Power emergency unstructured document directory construction method and device, computer equipment, storage medium and computer program product
The pre-trained title information prediction model automatically constructs the unstructured document catalog of power emergency, which solves the problem of inefficiency in traditional methods, realizes fast and efficient directory construction, and reduces the need for manual intervention.
Patent Information
- Application Number
- CN202411001624.5
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2024-07-25
- Publication Date
- 2024-12-03
AI Technical Summary
In the field of power grids, traditional methods are inefficient in catalog building of unstructured documents for power emergency, requiring a lot of manual intervention, resulting in low construction efficiency.
The pre-trained title information prediction model is adopted to obtain unstructured documents of pending power emergency, extract keywords, determine chapter title information and hierarchical relationships, and automatically build a catalog to avoid manual intervention.
It improves the efficiency of building unstructured document catalogs for power emergency, realizes rapid automated construction, and reduces time and labor costs.
Smart Images

Figure CN119066141A_ABST
Abstract
Description
Technical Field
[0001] The present application relates to the field of power grid technology, and in particular to a method, device, computer equipment, computer-readable storage medium and computer program product for constructing a power emergency unstructured document directory. Background Art
[0002] In the field of power grids, cataloging of unstructured documents related to power emergencies is crucial for quickly obtaining key information in the documents.
[0003] In traditional technology, manual construction is generally used to construct a directory for power emergency unstructured documents; however, this method is likely to consume a lot of time and manpower, resulting in low efficiency in constructing the power emergency unstructured document directory. Summary of the invention
[0004] Based on this, it is necessary to provide a method, device, computer equipment, computer-readable storage medium and computer program product for constructing an electric power emergency unstructured document directory that can improve the efficiency of constructing an electric power emergency unstructured document directory in response to the above technical problems.
[0005] In a first aspect, the present application provides a method for constructing a power emergency unstructured document directory, comprising:
[0006] Obtain unstructured documents of power emergencies to be processed;
[0007] Inputting the pending power emergency unstructured document and keywords in the pending power emergency unstructured document into a pre-trained title information prediction model to obtain title information of each chapter of the pending power emergency unstructured document;
[0008] Determine the target word vector corresponding to the title information of each chapter according to the title information of each chapter and the text information under the title information of each chapter;
[0009] Determining the hierarchical relationship between the title information of each chapter according to the target word vector corresponding to the title information of each chapter;
[0010] According to the title information of each chapter and the hierarchical relationship between the title information of each chapter, the directory information corresponding to the unstructured document of power emergency to be processed is constructed.
[0011] In one embodiment, determining the target word vector corresponding to the title information of each chapter according to the title information of each chapter and the text information under the title information of each chapter includes:
[0012] Performing feature extraction processing on the title information of each chapter to obtain a first word vector corresponding to the title information of each chapter, and performing feature extraction processing on the text information under the title information of each chapter to obtain a second word vector corresponding to the text information under the title information of each chapter;
[0013] Determine a first weight corresponding to the first word vector and a second weight corresponding to the second word vector;
[0014] The first word vector and the second word vector are fused according to the first weight and the second weight to obtain a target word vector corresponding to the title information of each chapter.
[0015] In one embodiment, determining the hierarchical relationship between the title information of each chapter according to the target word vector corresponding to the title information of each chapter includes:
[0016] Determining the similarity between the title information of each chapter according to the target word vector corresponding to the title information of each chapter;
[0017] According to the similarities between the title information of the chapters, the hierarchical relationship between the title information of the chapters is determined.
[0018] In one embodiment, after constructing the directory information corresponding to the unstructured document for power emergency to be processed according to the title information of each chapter and the hierarchical relationship between the title information of each chapter, the method further includes:
[0019] Verifying the directory information to obtain a verification result corresponding to the unstructured power emergency document to be processed;
[0020] When the verification result meets the preset conditions, the directory information is used as the target directory information corresponding to the unstructured document for power emergency to be processed.
[0021] In one embodiment, after obtaining the unstructured document of power emergency to be processed, the method further includes:
[0022] Preprocessing the unstructured power emergency document to be processed to obtain a preprocessed unstructured power emergency document;
[0023] The step of inputting the unstructured document of power emergency to be processed and the keywords in the unstructured document of power emergency to be processed into a pre-trained title information prediction model to obtain the title information of each chapter of the unstructured document of power emergency to be processed includes:
[0024] The pre-processed electric power emergency unstructured document and keywords in the pre-processed electric power emergency unstructured document are input into a pre-trained title information prediction model to obtain title information of each chapter of the unstructured electric power emergency document to be processed.
[0025] In one embodiment, the pre-trained title information prediction model is trained in the following manner:
[0026] Get sample power emergency unstructured documents;
[0027] Inputting the sample power emergency unstructured document and keywords in the sample power emergency unstructured document into the title information prediction model to be trained to obtain predicted title information of each chapter of the sample power emergency unstructured document;
[0028] According to the difference between the predicted title information and the actual title information of each chapter of the sample power emergency unstructured document, the title information prediction model to be trained is iteratively trained to obtain the pre-trained title information prediction model.
[0029] In a second aspect, the present application also provides a device for constructing a power emergency unstructured document directory, comprising:
[0030] A document acquisition module is used to acquire unstructured documents of power emergency to be processed;
[0031] An information prediction module, used for inputting the pending power emergency unstructured document and keywords in the pending power emergency unstructured document into a pre-trained title information prediction model to obtain title information of each chapter of the pending power emergency unstructured document;
[0032] A vector determination module, used to determine the target word vector corresponding to the title information of each chapter according to the title information of each chapter and the text information under the title information of each chapter;
[0033] A relationship determination module, used to determine the hierarchical relationship between the title information of each chapter according to the target word vector corresponding to the title information of each chapter;
[0034] The directory construction module is used to construct the directory information corresponding to the unstructured document of power emergency to be processed according to the title information of each chapter and the hierarchical relationship between the title information of each chapter.
[0035] In a third aspect, the present application further provides a computer device, including a memory and a processor, wherein the memory stores a computer program, and when the processor executes the computer program, the following steps are implemented:
[0036] Obtain unstructured documents of power emergencies to be processed;
[0037] Inputting the pending power emergency unstructured document and keywords in the pending power emergency unstructured document into a pre-trained title information prediction model to obtain title information of each chapter of the pending power emergency unstructured document;
[0038] Determine the target word vector corresponding to the title information of each chapter according to the title information of each chapter and the text information under the title information of each chapter;
[0039] Determining the hierarchical relationship between the title information of each chapter according to the target word vector corresponding to the title information of each chapter;
[0040] According to the title information of each chapter and the hierarchical relationship between the title information of each chapter, the directory information corresponding to the unstructured document of power emergency to be processed is constructed.
[0041] In a fourth aspect, the present application further provides a computer-readable storage medium having a computer program stored thereon, wherein when the computer program is executed by a processor, the following steps are implemented:
[0042] Obtain unstructured documents of power emergencies to be processed;
[0043] Inputting the pending power emergency unstructured document and keywords in the pending power emergency unstructured document into a pre-trained title information prediction model to obtain title information of each chapter of the pending power emergency unstructured document;
[0044] Determine the target word vector corresponding to the title information of each chapter according to the title information of each chapter and the text information under the title information of each chapter;
[0045] Determining the hierarchical relationship between the title information of each chapter according to the target word vector corresponding to the title information of each chapter;
[0046] According to the title information of each chapter and the hierarchical relationship between the title information of each chapter, the directory information corresponding to the unstructured document of power emergency to be processed is constructed.
[0047] In a fifth aspect, the present application further provides a computer program product, including a computer program, which implements the following steps when executed by a processor:
[0048] Obtain unstructured documents of power emergencies to be processed;
[0049] Inputting the pending power emergency unstructured document and keywords in the pending power emergency unstructured document into a pre-trained title information prediction model to obtain title information of each chapter of the pending power emergency unstructured document;
[0050] Determine the target word vector corresponding to the title information of each chapter according to the title information of each chapter and the text information under the title information of each chapter;
[0051] Determining the hierarchical relationship between the title information of each chapter according to the target word vector corresponding to the title information of each chapter;
[0052] According to the title information of each chapter and the hierarchical relationship between the title information of each chapter, the directory information corresponding to the unstructured document of power emergency to be processed is constructed.
[0053] The above-mentioned method, device, computer equipment, storage medium and computer program product for constructing a directory of electric power emergency unstructured documents first obtain the unstructured electric power emergency documents to be processed, and then input the unstructured electric power emergency documents to be processed and the keywords in the unstructured electric power emergency documents to be processed into a pre-trained title information prediction model to obtain the title information of each chapter of the unstructured electric power emergency documents to be processed, and then, according to the title information of each chapter and the text information under the title information of each chapter, determine the target word vector corresponding to the title information of each chapter, and then, according to the target word vector corresponding to the title information of each chapter, determine the hierarchical relationship between the title information of each chapter, and finally, according to the title information of each chapter and the hierarchical relationship between the title information of each chapter, construct the directory information corresponding to the unstructured electric power emergency documents to be processed. In this way, when constructing a directory for power emergency unstructured documents, the title information of each chapter of the power emergency unstructured documents to be processed can be quickly obtained through the pre-trained title information prediction model, and the hierarchical relationship between the title information of each chapter can be combined to quickly construct the directory information corresponding to the power emergency unstructured documents to be processed, which is conducive to improving the construction efficiency of the power emergency unstructured document directory; moreover, the entire process adopts an automated method without manual intervention, avoiding the defect of manual construction that easily consumes a lot of time and manpower, resulting in low construction efficiency of the power emergency unstructured document directory, thereby improving the construction efficiency of the power emergency unstructured document directory. BRIEF DESCRIPTION OF THE DRAWINGS
[0054] In order to more clearly illustrate the technical solutions in the embodiments of the present application or related technologies, the drawings required for use in the embodiments of the present application or related technical descriptions will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present application. For ordinary technicians in this field, other related drawings can be obtained based on these drawings without paying creative work.
[0055] Figure 1 A flowchart of a method for constructing a power emergency unstructured document directory in one embodiment;
[0056] Figure 2 A flowchart of the steps of determining the target word vector corresponding to the title information of each chapter in one embodiment;
[0057] Figure 3 A flowchart of the steps of determining the hierarchical relationship between the title information of each chapter in one embodiment;
[0058] Figure 4 A flowchart of a method for constructing a power emergency unstructured document directory in another embodiment;
[0059] Figure 5 A structural block diagram of a device for constructing a power emergency unstructured document directory in one embodiment;
[0060] Figure 6 FIG. 4 is a diagram showing the internal structure of a computer device in one embodiment. DETAILED DESCRIPTION
[0061] In order to make the purpose, technical solution and advantages of the present application more clearly understood, the present application is further described in detail below in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present application and are not used to limit the present application.
[0062] It should be noted that the user information (including but not limited to user device information, user personal information, etc.) and data (including but not limited to data used for analysis, stored data, displayed data, etc.) involved in this application are all information and data authorized by the user or fully authorized by all parties, and the collection, use and processing of relevant data must comply with relevant regulations.
[0063] In an exemplary embodiment, Figure 1As shown, a method for constructing an unstructured document directory for power emergency is provided. This embodiment uses the method applied to a server as an example for illustration; it can be understood that the method can also be applied to a terminal, and can also be applied to a system including a terminal and a server, and is implemented through the interaction between the terminal and the server. Among them, the terminal can be but is not limited to various personal computers, laptops, smart phones and tablets; the server can be implemented with an independent server or a server cluster composed of multiple servers. In this embodiment, the method includes the following steps:
[0064] Step S101, obtaining the unstructured power emergency document to be processed.
[0065] The unprocessed power emergency unstructured documents refer to the power emergency unstructured documents that need to be processed.
[0066] Among them, power emergency unstructured documents are used to represent data documents without clear structure generated in the power emergency field, such as emergency plans, accident case analysis reports, emergency rescue operation manuals, etc.
[0067] Exemplarily, in response to the directory construction instruction, the server selects a candidate power emergency unstructured document corresponding to the directory construction instruction from a plurality of candidate power emergency unstructured documents as the power emergency unstructured document to be processed.
[0068] Step S102: input the unstructured document of power emergency to be processed and the keywords in the unstructured document of power emergency to be processed into a pre-trained title information prediction model to obtain the title information of each chapter of the unstructured document of power emergency to be processed.
[0069] Among them, keywords refer to words with high importance extracted from the unstructured documents of power emergency to be processed, such as "power", "emergency", and "plan". For example, this application uses words with an importance greater than a preset importance in the unstructured documents of power emergency to be processed as keywords.
[0070] Among them, the title information prediction model refers to a network model that can use the pending power emergency unstructured documents and keywords in the pending power emergency unstructured documents to obtain the title information of each chapter of the pending power emergency unstructured documents, such as the LSTM (Long Short-Term Memory) model and the Bi-RNN (Bidirectional Recurrent Neural Network) model.
[0071] The title information of each chapter is used to represent the title corresponding to each chapter in the unstructured document of power emergency to be processed, such as "Power System Overview" and "Grid Structure".
[0072] Exemplarily, the server performs word segmentation processing on the unstructured document for electric power emergency to be processed, and obtains multiple words in the unstructured document for electric power emergency to be processed; then, the server selects words that meet preset conditions from the multiple words in the unstructured document for electric power emergency to be processed, and uses them as keywords in the unstructured document for electric power emergency to be processed; for example, the server determines the importance corresponding to the multiple words in the unstructured document for electric power emergency to be processed, and selects words whose corresponding importance is greater than the preset importance from the multiple words in the unstructured document for electric power emergency to be processed, and uses them as keywords in the unstructured document for electric power emergency to be processed; then, the server inputs the unstructured document for electric power emergency to be processed and the keywords in the unstructured document for electric power emergency to be processed into a pre-trained title information prediction model, and performs title information prediction through the pre-trained title information prediction model to obtain title information corresponding to the unstructured document for electric power emergency to be processed and the keywords, and uses them as title information of each chapter of the unstructured document for electric power emergency to be processed.
[0073] Step S103, determining the target word vector corresponding to the title information of each chapter according to the title information of each chapter and the text information under the title information of each chapter.
[0074] The text information refers to the specific content corresponding to the title information of each chapter.
[0075] Among them, the target word vector refers to the comprehensive representation vector corresponding to the title information of each chapter.
[0076] Exemplarily, the server determines the word vector corresponding to the title information of each chapter as the first word vector, and the word vector corresponding to the text information under the title information of each chapter as the second word vector; then, the server fuses the first word vector and the second word vector to obtain the fused word vector as the target word vector corresponding to the title information of each chapter.
[0077] Step S104, determining the hierarchical relationship between the title information of each chapter according to the target word vector corresponding to the title information of each chapter.
[0078] The hierarchical relationship is used to represent the hierarchical division and attribution between the title information of each chapter, such as the subordinate relationship. For example, "Power System Overview" and "Grid Structure" have a subordinate relationship. "Power System Overview" is the upper-level title of "Grid Structure", and "Grid Structure" is the lower-level title of "Power System Overview".
[0079] Exemplarily, the server determines the hierarchical relationship corresponding to the target word vector as the hierarchical relationship between the title information of each chapter based on the target word vector corresponding to the title information of each chapter and the correspondence between the target word vector and the hierarchical relationship.
[0080] Step S105: construct directory information corresponding to the unstructured document of power emergency to be processed according to the title information of each chapter and the hierarchical relationship between the title information of each chapter.
[0081] The directory information is used to characterize the document directory tree structure corresponding to the unstructured documents of power emergency to be processed.
[0082] Exemplarily, the server determines the hierarchical identifier corresponding to the title information of each chapter based on the title information of each chapter and the hierarchical relationship between the title information of each chapter (for example, the hierarchical identifier corresponding to "Overview of Power System" is 1.1, the hierarchical identifier corresponding to "Grid Structure" is 1.1.1, etc.); then, the server determines the page number information corresponding to the title information of each chapter (for example, the page number corresponding to "Overview of Power System" is page 2, and the page number corresponding to "Grid Structure" is page 6); then, the server constructs the directory information corresponding to the unstructured document for power emergency to be processed based on the title information of each chapter, the hierarchical relationship between the title information of each chapter, the hierarchical identifier and page number information corresponding to the title information of each chapter.
[0083] In the above-mentioned method for constructing a directory of power emergency unstructured documents, the power emergency unstructured documents to be processed are first obtained, and then the power emergency unstructured documents to be processed and the keywords in the power emergency unstructured documents to be processed are input into a pre-trained title information prediction model to obtain the title information of each chapter of the power emergency unstructured documents to be processed. Then, according to the title information of each chapter and the text information under the title information of each chapter, the target word vector corresponding to the title information of each chapter is determined. Then, according to the target word vector corresponding to the title information of each chapter, the hierarchical relationship between the title information of each chapter is determined. Finally, according to the title information of each chapter and the hierarchical relationship between the title information of each chapter, the directory information corresponding to the power emergency unstructured documents to be processed is constructed. In this way, when constructing a directory for power emergency unstructured documents, the title information of each chapter of the power emergency unstructured documents to be processed can be quickly obtained through the pre-trained title information prediction model, and the hierarchical relationship between the title information of each chapter can be combined to quickly construct the directory information corresponding to the power emergency unstructured documents to be processed, which is conducive to improving the construction efficiency of the power emergency unstructured document directory; moreover, the entire process adopts an automated method without manual intervention, avoiding the defect of manual construction that easily consumes a lot of time and manpower, resulting in low construction efficiency of the power emergency unstructured document directory, thereby improving the construction efficiency of the power emergency unstructured document directory.
[0084] In an exemplary embodiment, Figure 2 As shown, the above step S103, according to the title information of each chapter and the text information under the title information of each chapter, determines the target word vector corresponding to the title information of each chapter, and specifically includes the following steps:
[0085] Step S201, perform feature extraction processing on the title information of each chapter to obtain the first word vector corresponding to the title information of each chapter, and perform feature extraction processing on the text information under the title information of each chapter to obtain the second word vector corresponding to the text information under the title information of each chapter.
[0086] Step S202, determining a first weight corresponding to the first word vector and a second weight corresponding to the second word vector.
[0087] Step S203: According to the first weight and the second weight, the first word vector and the second word vector are fused to obtain the target word vector corresponding to the title information of each chapter.
[0088] Among them, feature extraction processing refers to the process of extracting the word vector corresponding to the information.
[0089] Among them, the first word vector refers to the word vector corresponding to the title information of each chapter.
[0090] The second word vector refers to the word vector corresponding to the text information under the title information of each chapter.
[0091] Among them, the first weight refers to the weight corresponding to the first word vector.
[0092] Among them, the second weight refers to the weight corresponding to the second word vector.
[0093] The fusion processing may refer to weighted summation processing.
[0094] Exemplarily, the server inputs the title information of each chapter into a pre-trained word vector model, performs feature extraction processing on the title information of each chapter through the pre-trained word vector model, and obtains the word vector corresponding to the title information of each chapter as the first word vector; then, the server inputs the text information under the title information of each chapter into the pre-trained word vector model, performs feature extraction processing on the text information under the title information of each chapter through the pre-trained word vector model, and obtains the word vector corresponding to the text information under the title information of each chapter, the second word vector; then, the server inputs the first word vector and the second word vector into the attention model respectively, and determines the first weight corresponding to the first word vector and the second weight corresponding to the second word vector through the attention model; then, the server fuses the first word vector and the second word vector according to the first weight and the second weight to obtain the fused word vector as the target word vector corresponding to the title information of each chapter; for example, the server performs weighted sum processing on the first word vector and the second word vector according to the first weight and the second weight to obtain the fused word vector as the target word vector corresponding to the title information of each chapter.
[0095] It should be noted that the pre-trained word vector models include the BERT (Bidirectional Encoder Representations from Transformers) model and the ELMo (Embeddings from Language Models) model.
[0096] In this embodiment, by extracting features from the title information of each chapter and the text information under the title information of each chapter respectively, and merging them after assigning different weights, the semantic features of each chapter can be represented more comprehensively and accurately, thereby improving the accuracy of determining the target word vector corresponding to the title information of each chapter.
[0097] In an exemplary embodiment, Figure 3 As shown, the above step S104, according to the target word vector corresponding to the title information of each chapter, determines the hierarchical relationship between the title information of each chapter, and specifically includes the following steps:
[0098] Step S301, determining the similarity between the title information of each chapter according to the target word vector corresponding to the title information of each chapter.
[0099] Step S302: determining the hierarchical relationship between the title information of each chapter according to the similarity between the title information of each chapter.
[0100] Herein, the similarity may refer to cosine similarity.
[0101] Exemplarily, the server inputs the target word vector corresponding to the title information of each chapter into the similarity prediction model, and determines the similarity between the title information of each chapter through the similarity prediction model; then, the server determines the hierarchical relationship corresponding to the similarity according to the similarity between the title information of each chapter, as the hierarchical relationship between the title information of each chapter; for example, the server uses the similarity threshold to judge the similarity between the title information of each two chapters; when the similarity between the title information of the two chapters is greater than the similarity threshold, it is determined that the hierarchical relationship between the title information of the two chapters is a subordinate relationship; when the similarity between the title information of the two chapters is less than or equal to the similarity threshold, it is determined that the hierarchical relationship between the title information of the two chapters is not a subordinate relationship.
[0102] In this embodiment, the hierarchical relationship is clarified by determining the similarity between the title information, which can more accurately reflect the logical structure of the document content, avoid the subjectivity and uncertainty of manual judgment, and lead to the defect of low accuracy in determining the hierarchical relationship between the title information of each chapter, thereby improving the accuracy of determining the hierarchical relationship between the title information of each chapter.
[0103] In an exemplary embodiment, the above step S105, after constructing the directory information corresponding to the unstructured document for power emergency to be processed according to the title information of each chapter and the hierarchical relationship between the title information of each chapter, specifically includes the following contents: verifying the directory information to obtain the verification result corresponding to the unstructured document for power emergency to be processed; when the verification result meets the preset conditions, using the directory information as the target directory information corresponding to the unstructured document for power emergency to be processed.
[0104] The verification result is used to characterize the verification status of the unstructured document of power emergency to be processed.
[0105] The preset conditions refer to pre-set judgment conditions used to judge the verification results, such as whether the titles at the same level in the directory information follow similar naming rules, whether there are missing or redundant levels in the directory information, etc. It should be noted that the preset conditions may be determined according to the circumstances.
[0106] The target directory information refers to the directory information whose corresponding verification results meet the preset conditions.
[0107] Exemplarily, the server verifies the directory information and obtains a verification result of the directory information as a verification result corresponding to the unstructured document for power emergency to be processed; when the verification result meets the preset conditions, the server uses the directory information as the target directory information corresponding to the unstructured document for power emergency to be processed; when the verification result does not meet the preset conditions, the server adjusts the directory information (such as eliminating incorrectly identified items in the directory information) and obtains the adjusted directory information as the target directory information corresponding to the unstructured document for power emergency to be processed.
[0108] In this embodiment, by verifying the directory information, it is possible to eliminate erroneous or unreasonable information that may exist in the directory information, which is conducive to ensuring that the directory information can accurately reflect the document content of the unstructured power emergency documents to be processed, thereby improving the quality and availability of the unstructured power emergency documents to be processed.
[0109] In an exemplary embodiment, the above step S101, after obtaining the unstructured power emergency document to be processed, specifically includes the following contents: preprocessing the unstructured power emergency document to be processed to obtain the preprocessed unstructured power emergency document.
[0110] Then, the keywords in the unstructured document for power emergency to be processed and the keywords in the unstructured document for power emergency to be processed are input into the pre-trained title information prediction model to obtain the title information of each chapter of the unstructured document for power emergency to be processed, which specifically includes the following contents: the keywords in the preprocessed unstructured document for power emergency to be processed and the keywords in the unstructured document for power emergency to be processed are input into the pre-trained title information prediction model to obtain the title information of each chapter of the unstructured document for power emergency to be processed.
[0111] Among them, preprocessing refers to pre-processing the unstructured documents of power emergency to be processed, such as removing irrelevant characters, punctuation marks and noise information in the unstructured documents of power emergency to be processed.
[0112] The pre-processed power emergency unstructured document refers to the unstructured power emergency document to be processed after pre-processing.
[0113] Exemplarily, the server identifies abnormal information (such as irrelevant characters, punctuation marks and noise information) in the unstructured document for electric power emergency to be processed, and deletes the abnormal information in the unstructured document for electric power emergency to be processed, so as to obtain the processed unstructured document for electric power emergency to be processed as the preprocessed unstructured document for electric power emergency; then, the server inputs the preprocessed unstructured document for electric power emergency and the keywords in the unstructured document for electric power emergency to a pre-trained title information prediction model, and performs title information prediction through the pre-trained title information prediction model, so as to obtain the title information corresponding to the preprocessed unstructured document for electric power emergency and the keywords as the title information of each chapter of the unstructured document for electric power emergency to be processed.
[0114] In this embodiment, by preprocessing the unstructured power emergency documents to be processed, noise, invalid information or format errors in the unstructured power emergency documents to be processed can be removed, so that subsequent processing and analysis are based on more accurate and valuable data, which is conducive to reducing the complexity and time consumption in the subsequent processing process.
[0115] In an exemplary embodiment, the method for constructing a power emergency unstructured document directory provided in the present application also includes a training step of a pre-trained title information prediction model, which specifically includes the following contents: obtaining a sample power emergency unstructured document; inputting the sample power emergency unstructured document and keywords in the sample power emergency unstructured document into the title information prediction model to be trained to obtain the predicted title information of each chapter of the sample power emergency unstructured document; iteratively training the title information prediction model to be trained according to the difference between the predicted title information and the actual title information of each chapter of the sample power emergency unstructured document to obtain a pre-trained title information prediction model.
[0116] The sample power emergency unstructured document refers to the power emergency unstructured document corresponding to the title information prediction model used for training.
[0117] The predicted title information refers to the predicted value corresponding to the title information.
[0118] The actual title information refers to the actual value corresponding to the title information.
[0119] Exemplarily, in response to a model training instruction for a title information prediction model to be trained, the server obtains a sample power emergency unstructured document from a database; then, the server identifies keywords in the sample power emergency unstructured document, and inputs the sample power emergency unstructured document and the keywords in the sample power emergency unstructured document into the title information prediction model to be trained to obtain predicted title information of each chapter of the sample power emergency unstructured document; then, the server obtains a loss value based on the difference between the predicted title information and the actual title information of each chapter of the sample power emergency unstructured document; then, the server adjusts the model parameters of the title information prediction model to be trained based on the loss value; then, the server re-trains the title information prediction model after the model parameters are adjusted until the loss value obtained by the trained title information prediction model is less than the loss value threshold, then stops the training, and uses the trained title information prediction model as the pre-trained title information prediction model.
[0120] In this embodiment, by pre-training the title information prediction model, it is convenient in practical applications to predict the title information of each chapter of the unstructured document for power emergency to be processed after obtaining the unstructured document for power emergency to be processed; moreover, the title information prediction model receives new data in each round of iteration, and performs internal improvements and optimizations on the model, so as to be able to make predictions more effectively, which is beneficial to improving the prediction accuracy of the title information prediction model.
[0121] In an exemplary embodiment, Figure 4 As shown, another method for constructing a power emergency unstructured document directory is provided, and the method is applied to a server for example, including the following steps:
[0122] Step S401, obtaining the unstructured power emergency document to be processed.
[0123] Step S402, preprocessing the unstructured power emergency document to be processed to obtain a preprocessed unstructured power emergency document.
[0124] Step S403, input the preprocessed electric power emergency unstructured document and the keywords in the preprocessed electric power emergency unstructured document into a pre-trained title information prediction model to obtain the title information of each chapter of the to-be-processed electric power emergency unstructured document.
[0125] Step S404, perform feature extraction processing on the title information of each chapter to obtain the first word vector corresponding to the title information of each chapter, and perform feature extraction processing on the text information under the title information of each chapter to obtain the second word vector corresponding to the text information under the title information of each chapter.
[0126] Step S405, determine the first weight corresponding to the first word vector and the second weight corresponding to the second word vector; according to the first weight and the second weight, fuse the first word vector and the second word vector to obtain the target word vector corresponding to the title information of each chapter.
[0127] Step S406, determining the similarity between the title information of each chapter based on the target word vector corresponding to the title information of each chapter; and determining the hierarchical relationship between the title information of each chapter based on the similarity between the title information of each chapter.
[0128] Step S407: construct directory information corresponding to the unstructured document of power emergency to be processed according to the title information of each chapter and the hierarchical relationship between the title information of each chapter.
[0129] Step S408, verifying the directory information to obtain a verification result corresponding to the unstructured document of the power emergency to be processed; when the verification result meets a preset condition, the directory information is used as the target directory information corresponding to the unstructured document of the power emergency to be processed.
[0130] In the above-mentioned method for constructing a directory of electric power emergency unstructured documents, when constructing a directory for electric power emergency unstructured documents, the title information of each chapter of the to-be-processed electric power emergency unstructured documents can be quickly obtained through a pre-trained title information prediction model, and the hierarchical relationship between the title information of each chapter can be combined to quickly construct the directory information corresponding to the to-be-processed electric power emergency unstructured documents, which is conducive to improving the construction efficiency of the electric power emergency unstructured document directory; moreover, the entire process adopts an automated method without the need for human intervention, thus avoiding the defect that the manual construction method is prone to consuming a lot of time and manpower, resulting in low construction efficiency of the electric power emergency unstructured document directory, thereby improving the construction efficiency of the electric power emergency unstructured document directory.
[0131] In an exemplary embodiment, in order to more clearly illustrate the method for constructing a directory of electric emergency unstructured documents provided in the embodiment of the present application, the method for constructing a directory of electric emergency unstructured documents is specifically described below with a specific embodiment. In one embodiment, the present application also provides an intelligent extraction method based on the directory of electric emergency unstructured documents. When constructing a directory of electric emergency unstructured documents, the electric emergency unstructured documents to be processed are first obtained, and then the electric emergency unstructured documents to be processed and the keywords in the electric emergency unstructured documents to be processed are input into the pre-trained title information prediction model to obtain the title information of each chapter of the electric emergency unstructured documents to be processed. Then, according to the title information of each chapter and the text information under the title information of each chapter, the target word vector corresponding to the title information of each chapter is determined. Then, according to the target word vector corresponding to the title information of each chapter, the hierarchical relationship between the title information of each chapter is determined. Finally, according to the title information of each chapter and the hierarchical relationship between the title information of each chapter, the directory information corresponding to the electric emergency unstructured documents to be processed is constructed. Specifically including the following contents:
[0132] This application proposes an intelligent extraction method for unstructured document directories based on power emergency. First, this method uses pre-trained word vector models (such as BERT, ELMo) to convert words in the document directory into vector form. Compared with traditional word vector models, these pre-trained models can better capture the semantic information between words and take into account the context of the context. In addition, based on the characteristics of the power emergency management field, we can combine domain knowledge to build a customized word vector model to improve the performance of the model in a specific field. Then, the system uses a deep learning model to perform named entity recognition in the document directory. We can use pre-trained models or design deep learning networks by ourselves to identify entities in the document directory, such as places, times, event names, etc., through sequence labeling methods. In recent years, named entity recognition technology based on deep learning has made significant progress. The BiLSTM-CRF (Bidirectional Long Short-Term Memory-Conditional Random Field) model structure can effectively capture contextual information between entities and improve recognition accuracy.
[0133] In the classification stage of the document directory, this method introduces a deep learning model with an attention mechanism. The attention mechanism can dynamically adjust the model's attention to the text sequence according to the importance of different words, so that the model can better capture key information. At the same time, we can combine the latest self-attention models such as Transformer to achieve better text representation and classification effects. In addition, we can also use the transfer learning method to use the pre-trained model parameters to learn a small number of samples of power emergency documents to improve the generalization ability of the model. It mainly includes the following steps:
[0134] 1. Preprocessing stage:
[0135] Clean the unstructured documents of power emergency and remove irrelevant characters, punctuation marks and noise information;
[0136] Using natural language processing technology for word segmentation and syntactic analysis, in power emergency documents, word segmentation can help identify key terms such as "power", "emergency", and "plan", which are often the basic elements of the document directory. Through word segmentation, keywords and common words in the document can be distinguished. Combined with syntactic analysis, the relationship between words, such as subject-predicate-object structure, modification relationship, etc., can be determined. This is crucial for understanding the internal logic and hierarchical structure of the document. In power emergency documents, syntactic analysis can help identify whether a word is a key node that describes the power system, emergency response process, or specific operation steps. By analyzing the sentence structure, it can be inferred that the outline or sub-chapter of the document is very helpful for extracting the document directory. It lays the foundation for subsequent directory extraction. At the same time, syntactic analysis can reveal the internal logical structure of the document, help identify the relationship between different paragraphs or chapters, and thus construct the hierarchical structure of the document. For example, syntactic analysis can identify the superior-subordinate relationship between "1.1 Power System Overview" and "1.1.1 Grid Structure", which is essential for building a directory tree structure.
[0137] 2. Document structure identification stage:
[0138] Use deep learning models, such as long short-term memory networks (LSTM) or bidirectional recurrent neural networks (Bi-RNN), combined with pre-trained word segmentation and syntactic analysis result embedding models, such as BERT or GPT (Generative Pre-trained Transformer), to train on the characteristics of power emergency documents and identify chapter titles and other key information with directory attributes (such as section titles, subtitles, keywords, and other structural and general content) in the documents.
[0139] 3. Directory information extraction stage:
[0140] Based on the identified chapter titles, a title and hierarchical relationship model is established in a supervised or semi-supervised manner. This model aims to identify and understand the hierarchical relationship of titles in the document, so as to accurately reflect the structure of the document; it can clearly give a top-down document directory tree structure. For example, starting from the level 1 title, the level 2 title is its child node, the level 3 title is the child node of the level 2 title, and so on, until the complete directory tree is built.
[0141] Combining the context and text structure information, the document directory tree structure is accurately extracted and constructed through the global optimal path search algorithm or the graph neural network (GNN) algorithm. At this time, a clear and hierarchical directory structure is automatically identified and constructed from the unstructured document text, which also includes the directory and the corresponding text content.
[0142] 4. Post-processing stage:
[0143] Verify the validity of the extracted directory information and eliminate incorrectly identified items. The means of validity verification generally include: (1) Consistency check: Verify whether the naming and hierarchy of the directory items conform to the style and structure of the document. For example, check whether the titles at the same level follow similar naming rules and whether there are missing or redundant levels. (2) Logical check: Ensure that the directory structure is logically clear and there are no hierarchical jumps or repetitions. For example, check whether there is a situation where a level 2 title jumps directly to a level 4 title, or whether the order of titles at the same level is reasonable. (3) Compare with the original document: Compare the extracted directory with the original document one by one to confirm the existence and accuracy of the directory items. You can use manual review or automated tools to compare. (4) Compare with other directory versions: If there are multiple versions of the same document, compare the directory differences between different versions to ensure the consistency of the extracted directory information between different versions.
[0144] The extracted standard directory information is stored in association with the original document. One document corresponds to multiple directory structures, and each directory structure has a hierarchical relationship and sequence. Each directory is associated with the corresponding content, which is convenient for subsequent query and retrieval.
[0145] 5. Application stage:
[0146] The directory information extracted by the above method can be integrated into the power emergency document management system to achieve rapid indexing and precise positioning of document content and improve emergency response efficiency.
[0147] In the above embodiment, when constructing the directory of the power emergency unstructured document, the title information of each chapter of the power emergency unstructured document to be processed can be quickly obtained through the pre-trained title information prediction model, and the hierarchical relationship between the title information of each chapter can be combined to quickly construct the directory information corresponding to the power emergency unstructured document to be processed, which is conducive to improving the construction efficiency of the power emergency unstructured document directory; moreover, the whole process adopts an automated method without manual intervention, avoiding the defect of manual construction that is easy to consume a lot of time and manpower, resulting in low construction efficiency of the power emergency unstructured document directory, thereby improving the construction efficiency of the power emergency unstructured document directory. At the same time, this method uses deep learning and natural language processing technology to realize the automated processing of large-scale unstructured document directories and improve work efficiency; using advanced deep learning models, it can accurately identify key information in the field of power emergency management and provide reliable support for emergency decision-making; this method is compatible with the existing power emergency management system, and can achieve seamless integration with the existing system to improve the overall efficiency of the system. The technical details of this application include key technologies such as pre-trained word vector models, deep learning models, and attention mechanisms. It combines the latest advances in the fields of deep learning and natural language processing. In practical applications, it can be integrated with the power emergency management system to achieve intelligent processing of unstructured document directories. By integrating the extracted information into the existing decision support system, the efficiency and accuracy of power emergency management can be improved, providing a more reliable basis for emergency decision-making. At the same time, this method can also be combined with big data technology to achieve real-time monitoring and analysis of power emergency information, providing more timely support for emergency decision-making. It provides a brand new solution for power emergency management.
[0148] It should be understood that, although the various steps in the flowcharts involved in the above-mentioned embodiments are displayed in sequence according to the indication of the arrows, these steps are not necessarily executed in sequence according to the order indicated by the arrows. Unless there is a clear explanation in this article, the execution of these steps does not have a strict order restriction, and these steps can be executed in other orders. Moreover, at least a part of the steps in the flowcharts involved in the above-mentioned embodiments can include multiple steps or multiple stages, and these steps or stages are not necessarily executed at the same time, but can be executed at different times, and the execution order of these steps or stages is not necessarily to be carried out in sequence, but can be executed in turn or alternately with other steps or at least a part of the steps or stages in other steps.
[0149] Based on the same inventive concept, the embodiment of the present application also provides a device for constructing an unstructured document directory for electric power emergency, which is used to implement the method for constructing an unstructured document directory for electric power emergency mentioned above. The implementation scheme for solving the problem provided by the device is similar to the implementation scheme recorded in the above method, so the specific limitations in one or more embodiments of the device for constructing an unstructured document directory for electric power emergency provided below can refer to the limitations of the method for constructing an unstructured document directory for electric power emergency mentioned above, and will not be repeated here.
[0150] In an exemplary embodiment, Figure 5 As shown, a device for constructing a power emergency unstructured document directory is provided, comprising: a document acquisition module 501, an information prediction module 502, a vector determination module 503, a relationship determination module 504 and a directory construction module 505, wherein:
[0151] The document acquisition module 501 is used to acquire the unstructured power emergency document to be processed.
[0152] The information prediction module 502 is used to input the pending power emergency unstructured document and the keywords in the pending power emergency unstructured document into a pre-trained title information prediction model to obtain the title information of each chapter of the pending power emergency unstructured document.
[0153] The vector determination module 503 is used to determine the target word vector corresponding to the title information of each chapter according to the title information of each chapter and the text information under the title information of each chapter.
[0154] The relationship determination module 504 is used to determine the hierarchical relationship between the title information of each chapter according to the target word vector corresponding to the title information of each chapter.
[0155] The directory construction module 505 is used to construct directory information corresponding to the unstructured document of power emergency to be processed according to the title information of each chapter and the hierarchical relationship between the title information of each chapter.
[0156] In an exemplary embodiment, the vector determination module 503 is also used to perform feature extraction processing on the title information of each chapter to obtain a first word vector corresponding to the title information of each chapter, and to perform feature extraction processing on the text information under the title information of each chapter to obtain a second word vector corresponding to the text information under the title information of each chapter; determine a first weight corresponding to the first word vector, and a second weight corresponding to the second word vector; and fuse the first word vector and the second word vector according to the first weight and the second weight to obtain a target word vector corresponding to the title information of each chapter.
[0157] In an exemplary embodiment, the relationship determination module 504 is also used to determine the similarity between the title information of each chapter based on the target word vector corresponding to the title information of each chapter; and determine the hierarchical relationship between the title information of each chapter based on the similarity between the title information of each chapter.
[0158] In an exemplary embodiment, the power emergency unstructured document directory construction device also includes a directory verification module, which is used to verify the directory information and obtain a verification result corresponding to the power emergency unstructured document to be processed; when the verification result meets the preset conditions, the directory information is used as the target directory information corresponding to the power emergency unstructured document to be processed.
[0159] In an exemplary embodiment, the power emergency unstructured document directory construction device further includes a document preprocessing module, which is used to preprocess the power emergency unstructured document to be processed to obtain the preprocessed power emergency unstructured document.
[0160] In an exemplary embodiment, the information prediction module 502 is also used to input the preprocessed power emergency unstructured document and the keywords in the preprocessed power emergency unstructured document into a pre-trained title information prediction model to obtain the title information of each chapter of the power emergency unstructured document to be processed.
[0161] In an exemplary embodiment, the power emergency unstructured document directory construction device also includes a model training module, which is used to obtain sample power emergency unstructured documents; input the sample power emergency unstructured documents and keywords in the sample power emergency unstructured documents into the title information prediction model to be trained to obtain the predicted title information of each chapter of the sample power emergency unstructured document; according to the difference between the predicted title information and the actual title information of each chapter of the sample power emergency unstructured document, the title information prediction model to be trained is iteratively trained to obtain a pre-trained title information prediction model.
[0162] Each module in the above-mentioned power emergency unstructured document directory construction device can be implemented in whole or in part by software, hardware and their combination. The above-mentioned modules can be embedded in or independent of the processor in the computer device in the form of hardware, or can be stored in the memory of the computer device in the form of software, so that the processor can call and execute the operations corresponding to the above modules.
[0163] In an exemplary embodiment, a computer device is provided. The computer device may be a server, and its internal structure diagram may be as shown in FIG. Figure 6As shown. The computer device includes a processor, a memory, an input / output interface (Input / Output, referred to as I / O) and a communication interface. Among them, the processor, the memory and the input / output interface are connected through a system bus, and the communication interface is connected to the system bus through the input / output interface. Among them, the processor of the computer device is used to provide computing and control capabilities. The memory of the computer device includes a non-volatile storage medium and an internal memory. The non-volatile storage medium stores an operating system, a computer program and a database. The internal memory provides an environment for the operation of the operating system and the computer program in the non-volatile storage medium. The database of the computer device is used to store data such as title information and hierarchical relationships. The input / output interface of the computer device is used to exchange information between the processor and an external device. The communication interface of the computer device is used to communicate with an external terminal through a network connection. When the computer program is executed by the processor, a method for constructing an unstructured document directory for power emergency is implemented.
[0164] Those skilled in the art will understand that Figure 6 The structure shown in the figure is only a block diagram of a part of the structure related to the solution of the present application, and does not constitute a limitation on the computer device to which the solution of the present application is applied. The specific computer device may include more or fewer components than those shown in the figure, or combine certain components, or have a different arrangement of components.
[0165] In an exemplary embodiment, a computer device is further provided, including a memory and a processor, wherein a computer program is stored in the memory, and the processor implements the steps in the above-mentioned method embodiments when executing the computer program.
[0166] In an exemplary embodiment, a computer-readable storage medium is provided, on which a computer program is stored. When the computer program is executed by a processor, the steps in the above-mentioned method embodiments are implemented.
[0167] In an exemplary embodiment, a computer program product is provided, including a computer program, and when the computer program is executed by a processor, the steps in the above method embodiments are implemented.
[0168] Those skilled in the art can understand that all or part of the processes in the above-mentioned embodiment methods can be completed by instructing the relevant hardware through a computer program, and the computer program can be stored in a non-volatile computer-readable storage medium. When the computer program is executed, it can include the processes of the embodiments of the above-mentioned methods. Among them, any reference to the memory, database or other medium used in the embodiments provided in this application can include at least one of non-volatile and volatile memory. Non-volatile memory can include read-only memory (ROM), magnetic tape, floppy disk, flash memory, optical memory, high-density embedded non-volatile memory, resistive random access memory (ReRAM), magnetoresistive random access memory (MRAM), ferroelectric random access memory (FRAM), phase change memory (PCM), graphene memory, etc. Volatile memory can include random access memory (RAM) or external cache memory, etc. As an illustration and not limitation, RAM can be in various forms, such as static random access memory (SRAM) or dynamic random access memory (DRAM). The database involved in each embodiment provided in this application may include at least one of a relational database and a non-relational database. Non-relational databases may include distributed databases based on blockchains, etc., but are not limited to this. The processor involved in each embodiment provided in this application may be a general-purpose processor, a central processing unit, a graphics processor, a digital signal processor, a programmable logic device, a data processing logic device based on quantum computing, etc., but are not limited to this.
[0169] The technical features of the above embodiments may be arbitrarily combined. To make the description concise, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.
[0170] The above-described embodiments only express several implementation methods of the present application, and the descriptions thereof are relatively specific and detailed, but they cannot be understood as limiting the scope of the present application. It should be pointed out that, for a person of ordinary skill in the art, several variations and improvements can be made without departing from the concept of the present application, and these all belong to the protection scope of the present application. Therefore, the protection scope of the present application shall be subject to the attached claims.
Claims
1. A method for constructing a power emergency unstructured document directory, characterized in that: The method comprises: Obtain unstructured documents of power emergencies to be processed; Inputting the pending power emergency unstructured document and keywords in the pending power emergency unstructured document into a pre-trained title information prediction model to obtain title information of each chapter of the pending power emergency unstructured document; Determine the target word vector corresponding to the title information of each chapter according to the title information of each chapter and the text information under the title information of each chapter; Determining the hierarchical relationship between the title information of each chapter according to the target word vector corresponding to the title information of each chapter; According to the title information of each chapter and the hierarchical relationship between the title information of each chapter, the directory information corresponding to the unstructured document of power emergency to be processed is constructed.
2. The method according to claim 1, characterized in that Determining the target word vector corresponding to the title information of each chapter according to the title information of each chapter and the text information under the title information of each chapter includes: Performing feature extraction processing on the title information of each chapter to obtain a first word vector corresponding to the title information of each chapter, and performing feature extraction processing on the text information under the title information of each chapter to obtain a second word vector corresponding to the text information under the title information of each chapter; Determine a first weight corresponding to the first word vector and a second weight corresponding to the second word vector; The first word vector and the second word vector are fused according to the first weight and the second weight to obtain a target word vector corresponding to the title information of each chapter.
3. The method according to claim 1, characterized in that The step of determining the hierarchical relationship between the title information of each chapter according to the target word vector corresponding to the title information of each chapter includes: Determining the similarity between the title information of each chapter according to the target word vector corresponding to the title information of each chapter; According to the similarities between the title information of the chapters, the hierarchical relationship between the title information of the chapters is determined.
4. The method according to claim 1, characterized in that: After constructing the directory information corresponding to the unstructured document for power emergency to be processed according to the title information of each chapter and the hierarchical relationship between the title information of each chapter, the method further includes: Verifying the directory information to obtain a verification result corresponding to the unstructured power emergency document to be processed; When the verification result meets the preset conditions, the directory information is used as the target directory information corresponding to the unstructured power emergency document to be processed.
5. The method according to claim 1, characterized in that After obtaining the unstructured documents of power emergency to be processed, it also includes: Preprocessing the unstructured power emergency document to be processed to obtain a preprocessed unstructured power emergency document; The step of inputting the unstructured document of power emergency to be processed and the keywords in the unstructured document of power emergency to be processed into a pre-trained title information prediction model to obtain the title information of each chapter of the unstructured document of power emergency to be processed includes: The pre-processed electric power emergency unstructured document and keywords in the pre-processed electric power emergency unstructured document are input into a pre-trained title information prediction model to obtain title information of each chapter of the unstructured electric power emergency document to be processed.
6. The method according to any one of claims 1 to 5, characterized in that The pre-trained title information prediction model is trained in the following way: Get sample power emergency unstructured documents; Inputting the sample power emergency unstructured document and keywords in the sample power emergency unstructured document into the title information prediction model to be trained to obtain predicted title information of each chapter of the sample power emergency unstructured document; According to the difference between the predicted title information and the actual title information of each chapter of the sample power emergency unstructured document, the title information prediction model to be trained is iteratively trained to obtain the pre-trained title information prediction model.
7. A device for constructing a power emergency unstructured document directory, characterized in that: The device comprises: A document acquisition module is used to acquire unstructured documents of power emergency to be processed; An information prediction module, used for inputting the pending power emergency unstructured document and keywords in the pending power emergency unstructured document into a pre-trained title information prediction model to obtain title information of each chapter of the pending power emergency unstructured document; A vector determination module, used to determine the target word vector corresponding to the title information of each chapter according to the title information of each chapter and the text information under the title information of each chapter; A relationship determination module, used to determine the hierarchical relationship between the title information of each chapter according to the target word vector corresponding to the title information of each chapter; The directory construction module is used to construct the directory information corresponding to the unstructured document of power emergency to be processed according to the title information of each chapter and the hierarchical relationship between the title information of each chapter.
8. A computer device comprising a memory and a processor, wherein the memory stores a computer program, wherein: When the processor executes the computer program, the steps of the method according to any one of claims 1 to 6 are implemented.
9. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the steps of the method according to any one of claims 1 to 6 are implemented.
10. A computer program product, comprising a computer program, characterized in that When the computer program is executed by a processor, the steps of the method according to any one of claims 1 to 6 are implemented.
Citation Information
Cited By
Non-standard document slicing processing method
CN119538921A
Document information extraction method and device, electronic equipment and medium
CN119783658A
Document identification method, intelligent interaction method, related device, equipment and medium
CN119990112A
Document Recognition Method, Intelligent Interaction Method, and Related Devices, Equipment, and Media
CN119990112B
Method and device for quickly positioning and extracting key information of security market announcement file, storage medium and electronic equipment
CN120578761A
