description of the relationship between related objects, indicating that there can be three kinds of relationships. For example, object A and / or object B can represent three situations: object A exists alone, object A and object B exist simultaneously, and object B exists alone.
[0089] It should be noted that, in this document, 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. Unless otherwise specified, 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 that element.
[0090] In the several embodiments provided in this application, it should be understood that the disclosed devices and methods can be implemented in other ways. The embodiments described above are merely illustrative. For example, the division of modules is only a logical functional division, and in actual implementation, there may be other division methods, such as: multiple modules or components can be combined, or integrated into another system, or some features can be ignored or not executed. In addition, the coupling, direct coupling, or communication connection between the various components shown or discussed can be through some interfaces, and the indirect coupling or communication connection between devices or modules can be electrical, mechanical, or other forms.
[0091] The modules described above as separate components may or may not be physically separate. The components shown as modules may or may not be physical modules. They may be located in one place or distributed across multiple network units. Some or all of the modules may be selected to achieve the purpose of this embodiment according to actual needs.
[0092] In addition, each functional module in the various embodiments of this application can be integrated into one processing unit, or each module can be a separate unit, or two or more modules can be integrated into one unit; the integrated modules can be implemented in hardware or in the form of hardware plus software functional units.
[0093] Those skilled in the art will understand that all or part of the steps of the above method embodiments can be implemented by hardware related to program instructions. The aforementioned program can be stored in a computer-readable storage medium. When the program is executed, it performs the steps of the above method embodiments. The aforementioned storage medium includes various media that can store program code, such as mobile storage devices, read-only memory (ROM), magnetic disks, or optical disks.
[0094] Alternatively, if the integrated units described above are implemented as software functional modules and sold or used as independent products, they can also be stored in a computer-readable storage medium. Based on this understanding, the technical solutions of the embodiments of this application, or the parts that contribute to related technologies, 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 an electronic device to execute all or part of the methods described in the various embodiments of this application. The aforementioned storage medium includes various media capable of storing program code, such as mobile storage devices, ROMs, magnetic disks, or optical disks.
[0095] The methods disclosed in the several method embodiments provided in this application can be arbitrarily combined without conflict to obtain new method embodiments.
[0096] The features disclosed in the several product embodiments provided in this application can be arbitrarily combined without conflict to obtain new product embodiments.
[0097] The features disclosed in the several method or device embodiments provided in this application can be arbitrarily combined without conflict to obtain new method or device embodiments.
[0098] The above description is merely an embodiment of this application, but the scope of protection of this application is not limited thereto. Any variations or substitutions that can be easily conceived by those skilled in the art within the scope of the technology disclosed in this application should be included within the scope of protection of this application. Therefore, the scope of protection of this application should be determined by the scope of the claims.
Claims
1. A method for identifying key work orders, characterized in that, The method includes: The work order to be identified is subjected to key keyword identification to obtain the first identification result; If the first identification result indicates that the work order to be identified is an important work order, then the work order to be identified will be marked as an important work order. If the first identification result is non-priority, the work order to be identified is input into the key work order identification model to obtain the second identification result; If the second identification result indicates that the work order to be identified is an important work order, then the work order to be identified will be marked as an important work order. If the second identification result is non-priority, the work order to be identified is marked as a regular work order.
2. The method according to claim 1, characterized in that, The process of identifying key keywords in the work order to be identified, and obtaining the first identification result, includes: The work order to be identified is preprocessed to obtain the work order text information; The work order text information is examined using at least one preset regular expression matching function to obtain a first recognition result.
3. The method according to claim 2, characterized in that, Each of the regular expression functions has a corresponding entity category. The step of checking the work order text information using at least one preset regular expression matching function to obtain a first recognition result includes: Call at least one regular expression matching function to check in turn whether there are key keywords representing the corresponding entity category in the work order text information; If the work order text information includes at least one key keyword, the first identification result is determined to be a key keyword. If the work order text information does not contain key keywords, the first identification result is determined to be non-keywords.
4. The method according to claim 1, characterized in that, The step of inputting the work order to be identified into the key work order identification model to obtain the second identification result includes: The work order to be identified is preprocessed to obtain the work order text information; The work order text information is segmented using the Bert word segmenter to obtain a keyword sequence; The keyword sequence is input into the key work order identification model to obtain the entity label sequence; The entity label sequence is examined by at least one preset regular expression matching function to obtain a second recognition result.
5. The method according to claim 2 or 4, characterized in that, The preprocessing process includes removing special symbols and text standardization.
6. The method according to claim 2, characterized in that, The method further includes: The marked work order to be identified is sent to the user terminal for review; If the work order to be identified is marked based on the first identification result, and the review result returned by the user terminal is a marking error, the regular expression matching function is adjusted based on the work order to be identified.
7. The method according to claim 1, characterized in that, The method further includes: The marked work order to be identified is sent to the user terminal for review; If the work order to be identified is marked based on the second identification result, and the review result returned by the user terminal is that the marking is incorrect, the work order to be identified is added to the training set of the key work order identification model. The work order recognition model was retrained using the updated training set.
8. A key work order identification device, characterized in that, The device includes: The first identification module is used to identify key keywords in the work order to be identified and obtain the first identification result. The first marking module is used to mark the work order to be identified as a key work order when the first identification result indicates that it is a key work order; The second identification module is used to input the work order to be identified into the key work order identification model when the first identification result is non-key, and obtain the second identification result; The second marking module is used to mark the work order to be identified as a key work order when the second identification result indicates that it is a key work order; The third marking module is used to mark the work order to be identified as a regular work order when the second identification result is non-priority.
9. An electronic device comprising a memory and a processor, the memory storing a computer program executable on the processor, characterized in that, When the processor executes the program, it implements the steps of the method according to any one of claims 1 to 7.
10. A computer-readable storage medium having a computer program stored thereon, characterized in that, When the computer program is executed by a processor, it implements the method as described in any one of claims 1 to 7.