Work order title generation method and device, equipment and storage medium

Abstract

The invention discloses a work order title generation method and device, equipment and a storage medium, and relates to the technical field of natural language processing, and the method comprises the steps: obtaining an initial model input sequence corresponding to a work order; segmenting the initial model input sequence according to feature types in the initial model input sequence to obtain target fields, and connecting the target fields based on importance of the target fields to obtain a target model input sequence and position codes of the target fields; inputting the target model input sequence into a work order title generation model to obtain a feature type coding sequence, and generating a work order title based on the target model input sequence, the feature type coding sequence and the position code; and sending the work order title to the client, so that the client receives and displays the work order title. The sequence of the input model is recombined according to the importance of the work order feature type, and the work order title is generated by using the recombined sequence and the large model, so that the efficiency and the accuracy of the title generation process are improved.

Description

Technical Field

[0001] The present invention relates to the technical field of natural language processing, and particularly relates to a method, device, equipment and storage medium for generating work order titles. Background Art

[0002] In the current information technology management and customer service fields, work order management is an important component. Work order management not only involves the reporting, tracking and resolution of problems, but also includes the effective recording and classification of these problems. A clear and accurate work order title can help the operation and maintenance team quickly identify the core content of the problem and formulate reasonable solutions. However, the traditional way of generating work order titles mainly relies on manual input, and this way has many limitations.

[0003] Currently, the traditional way of generating work order titles is to rely on manual work to generate work order titles. This way of generating work order titles depends on the experience of the staff. Therefore, the method of using manual work to generate work order titles has problems such as low work efficiency and poor accuracy of the generated work order titles. For this reason, how to efficiently generate accurate work order titles has become a technical problem to be solved at present. Summary of the Invention

[0004] In view of this, the purpose of the present invention is to provide a method, device, equipment and storage medium for generating work order titles, which can reorganize the sequence input into the model according to the importance of the work order feature types, and use the reorganized sequence and the large model to generate work order titles, improving the efficiency and accuracy of the title generation process. The specific solutions are as follows:

[0005] In a first aspect, the present application provides a method for generating a work order title, which is applied to a server and includes:

[0006] Obtain the parsed data corresponding to the target work order sent by the client, and obtain the initial model input sequence based on the parsed data;

[0007] Perform a segmentation process on the initial model input sequence according to the feature types in the initial model input sequence to obtain target fields corresponding to the respective feature types, and connect the target fields based on the importance of the respective target fields to obtain a target model input sequence and position encodings corresponding to the respective target fields;

[0008] Input the target model input sequence into a preset work order title generation model to obtain a feature type encoding sequence, and generate a work order title corresponding to the target work order based on the target model input sequence, the feature type encoding sequence and the position encoding;

[0009] Send the work order title to the client so that the client can receive and display the work order title.

[0010] Optionally, the obtaining of the initial model input sequence based on the parsed data includes:

[0011] Perform data cleaning on the parsed data to obtain corresponding cleaned data, and perform data format conversion on the cleaned data to obtain the initial model input sequence corresponding to the parsed data.

[0012] Optionally, the connecting of the target fields based on the importance of each target field includes:

[0013] Add a preset delimiter mark at the end of each target field to obtain the marked fields corresponding to each target field, and connect the marked fields based on the importance of each target field.

[0014] Optionally, the inputting of the target model input sequence into a preset work order title generation model to obtain a feature type coding sequence includes:

[0015] Input the target model input sequence into the preset work order title generation model, and use the preset work order title generation model to generate identifiers corresponding to each feature type in the target model input sequence;

[0016] Convert each identifier into a target dense vector, and obtain the feature type coding sequence based on each target dense vector.

[0017] Optionally, the generating of the work order title corresponding to the target work order based on the target model input sequence, the feature type coding sequence, and the position coding includes:

[0018] Obtain a first target vector corresponding to the target model input sequence and second target vectors corresponding to each feature type in the feature type coding sequence, and generate an initial embedding vector according to the first target vector, the second target vectors, and the position coding;

[0019] Use the preset work order title generation model to iterate the initial embedding vector to obtain a corresponding target embedding vector, and obtain the work order title corresponding to the target work order according to the target embedding vector.

[0020] In a second aspect, the present application provides a work order title generation method, which is applied to a client and includes:

[0021] Obtain the work order information corresponding to the target work order, and parse the work order information to obtain the parsed data corresponding to the target work order;

[0022] Send the parsed data to the server, so that the server can obtain the parsed data, obtain the initial model input sequence based on the parsed data, perform segmentation processing on the initial model input sequence according to the feature types in the initial model input sequence to obtain target fields corresponding to each feature type, and connect the target fields based on the importance of each target field to obtain the target model input sequence and the position encoding corresponding to each target field. Finally, generate the work order title corresponding to the target work order based on the target model input sequence, the feature type encoding sequence, and the position encoding, and send the work order title to the client; wherein, the feature type encoding sequence is a sequence generated by a preset work order title generation model using the target model input sequence;

[0023] Receive the work order title sent by the server and display the work order title.

[0024] Optionally, after displaying the work order title, it further includes:

[0025] Start a preset feedback acquisition mechanism, use the preset feedback acquisition mechanism to obtain the feedback information of the target user on the work order title, and save the feedback information to a preset storage space locally.

[0026] In a third aspect, the present application provides a work order title generation device, which is applied to the server and includes:

[0027] A data acquisition module, configured to acquire the parsed data corresponding to the target work order sent by the client, and obtain the initial model input sequence based on the parsed data;

[0028] A sequence segmentation module, configured to perform segmentation processing on the initial model input sequence according to the feature types in the initial model input sequence to obtain target fields corresponding to each feature type, and connect the target fields based on the importance of each target field to obtain the target model input sequence and the position encoding corresponding to each target field;

[0029] A work order title generation module, configured to input the target model input sequence into a preset work order title generation model to obtain a feature type encoding sequence, and generate the work order title corresponding to the target work order based on the target model input sequence, the feature type encoding sequence, and the position encoding;

[0030] A work order title sending module, configured to send the work order title to the client, so that the client can receive the work order title and display the work order title.

[0031] In a fourth aspect, the present application provides an electronic device, including:

[0032] A memory, configured to store a computer program;

[0033] A processor, configured to execute the computer program to implement the foregoing work order title generation method.

[0034] In a fifth aspect, the present application provides a computer-readable storage medium, configured to store a computer program, and when the computer program is executed by a processor, the foregoing work order title generation method is implemented.

[0035] In the present application, the server first obtains the parsed data corresponding to the target work order sent by the client, and obtains an initial model input sequence based on the parsed data. Then, according to the feature types in the initial model input sequence, the initial model input sequence is segmented to obtain target fields corresponding to each feature type, and the target fields are connected based on the importance of each target field to obtain a target model input sequence and position encodings corresponding to each target field. After that, the target model input sequence is input into a preset work order title generation model to obtain a feature type encoding sequence, and a work order title corresponding to the target work order is generated based on the target model input sequence, the feature type encoding sequence, and the position encoding. Finally, the work order title is sent to the client, so that the client can receive the work order title and display the work order title. It can be seen that the present application improves the work efficiency of work order title generation by using a work order title generation model; by splitting and reorganizing the initial model input sequence corresponding to the target work order according to the importance of the feature information corresponding to the target work order, and inputting the reorganized sequence into the model, the key points of the work order information can be highlighted, enabling the model to deepen the understanding of the work order information, and thus improving the accuracy of the generated work order title. Description of the Drawings

[0036] In order to more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the following will briefly introduce the drawings required for use in the description of the embodiments or the prior art. Obviously, the drawings in the following description are only the embodiments of the present invention. For those of ordinary skill in the art, other drawings can be obtained according to the provided drawings without creative efforts.

[0037] Figure 1 It is a flowchart of a work order title generation method disclosed in the present application;

[0038] Figure 2 Schematic diagram of the process of a work order title generation method disclosed in this application;

[0039] Figure 3 Schematic diagram of the working process of a work order title generation model disclosed in this application;

[0040] Figure 4 Flowchart of a work order title generation method disclosed in this application;

[0041] Figure 5 Schematic diagram of the structure of a work order title generation device disclosed in this application;

[0042] Figure 6 Schematic diagram of the structure of an electronic device disclosed in this application. Detailed implementation manners

[0043] Next, the technical solutions in the embodiments of the present invention will be clearly and completely described in conjunction with the accompanying drawings in the embodiments of the present invention. Obviously, the described embodiments are only a part of the embodiments of the present invention, rather than all the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those of ordinary skill in the art without creative efforts shall fall within the protection scope of the present invention.

[0044] Currently, the traditional way to generate a work order title is to rely on manual generation. This way has problems such as low work efficiency and poor accuracy of the generated work order title. For this reason, this application provides a work order title generation method. By reorganizing the sequence input to the model according to the importance of the work order feature type and using the reorganized sequence and the large model to generate the work order title, the efficiency and accuracy of the title generation process are improved.

[0045] See Figure 1 As shown, an embodiment of the present invention discloses a work order title generation method, which is applied to a server and includes:

[0046] Step S11, obtain the parsed data corresponding to the target work order sent by the client, and obtain the initial model input sequence based on the parsed data.

[0047] The overall process of generating the work order title in this embodiment is as Figure 2As shown, the user first creates a work order and uses the work order intelligent assistant to send the work order information to the work order title generation model to generate the corresponding work order title using the model. In this embodiment, it is first necessary to obtain the parsed data corresponding to the target work order and obtain the initial model input sequence based on the above parsed data. Among them, the process of obtaining the initial model input sequence based on the parsed data may specifically include: performing data cleaning on the parsed data to obtain the corresponding cleaned data, and performing data format conversion on the cleaned data to obtain the initial model input sequence corresponding to the parsed data; specifically, the system needs to preprocess various text data in the work order, that is, the parsed data, such as fields like problem description, creator information, region, and product line. It can be understood that the main purpose of the preprocessing is to clean and standardize the data to ensure that the data input into the model is clean and consistent; among them, text cleaning needs to remove noises such as HTML (HyperText Markup Language) tags, special symbols, non-printable characters, etc., and then convert all texts into a unified character set, such as UTF-8 (Universal Character Set / Unicode Transformation Format), that is, the initial model input sequence. The above process is also the operation of standardizing the text, including operations such as unifying the case of text data, and replacing or deleting numbers and special characters. By performing data cleaning and data format conversion on the text data corresponding to the target work order, the reliability of the data is ensured, and at the same time, it is convenient for the work order title generation model to understand the data, thereby improving the accuracy of work order title generation.

[0048] It should be noted that the work order title generation model in this embodiment is a pre-trained model. In the training stage, the basic representation model part of the model is pre-trained based on large-scale language data. By pre-training the model, the model is enabled to have basic language understanding ability, and on this basis, the model is enhanced according to the input structure characteristics of the work order, and further fine-tuned training is performed on the production data in the specific field of work order title generation. During the training process, the model will continuously adjust its internal parameters and use techniques such as cross-validation and early stopping to ensure that the model reaches the best performance. Model optimization not only includes adjusting model parameters but also includes periodically re-training the model to adapt to new business scenarios and data. In addition, by monitoring the performance of the model in real time and collecting user feedback, the system can continuously optimize the model to make it more accurate and reliable. It should be noted that, as Figure 3As shown in the figure, the work order title generation model mainly consists of two parts: the work order representation model and the title text generation model. Among them, the core of the work order representation model is a pre-trained language model based on the Bert (Bidirectional Encoder Representations from Transformers) architecture. The Bert model is a model based on the Transformer architecture. By pre-training a large amount of language data, it can understand the deep semantics of language. In this system, the Bert model is used to extract and understand the semantic information of the work order text. In order to enhance the model's ability to understand different types of work order attributes, a feature type embedding module is added on the basis of the Bert model. The main function of this module is to perform feature embedding on various attributes in the work order and convert them into a format that the model can understand. In this way, the model can not only understand the text content, but also perform more accurate analysis according to the specific attributes of the work order; while the title text generation model uses the decoder module in the Seq2Seq (Sequence to Sequence) architecture. This module directly uses the sequence of feature vectors output by the work order representation model as input; the task of the decoder is to extract the most critical information from the work order feature vectors and generate the corresponding title text. In this process, the decoder will use the attention mechanism to focus on the important information in the input sequence, so as to generate the title text that best matches the work order content. Through the work order title generation model, the system can generate accurate work order titles according to the parsed data corresponding to the target work order.

[0049] Step S12: Perform segmentation processing on the initial model input sequence according to the feature types in the initial model input sequence to obtain target fields corresponding to each feature type, and connect the target fields based on the importance of each target field to obtain a target model input sequence and the position encoding corresponding to each target field.

[0050] In this embodiment, first, it is necessary to split the sequence according to the feature types of the data in the initial model input sequence. Specifically, for the given work order data object in the initial model input sequence, the text contents of the problem description, creator information, region, and product line fields are tokenized. This is a process of decomposing a continuous text string into meaningful units (words or tokens). Tokenization is not just simply splitting by spaces, but also needs to consider language characteristics, such as abbreviations in English and non-space writing in Chinese. After tokenization, obtain , , and , that is, each target field.

[0051] In this embodiment, the process of connecting each target field based on the importance of each target field may specifically include: adding a preset delimiter mark at the end of each target field to obtain the marked fields corresponding to each target field, and connecting the marked fields based on the importance of each target field; specifically, connecting all the preprocessed and tokenized field text sequences to form a single input sequence, that is, a token sequence, which is also the target model input sequence; it can be understood that the connection order and method of tokenization are crucial for the model's understanding. Usually, the tokenizations are arranged according to their importance. For example, the problem description is placed at the beginning, followed by other auxiliary information such as the creator, region, and product line. This order can help the model capture key information more effectively. Before connecting each field, it is necessary to use the [SEP] mark at the end of each field except the last one. Adding the mark at the end of the field helps the work order title generation model build a clear information boundary internally. After connecting the marked fields, the obtained token sequence, such as is used as the input of the model, where is a special mark at the head of the input sequence, is a special mark for separating different field feature sequences. By connecting each field according to its importance, the model's understanding of the text is improved, ensuring the accuracy of the generated work order title; by adding delimiters at the end of each field, the model's understanding of the data can be further deepened, ensuring the reliability of the generated work order title.

[0052] Step S13: Input the target model input sequence into a preset work order title generation model to obtain a feature type coding sequence, and generate the work order title corresponding to the target work order based on the target model input sequence, the feature type coding sequence, and the position coding.

[0053] In this embodiment, the process of inputting the target model input sequence into the work order title generation model to obtain the feature type encoding sequence may specifically include: inputting the target model input sequence into a preset work order title generation model, and using the preset work order title generation model to generate identifiers corresponding to each feature type in the target model input sequence; converting each identifier into a target dense vector, and obtaining the feature type encoding sequence based on each target dense vector; it can be understood that in constructing an efficient and accurate work order intelligent classification model, feature type encoding is a crucial link, which allows the model to appropriately process and understand various input features; specifically, the above process of feature type encoding is composed of a learnable embedding layer and a corresponding feature type dictionary in the work order title generation model; it should be noted that before performing feature type encoding, the model first constructs a feature type dictionary, which includes all possible input feature types, such as work order problem description, work order creator information, region, product line, and special markers for sequence construction such as [CLS] and [SEP], and the dictionary assigns a unique ID (Identity document) to each feature, that is, the above-mentioned identifier, and these IDs will be used in the subsequent encoding process; in addition, the above embedding layer is learnable, which means that its parameters will be updated during the model training process. The main role of the embedding layer is to convert the ID of the feature type into a dense vector of a fixed length, and the dense vector can express the semantic and syntactic attributes of each feature in the model. The dimension of the embedding vector usually depends on the complexity of the model and the number of features to be processed. The larger the dimension, the richer the information that can be expressed, but at the same time the computational cost is also greater. In addition, the parameters of the embedding layer are usually randomly initialized before training, and then optimized during the training process by the backpropagation algorithm. Regarding the feature type encoding as a training parameter to participate in the training, after obtaining the above sequence X, record the feature type information corresponding to the elements in the input sequence , where represents the feature type ID of the element at position i in X, and the feature types include work order title, work order problem description, work order creator information, region, product line, and special markers. Input C into the embedding layer to obtain the feature type encoding sequence .

[0054] In this embodiment, the process of generating a work order title corresponding to a target work order based on a target model input sequence, a feature type encoding sequence, and a position encoding may specifically include: obtaining a first target vector corresponding to the target model input sequence and second target vectors corresponding to each feature type in the feature type encoding sequence, and generating an initial embedding vector according to the first target vector, the second target vectors, and the position encoding; using a preset work order title generation model to iterate on the initial embedding vector to obtain a corresponding target embedding vector, and obtaining the work order title corresponding to the target work order according to the target embedding vector; specifically, the first step of work order representation, that is, work order title generation, is to perform embedding calculation on the input sequence to convert each input text unit (such as a word or a token) into a corresponding vector form. This process usually involves the following three types of embeddings: Token embedding, which converts each word or symbol in the text into a predefined vector; it should be noted that these vectors are usually learned during the pre-training process of the work order title generation model and can capture the semantic information of the vocabulary; Feature type encoding, which provides a vector of its type information for each element in the input sequence, and the feature type encoding helps the model understand the role of each vocabulary in the text; Position encoding, since the model architecture used (such as Transformer) does not naturally process the order information of sequence data, the position encoding supplements this by adding the position information of each vocabulary, which enables the model to handle the order relationship of vocabulary, such as the semantics of phrase or sentence structure. All these embedding vectors are in the same dimensional space and are combined together by element-wise addition to form the final embedding vector , that is, the above initial embedding vector, and the calculation method is , where 、 and are token embedding, feature type encoding, and position encoding respectively. Finally, obtained after n calculations by the representation model encoding layer , that is, the above target embedding vector, and the calculation process is . Finally, using the vector sequence output in the previous step, the decoder module in the Seq2Seq architecture is used to generate the work order title text; it should be noted that the decoder extracts the most critical information from the work order feature vector, that is, the above target embedding vector, and generates the corresponding title text. During the decoding process, the attention mechanism is used to focus on the important information in the input sequence to ensure that the generated title text is highly relevant to the work order content. By vectorizing the data input to the model, the understanding degree of the model for text data can be improved, and thus the accuracy of the generated work order title is improved.

[0055] Step S14: Send the work order title to the client so that the client can receive the work order title and display the work order title.

[0056] In this embodiment, after the server generates the work order title corresponding to the target work order by using the work order title generation model, it is also necessary to send the work order title to the client so that the client can display the work order title and obtain the corresponding feedback information.

[0057] It can be seen that in this application, by performing data cleaning and data format conversion on the text data corresponding to the target work order, the reliability of the data is ensured, and at the same time, it is convenient for the work order title generation model to understand the data, thereby improving the accuracy of the work order title generation; by connecting each field according to the importance of the field, the understanding degree of the model for the text is improved, and the accuracy of the generated work order title is ensured; by adding a delimiter at the end of each field, the understanding of the data by the model can be further deepened, and the reliability of the generated work order title is ensured.

[0058] Based on the foregoing embodiments, this application describes the overall process that occurs on the server during the work order title generation process. Next, this application will elaborate on the work order title generation process on the client side. Refer to Figure 4 As shown, an embodiment of the present invention discloses a work order title generation method applied to a client, including:

[0059] Step S21, obtain the work order information corresponding to the target work order, and parse the work order information to obtain the parsed data corresponding to the target work order.

[0060] In this embodiment, it is first necessary to receive the work order data created by the user. Specifically, the front-end interface of the system provides an intuitive and easy-to-use work order creation page where the user can input work order description information, including but not limited to problem description, creator information, area, product line, etc.; this interface will support various input methods such as text input, selection boxes, date pickers, etc. to adapt to different types of work order requirements; in addition, the front-end also has an input verification function to ensure the correctness of the data format, such as verifying the email format, phone number, etc., reducing invalid requests for back-end processing, and moreover, the front-end can optimize the interface layout through user behavior analysis, enabling the user to fill in information more quickly and accurately. After obtaining the work order created by the user, it is also necessary to use the intelligent assistant to parse the input text, extract key information using natural language processing technology, and convert it into structured data that can be processed by the model, that is, obtain the parsed data corresponding to the target work order; it should be noted that the work order intelligent assistant is the main interface for the user to interact with the system, and it is integrated into the work order creation page of the work order system. The main functions of the intelligent assistant include receiving the problem description input by the user, converting this information into a structured model input format, sending it to the work order title generation model at the back end, and presenting the output result of the model to the user. By validating the user input data, the time for the system to process invalid information is reduced, and the reliability of the generated work order is improved; by optimizing the interface layout according to the user behavior characteristics, the user operation experience can be improved.

[0061] Step S22: Send the parsed data to the server so that the server can obtain the parsed data, obtain an initial model input sequence based on the parsed data, perform a segmentation process on the initial model input sequence according to the feature types in the initial model input sequence to obtain target fields corresponding to each feature type, and connect the target fields based on the importance of each target field to obtain a target model input sequence and position encodings corresponding to each target field. Finally, generate a work order title corresponding to the target work order based on the target model input sequence, feature type encoding sequence, and the position encodings, and send the work order title to the client; wherein, the feature type encoding sequence is a sequence generated by a preset work order title generation model using the target model input sequence.

[0062] In this embodiment, the parsed and structured data will be sent to the deep learning work order title generation model at the backend. It can be understood that this model is deployed on a high-performance server, capable of handling high-concurrency data requests to ensure the classification results are returned within the user's waiting time. The model uses pre-trained models such as BERT and GPT (Generative Pre-Trained Transformer, a natural language processing model), and is fine-tuned according to actual business requirements. The model deployment adopts distributed computing technologies such as Kubernetes container management to optimize resource utilization and improve the system's response speed and processing capacity. It should be noted that before sending the work order title back to the front end, the backend will also verify the generated work order results to ensure the validity of the work order title returned to the front end.

[0063] Step S23: Receive the work order title sent by the server and display the work order title.

[0064] In this embodiment, after the front end displays the obtained work order title, it further includes: starting a preset feedback acquisition mechanism, using the preset feedback acquisition mechanism to obtain the feedback information of the target user on the work order title, and saving the feedback information to a preset storage space locally. Specifically, the system fills back the work order title generation result to the order creation page and allows the user to view and modify the work order title to ensure that the finally submitted work order information meets the actual needs of the user. At the same time, the front end uses reasonable user interaction design to enable the user to easily modify the work order title. In addition, the system also realizes real-time data binding between the front and back ends to ensure timely response to user operations. To further optimize the work order title generation model and user experience, the system provides a feedback mechanism. Users can feedback the accuracy of the classification result through a simple click operation: "Correct" and "Incorrect" buttons are set beside each classification result, and the user can select according to the actual situation. Finally, the collected user feedback will be used as important data input into the subsequent model training and optimization process to help the system better learn and adapt to the specific needs of different users. By obtaining the user's feedback on the work order title and retraining the model according to the user's feedback information, the data understanding ability of the work order title generation model is further improved, ensuring the reliability of the generated work order data.

[0065] It can be seen that by using a deep learning model in this application, the system can generate a work order title within a few seconds, significantly shortening the time for work order processing, making the system no longer rely on manual input and judgment, thus improving the processing efficiency; by training the model with a large amount of historical work order data, the model can identify and understand important information in the problem description, thereby generating an accurate work order title. Compared with manual generation, the model can learn complex patterns and relationships from the data, reducing the impact of human errors and subjective biases and improving the accuracy of title generation; by training and optimizing with historical work order data, the intelligence level and efficiency of the model can be continuously improved; by analyzing historical data, the system can identify common problem patterns and trends, thereby formulating more effective problem-solving solutions.

[0066] See Figure 5 As shown, an embodiment of this application discloses a work order title generation device, which is applied to a server and includes:

[0067] A data acquisition module 11, configured to acquire parsed data corresponding to a target work order sent by a client, and acquire an initial model input sequence based on the parsed data;

[0068] A sequence splitting module 12, configured to perform splitting processing on the initial model input sequence according to the feature types in the initial model input sequence, so as to obtain target fields corresponding to the respective feature types, and connect the target fields based on the importance of the respective target fields, so as to obtain a target model input sequence and position encodings corresponding to the respective target fields;

[0069] A work order title generation module 13, configured to input the target model input sequence into a preset work order title generation model, so as to obtain a feature type coding sequence, and generate a work order title corresponding to the target work order based on the target model input sequence, the feature type coding sequence, and the position encoding;

[0070] A work order title sending module 14, configured to send the work order title to the client, so that the client receives the work order title and displays the work order title.

[0071] It can be seen that by using a work order title generation model to generate a work order title in this application, the work efficiency of work order title generation is improved; by splitting and reorganizing the initial model input sequence corresponding to the target work order according to the importance of the feature information corresponding to the target work order, and inputting the reorganized sequence into the model, the key points of the work order information can be highlighted, enabling the model to deepen its understanding of the work order information, and thus improving the accuracy of the generated work order title.

[0072] In some specific embodiments, the data acquisition module 11 may specifically include:

[0073] A data cleaning unit is configured to clean the parsed data to obtain corresponding cleaned data, and convert the data format of the cleaned data to obtain the initial model input sequence corresponding to the parsed data.

[0074] In some specific embodiments, the sequence splitting module 12 may specifically include:

[0075] A delimiter adding unit is configured to add a preset delimiter mark at the end of each target field to obtain a field with a mark corresponding to each target field, and connect the fields with marks based on the importance of each target field.

[0076] In some specific embodiments, the work order title generation module 13 may specifically include:

[0077] An identifier generation unit is configured to input the target model input sequence into the preset work order title generation model, and use the preset work order title generation model to generate identifiers corresponding to each feature type in the target model input sequence;

[0078] An identifier conversion unit is configured to convert each identifier into a target dense vector, and obtain a feature type coding sequence based on each target dense vector.

[0079] In some specific embodiments, the work order title generation module 13 may specifically include:

[0080] A vector obtaining unit is configured to obtain a first target vector corresponding to the target model input sequence and second target vectors corresponding to each feature type in the feature type coding sequence, and generate an initial embedding vector according to the first target vector, the second target vectors, and the position coding;

[0081] A work order title generation unit is configured to iterate the initial embedding vector by using the preset work order title generation model to obtain a corresponding target embedding vector, and obtain the work order title corresponding to the target work order according to the target embedding vector.

[0082] Furthermore, an embodiment of the present application also discloses an electronic device, Figure 6 It is a structural diagram of an electronic device 20 shown according to an exemplary embodiment, and the content in the figure should not be regarded as any limitation to the scope of use of the present application.

[0083] Figure 6Schematic diagram of the structure of an electronic device 20 provided by an embodiment of the present application. The electronic device 20 may specifically include: at least one processor 21, at least one memory 22, a power supply 23, a communication interface 24, an input / output interface 25, and a communication bus 26. Among them, the memory 22 is used to store a computer program, and the computer program is loaded and executed by the processor 21 to implement the relevant steps in the work order title generation method disclosed in any of the foregoing embodiments. In addition, the electronic device 20 in this embodiment may specifically be an electronic computer.

[0084] In this embodiment, the power supply 23 is used to provide operating voltages for the various hardware devices on the electronic device 20; the communication interface 24 can create a data transmission channel between the electronic device 20 and external devices, and the communication protocol it follows can be any communication protocol applicable to the technical solution of the present application, and specific limitations are not imposed here; the input / output interface 25 is used to obtain external input data or output data to the outside, and the specific interface type can be selected according to specific application requirements, and specific limitations are not imposed here.

[0085] In addition, as a carrier for resource storage, the memory 22 may be a read-only memory, a random access memory, a magnetic disk, or an optical disc, etc., and the resources stored thereon may include an operating system 221, a computer program 222, etc., and the storage method may be temporary storage or permanent storage.

[0086] Among them, the operating system 221 is used to manage and control the various hardware devices and the computer program 222 on the electronic device 20, and it may be Windows Server, Netware, Unix, Linux, etc. In addition to the computer program that can be used to complete the work order title generation method executed by the electronic device 20 disclosed in any of the foregoing embodiments, the computer program 222 may further include a computer program that can be used to complete other specific tasks.

[0087] Further, the present application also discloses a computer-readable storage medium for storing a computer program; wherein, when the computer program is executed by a processor, the work order title generation method disclosed above is implemented. For the specific steps of this method, reference may be made to the corresponding content disclosed in the foregoing embodiments, and details are not described herein again.

[0088] In the present specification, the various embodiments are described in a progressive manner. Each embodiment focuses on the differences from other embodiments, and the same or similar parts among the various embodiments can be referred to each other. For the device disclosed in the embodiments, since it corresponds to the method disclosed in the embodiments, the description is relatively simple, and the relevant parts can be referred to the description of the method part.

[0089] Those skilled in the art may further realize that the units and algorithm steps of each example described in combination with the embodiments disclosed herein can be implemented by electronic hardware, computer software, or a combination of the two. To clearly illustrate the interchangeability of hardware and software, the composition and steps of each example have been generally described according to functions in the above description. Whether these functions are executed in a hardware or software manner depends on the specific application and design constraints of the technical solution. Those skilled in the art can use different methods to implement the described functions for each specific application, but such implementation should not be considered to exceed the scope of this application.

[0090] The steps of the methods or algorithms described in combination with the embodiments disclosed herein can be directly implemented by hardware, software modules executed by a processor, or a combination of the two. The software modules can be placed in a random access memory (RAM), internal memory, read-only memory (ROM), electrically programmable ROM, electrically erasable programmable ROM, registers, hard disk, removable disk, CD-ROM, or any other form of storage medium known in the technical field.

[0091] Finally, it should also be noted that in this document, relational terms such as "first" and "second" are only used to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply any actual relationship or order between these entities or operations. Moreover, the term "comprising", "including" or any other variant thereof is intended to cover non-exclusive inclusion, so that a process, method, article or device comprising a series of elements not only includes those elements, but also includes other elements not expressly listed, or elements inherent to such process, method, article or device. Without further limitation, an element defined by the statement "comprising a..." does not exclude the existence of additional identical elements in the process, method, article or device comprising the element.

[0092] The technical solutions provided in this application have been introduced in detail above. Specific examples have been used herein to elaborate on the principles and implementation manners of this application. The description of the above embodiments is only used to help understand the method and its core idea of this application; at the same time, for those of ordinary skill in the art, according to the idea of this application, there will be changes in the specific implementation manners and application scopes. In summary, the content of this specification should not be construed as a limitation to this application.

Claims

1. A method for generating a work order title, characterized in that: Applied to the server, including: Obtaining parsed data corresponding to the target work order sent by the client, and obtaining an initial model input sequence based on the parsed data; Segmenting the initial model input sequence according to the feature types in the initial model input sequence to obtain target fields corresponding to the feature types, and connecting the target fields based on the importance of the target fields to obtain a target model input sequence and position codes corresponding to the target fields; Inputting the target model input sequence into a preset work order title generation model to obtain a feature type code sequence, and generating a work order title corresponding to the target work order based on the target model input sequence, the feature type code sequence and the position code; The work order title is sent to the client, so that the client receives the work order title and displays the work order title.

2. The work order title generation method according to claim 1, characterized in that: The step of obtaining an initial model input sequence based on the parsed data comprises: The parsed data is cleaned to obtain corresponding cleaned data, and the cleaned data is converted into a data format to obtain the initial model input sequence corresponding to the parsed data.

3. The work order title generation method according to claim 1, characterized in that: The connecting the target fields based on the importance of the target fields comprises: A preset separation mark is added at the end of each target field to obtain a post-mark field corresponding to each target field, and each post-mark field is connected based on the importance of each target field.

4. The work order title generation method according to claim 1, characterized in that: The step of inputting the target model input sequence into a preset work order title generation model to obtain a feature type coding sequence includes: Inputting the target model input sequence into the preset work order title generation model, and using the preset work order title generation model to generate an identifier corresponding to each of the feature types in the target model input sequence; Each of the identifiers is converted into a target dense vector, and the feature type encoding sequence is obtained based on each of the target dense vectors.

5. The work order title generation method according to claim 1, characterized in that: The generating a work order title corresponding to the target work order based on the target model input sequence, the feature type code sequence, and the position code comprises: Obtaining a first target vector corresponding to the target model input sequence and a second target vector corresponding to each of the feature types in the feature type encoding sequence, and generating an initial embedding vector according to the first target vector, the second target vector and the position encoding; The initial embedding vector is iterated using the preset work order title generation model to obtain a corresponding target embedding vector, and the work order title corresponding to the target work order is obtained according to the target embedding vector.

6. A method for generating a work order title, characterized in that: Applied to the client, including: Obtaining work order information corresponding to the target work order, and parsing the work order information to obtain parsed data corresponding to the target work order; The parsed data is sent to the server so that the server can obtain the parsed data, an initial model input sequence is obtained based on the parsed data, the initial model input sequence is segmented according to the feature type in the initial model input sequence to obtain the target field corresponding to each feature type, and each target field is connected based on the importance of each target field to obtain the target model input sequence and the position code corresponding to each target field, and finally a work order title corresponding to the target work order is generated based on the target model input sequence, the feature type code sequence and the position code, and the work order title is sent to the client; wherein the feature type code sequence is a sequence generated by a preset work order title generation model using the target model input sequence; Receive the work order title sent by the server, and display the work order title.

7. The work order title generation method according to claim 6, characterized in that: After displaying the work order title, the method further includes: A preset feedback acquisition mechanism is started, and feedback information of a target user on the work order title is acquired by using the preset feedback acquisition mechanism, and the feedback information is saved in a local preset storage space.

8. A work order title generating device, characterized in that: Applied to the server, including: A data acquisition module, used to acquire parsed data corresponding to the target work order sent by the client, and acquire an initial model input sequence based on the parsed data; A sequence segmentation module, used for segmenting the initial model input sequence according to the feature types in the initial model input sequence to obtain target fields corresponding to each feature type, and connecting each target field based on the importance of each target field to obtain a target model input sequence and a position code corresponding to each target field; A work order title generation module, used to input the target model input sequence into a preset work order title generation model to obtain a feature type code sequence, and generate a work order title corresponding to the target work order based on the target model input sequence, the feature type code sequence and the position code; The work order title sending module is used to send the work order title to the client so that the client receives the work order title and displays the work order title.

9. An electronic device, characterized in that: include: Memory, used to store computer programs; A processor, used to execute the computer program to implement the work order title generation method as described in any one of claims 1 to 7.

10. A computer-readable storage medium, characterized in that: Used to store a computer program, which, when executed by a processor, implements the work order title generation method as described in any one of claims 1 to 7.

Images