Automatic labeling method and device in well drilling field, terminal equipment and storage medium

Abstract

The embodiment of the invention provides an automatic labeling method and device in the well drilling field, terminal equipment and a storage medium, and relates to the technical field of information. The method comprises the following steps: acquiring a labeling result of a first file in the well drilling field; storing the marking result of the first file into a drilling knowledge graph, wherein the drilling knowledge graph is a knowledge graph in the drilling field; training the extraction model based on the drilling knowledge graph; and automatically labeling the file by adopting the extraction model to obtain a labeling result of the file. By adopting the technical scheme provided by the embodiment of the invention, the labeling efficiency can be improved.

Description

Technical Field

[0001] This application relates to the field of information technology, and in particular to an automatic labeling method, apparatus, terminal equipment and storage medium in the field of drilling. Background Technology

[0002] Data annotation refers to the process of marking or annotating specific parts or attributes of data.

[0003] In related technologies, manually labeling samples is inefficient. Summary of the Invention

[0004] This application provides an automatic annotation method, apparatus, terminal device, and storage medium in the drilling field, which can improve annotation efficiency. The technical solution provided by this application is as follows:

[0005] According to one aspect of the embodiments of this application, an automatic annotation method in the field of drilling is provided, the method comprising:

[0006] Obtain the annotation results of the first document in the drilling field;

[0007] The annotation results of the first file are saved to the drilling knowledge graph, which is a knowledge graph in the drilling field;

[0008] Based on the drilling knowledge graph, an extraction model is trained;

[0009] The extraction model is used to automatically annotate the file, and the annotation results of the file are obtained.

[0010] According to one aspect of the embodiments of this application, an automatic marking device for the drilling field is provided, the device comprising:

[0011] The results acquisition module is used to acquire the annotation results of the first document in the drilling field.

[0012] The result saving module is used to save the annotation results of the first file to the drilling knowledge graph, which is a knowledge graph in the drilling field.

[0013] The model training module is used to train the extraction model based on the drilling knowledge graph;

[0014] The automatic annotation module is used to automatically annotate the file using the extraction model to obtain the annotation results of the file.

[0015] According to one aspect of the present application, a computer device is provided, the computer device including a processor and a memory, the memory storing a computer program, the computer program being loaded and executed by the processor to implement the above-described automatic labeling method in the drilling field.

[0016] According to one aspect of the embodiments of this application, a computer-readable storage medium is provided, wherein a computer program is stored in the computer-readable storage medium, the computer program being loaded and executed by a processor to implement the above-described automatic annotation method in the field of drilling.

[0017] According to one aspect of the embodiments of this application, a computer program product is provided, which is loaded and executed by a processor to implement the above-described automatic annotation method in the field of drilling.

[0018] The technical solutions provided in this application embodiment may have the following beneficial effects:

[0019] By storing the annotation results obtained from drilling-related documents in the drilling knowledge graph, the drilling knowledge graph can provide model training materials for drilling-related models to train extraction models. The trained extraction models can then automatically annotate new documents to obtain annotation results for new documents. These annotation results can then be saved and expanded into the drilling knowledge graph, thereby achieving automatic annotation in the drilling field and improving annotation efficiency.

[0020] It should be understood that the above general description and the following detailed description are exemplary and explanatory only, and do not limit this application. Attached Figure Description

[0021] To more clearly illustrate the technical solutions in the embodiments of this application, the accompanying drawings used in the description of the embodiments will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of this application. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0022] Figure 1 This is a flowchart of an automatic annotation method in the drilling field provided in one embodiment of this application;

[0023] Figure 2 This is a schematic diagram of a model training method provided in one embodiment of this application;

[0024] Figure 3 This is a schematic diagram of a standard page provided in one embodiment of this application;

[0025] Figure 4 This is a schematic diagram of model training provided in another embodiment of this application;

[0026] Figure 5 This is a schematic diagram of data extraction provided in one embodiment of this application;

[0027] Figure 6 This is a schematic diagram of a display page provided in one embodiment of this application;

[0028] Figure 7 This is a schematic diagram of a display page provided in another embodiment of this application;

[0029] Figure 8 This is a block diagram of an automatic marking device in the drilling field provided in one embodiment of this application;

[0030] Figure 9 This is a block diagram of a terminal device provided in one embodiment of this application. Detailed Implementation

[0031] Exemplary embodiments will now be described in detail, examples of which are illustrated in the accompanying drawings. When the following description relates to the drawings, unless otherwise indicated, the same numbers in different drawings denote the same or similar elements. The embodiments described in the following exemplary embodiments do not represent all embodiments consistent with this application. Rather, they are merely examples of methods consistent with some aspects of this application as detailed in the appended claims.

[0032] The method provided in this application embodiment can be executed by a terminal device, which refers to an electronic device with data computing, processing, and storage capabilities. This terminal device can be such as a PC (Personal Computer), tablet computer, smartphone, wearable device, intelligent robot, etc.

[0033] The technical solution of this application will be described and illustrated below through several embodiments.

[0034] Please refer to Figure 1 This document illustrates a flowchart of an automatic annotation method in the drilling field provided by an embodiment of this application. In this embodiment, the method is primarily illustrated by its application to the terminal device described above. The method may include at least one of the following steps (110-140).

[0035] Step 110: Obtain the annotation results of the first document in the drilling field.

[0036] In some embodiments, data annotation refers to the process of marking or annotating specific parts or attributes of data, involving different types of data such as text, images, audio, and video. The annotated content can include entities, relationships, attributes, sentiment, classifications, etc. Data annotation can provide high-quality training samples for artificial intelligence applications such as knowledge graphs and large-scale models.

[0037] In some embodiments, the annotation results of the first file can be obtained by manually annotating with annotation tools or by automatically annotating using an annotation model.

[0038] In some embodiments, the first file may be a Word file, a TXT file, or a PDF (Portable Document Format) file.

[0039] Step 120: Save the annotation results of the first file to the drilling knowledge graph, which is a knowledge graph in the drilling field.

[0040] In some embodiments, the drilling knowledge graph can be an indicator graph extended from a general knowledge graph, or it can be generated and expanded from scratch using documents in the drilling domain. When training corpus related to the drilling domain is needed, it can be extracted from the drilling knowledge graph. The annotation tool obtains entity and relation information from the drilling and completion domain knowledge graph as auxiliary information for annotation. The data after annotation is completed is stored in the knowledge graph and can be used as training corpus for the model.

[0041] Step 130: Train the extraction model based on the drilling knowledge graph.

[0042] In some embodiments, the extraction model is an information extraction model with a Transformer network architecture. In some embodiments, the initial extraction model can be a general domain extraction model, and then the extraction model can be trained using drilling domain training samples (such as drilling domain training corpus) to improve the extraction model's annotation accuracy and automatic annotation effect in the drilling domain.

[0043] Step 140: The document is automatically labeled using an extraction model to obtain the labeling results.

[0044] In some embodiments, the extraction model is trained using manually labeled model training corpus to obtain a network model that can be automatically labeled, i.e., the extraction model.

[0045] In some embodiments, such as Figure 2As shown, the annotation tool 21 can be used to manually annotate the multi-source data 22 to obtain the annotation results, which are then saved in the drilling knowledge graph 23. The annotation results are stored as training corpus 24, and the extraction model 25 is trained using the training corpus 24. The annotation results obtained by the automatic annotation of the extraction model 25 are then saved in the drilling knowledge graph 23.

[0046] In summary, the technical solution provided in this application, by storing the annotation results obtained from documents in the drilling field in a drilling knowledge graph, enables the drilling knowledge graph to provide model training materials for the drilling field to train the extraction model. The trained extraction model can then automatically annotate new documents to obtain annotation results for the new documents. These annotation results can then be saved and expanded into the drilling knowledge graph, thereby achieving automatic annotation in the drilling field and improving annotation efficiency.

[0047] In some possible implementations, obtaining the annotation results of the first document in the drilling field includes the following steps:

[0048] 1. If the first file is a PDF file, convert the elements in the first file into image form to obtain the image layer of the labeled page, and the image layer includes at least one image;

[0049] 2. Generate a text layer for the annotation page based on the first file. The text layer includes text content obtained from the first file and is located above the image layer. The annotation page also includes an annotation layer, which is located between the text layer and the image layer.

[0050] 3. In response to the annotation operation on the first text segment in the text content, generate the annotation result of the first file in the annotation layer corresponding to the position of the first text segment.

[0051] In some embodiments, the PDF.js tool is used to parse the PDF file and construct an image layer to render all the visual content of the PDF page. This layer is equivalent to the visual representation of the PDF page, i.e., the image layer. Each element in the PDF file is converted into an image in the image layer. Then, the page.getTextContent() function is called to retrieve the text content from the PDF file page, which is then processed and rendered, and finally displayed in the text layer. In some embodiments, such as... Figure 3 As shown, the annotation page 30 includes a text layer 31, an annotation layer 32, and an image layer 33. The text layer 31 is located above the image layer 33, and the annotation layer 32 exists between the text layer 31 and the image layer 33. In some embodiments, the annotation layer is used to display the annotation results and can clearly show the annotated content. In some embodiments, the annotation method can include at least one of the following: rectangle, highlighting, label, etc.

[0052] In some embodiments, the annotation operation can be the selection of a first text segment by the mouse. When text is selected for annotation, the user's text selection is captured by a designed mouse event handler function. If no text is selected, the function will return immediately without performing any subsequent operations. When the user selects a text segment (such as the first text segment), the system will trigger the mouse event handler function to begin processing the annotation operation. In this function, information about the first text segment selected by the user will be obtained, including the text content of the first text segment, the index of the page where the first text segment is located, etc.

[0053] In some embodiments, generating annotation results for the first document at positions corresponding to the first text segment in the annotation layer includes the following steps:

[0054] 1. Based on the annotation operation, determine the first coordinate information of the first text segment on the screen;

[0055] 2. Convert the first coordinate information into the second coordinate information in the coordinate system corresponding to the first file;

[0056] 3. Based on the first coordinate information, generate the annotation results of the first file in the annotation layer.

[0057] In some embodiments, the coordinates of the rectangular area of ​​the user-selected text on the screen are calculated and then converted into coordinate points in the PDF coordinate system. After coordinate transformation, corresponding annotations are created in the annotation layer in the form of rectangles, highlights, etc., according to the position and size of the selected text, to highlight entities or relationships.

[0058] In some embodiments, generating a text layer for the annotation page based on a first file includes the following steps:

[0059] 1. Call the first function to retrieve the text of the page displayed by the first file;

[0060] 2. Process and render the text to obtain the processed and rendered text content;

[0061] 3. Display text content in the text layer.

[0062] In some embodiments, when processing Word type files, the text in the Word document is extracted by calling Apache's POI and converted into a TXT file to cope with the complexity of rich text structures, ensure high-quality data extraction, and thus achieve more comprehensive and detailed annotation.

[0063] In this embodiment, TXT, PDF, and Word documents are converted into drilling knowledge graphs through manual annotation and automated model extraction. The cyclical self-feedback of manual annotation and automated model extraction efficiently utilizes the labeled data, improving the accuracy and efficiency of the extracted model.

[0064] In some possible implementations, an extraction model is trained based on a drilling knowledge graph, including the following steps:

[0065] 1. Obtain at least one entity and information about the relationships between entities from the knowledge graph;

[0066] 2. Train the extraction model using at least one entity and the relationship information between entities.

[0067] In some embodiments, because knowledge graphs contain entities, attributes, and relationships, common relational databases such as MySQL cannot adequately represent these characteristics of the data. Therefore, knowledge graph data is generally stored using a graph database. Neo4j is a type of graph database that can effectively represent and retrieve connected data. Therefore, this application organizes the knowledge, such as concepts, relationships, and attributes, based on the knowledge structure obtained from the ontology construction module and stores them in the Neo4j graph database.

[0068] In some embodiments, the training extraction model can fine-tune a general-domain extraction model based on a newly labeled dataset (such as drilling-related documents or drilling datasets) to improve information extraction capabilities in new text environments (i.e., drilling-related texts). Figure 4 As shown, training an extraction model can include at least the following steps:

[0069] 1. Obtain the dataset 41 labeled using the annotation tool.

[0070] 2. Use the training script to divide the dataset into training and test sets, and start the training script to train the general domain extraction model 42.

[0071] 3. After training, an extraction model 43 is obtained under the new sample environment. This extraction model 43 is used to extract data (i.e., automatically label) from the new text material 44, and the newly obtained data is saved to the drilling knowledge graph 45.

[0072] In some embodiments, such as Figure 4 As shown, data in the drilling knowledge graph, including newly added data, can be viewed through browser 46. In some embodiments, to obtain training progress, the browser sends a request to the training script, which returns the percentage of training progress completed at the current stage, such as 84% ​​completion. The complete model training process is as follows: Figure 4 As shown.

[0073] This application's embodiments introduce a knowledge graph, integrating file processing, annotation, model training, and data extraction into a complete annotation and training system. This system enables automated extraction of entity and relation data, allowing the annotation results to be stored as high-quality training corpora to improve the performance of subsequent models. It also feeds the annotated data back into the drilling knowledge graph, enabling continuous accumulation and optimization of knowledge (i.e., expanding and optimizing the drilling knowledge graph), thereby improving the accuracy and efficiency of data annotation.

[0074] In some embodiments, the extraction model is designed to transform unstructured text data into structured triple data.

[0075] In some embodiments, such as Figure 5 As shown, the UIE (Universal Information Extraction) model 51, an information extraction model with a Transformer network architecture, is used to further fine-tune the UIE model 51 when facing drilling data extraction tasks. Through the model training method described above, the extraction capability of the UIE model 51 in the drilling data domain can be effectively improved. In some embodiments, since the UIE model 51 extracts by polling all entity concepts, its time complexity is O(n), where n is the number of entity concepts, and only a small number of concepts can be extracted from a text. To solve this problem, the M3E text embedding model 52 is used to calculate the semantic vectors of the source text and concept entities, and similarity scores are calculated. The k concepts with the highest scores are then used for subsequent knowledge extraction, effectively shortening the extraction time. Since the semantic vectors of entities are frequently used, caching them can further accelerate the extraction algorithm.

[0076] In some embodiments, when the first file is a PDF file, the display page 60 of the first file is as follows: Figure 6 , 7 As shown, page 60 displays a document display area 61, a toolbar 62, and an entity list 63. The document display area 61 displays the document's content, the toolbar 62 displays controls for text display and annotation, and the entity list 63 displays entity and attribute relationship information obtained from the drilling knowledge graph, such as the type to be annotated. To annotate, select the entity type in the entity list 63, then draw a line in the document display area 61 to select the text for annotation. The annotation result can be shown as follows: Figure 7 As shown. In some embodiments, text annotation methods can include highlighting, bolding, underlining, etc.

[0077] This application's embodiments use a combination of manual annotation and automated model extraction to convert TXT, PDF, and Word documents into knowledge graphs for the drilling and completion field. Through a cyclical self-feedback process of manual annotation and automated model extraction, it efficiently utilizes the annotated data, thereby improving the accuracy and efficiency of the extracted models.

[0078] The following are embodiments of the apparatus described in this application, which can be used to execute the embodiments of the method described in this application. For details not disclosed in the apparatus embodiments of this application, please refer to the embodiments of the method described in this application.

[0079] Please refer to Figure 8 This diagram illustrates a block diagram of an automatic annotation device for the drilling field according to an embodiment of this application. The device has the functionality to implement the aforementioned automatic annotation method example for the drilling field. This functionality can be implemented in hardware or by hardware executing corresponding software. The device can be the terminal device described above, or it can be installed on a terminal device. The device 800 may include: a result acquisition module 810, a result storage module 820, a model training module 830, and an automatic annotation module 840.

[0080] The result acquisition module 810 is used to acquire the annotation results of the first document in the drilling field.

[0081] The result storage module 820 is used to save the annotation results of the first file to the drilling knowledge graph, which is a knowledge graph in the drilling field.

[0082] The model training module 830 is used to train the extraction model based on the drilling knowledge graph.

[0083] The automatic annotation module 840 is used to automatically annotate the file using the extraction model to obtain the annotation result of the file.

[0084] In some embodiments, the automatic annotation module 840 is configured to:

[0085] If the first file is a portable document format PDF file, the elements in the first file are converted into image form to obtain an image layer of the labeled page, wherein the image layer includes at least one image;

[0086] The annotation page generates a text layer based on the first file. The text layer includes text content obtained from the first file. The text layer is located above the image layer. The annotation page also includes an annotation layer located between the text layer and the image layer.

[0087] In response to an annotation operation on a first text segment in the text content, an annotation result for the first file is generated in the annotation layer corresponding to the position of the first text segment.

[0088] In some embodiments, the automatic annotation module 840 is configured to:

[0089] Based on the annotation operation, determine the first coordinate information of the first text segment on the screen;

[0090] The first coordinate information is converted into the second coordinate information in the coordinate system corresponding to the first file;

[0091] Based on the first coordinate information, the annotation result of the first file is generated in the annotation layer.

[0092] In some embodiments, the automatic annotation module 840 is configured to:

[0093] Call the first function to retrieve the text of the page displayed by the first file;

[0094] The text is processed and rendered to obtain the processed and rendered text content;

[0095] The text content is displayed in the text layer.

[0096] In some embodiments, the model training module 830 is configured to:

[0097] Obtain at least one entity and the relationship information between the entities from the knowledge graph;

[0098] The extraction model is trained using the information about at least one entity and the relationships between the entities.

[0099] In some embodiments, the extraction model is an information extraction model with Transformer as the network architecture.

[0100] In summary, the technical solution provided in this application, by storing the annotation results obtained from documents in the drilling field in a drilling knowledge graph, enables the drilling knowledge graph to provide model training materials for the drilling field to train the extraction model. The trained extraction model can then automatically annotate new documents to obtain annotation results for the new documents. These annotation results can then be saved and expanded into the drilling knowledge graph, thereby achieving automatic annotation in the drilling field and improving annotation efficiency.

[0101] It should be noted that the apparatus provided in the above embodiments is only illustrated by the division of the above functional modules when implementing its functions. In actual applications, the above functions can be assigned to different functional modules as needed, that is, the internal structure of the device can be divided into different functional modules to complete all or part of the functions described above. In addition, the apparatus and method embodiments provided in the above embodiments belong to the same concept, and the specific implementation process can be found in the method embodiments, which will not be repeated here.

[0102] Please refer to Figure 9 This diagram illustrates a structural block diagram of a terminal device 900 provided in one embodiment of this application. The terminal device 900 can be an electronic device such as a mobile phone, tablet computer, game console, e-book reader, multimedia playback device, wearable device, or PC. This terminal device is used to implement the automatic annotation method in the drilling field provided in the above embodiments. Specifically:

[0103] Typically, terminal device 900 includes a processor 901 and a memory 902.

[0104] Processor 901 may include one or more processing cores, such as a quad-core processor, an octa-core processor, etc. Processor 901 may be implemented using at least one hardware form selected from DSP (Digital Signal Processing), FPGA (Field Programmable Gate Array), and PLA (Programmable Logic Array). Processor 901 may also include a main processor and a coprocessor. The main processor, also known as a CPU (Central Processing Unit), is used to process data in the wake-up state; the coprocessor is a low-power processor used to process data in the standby state. In some embodiments, processor 901 may integrate a GPU (Graphics Processing Unit), which is responsible for rendering and drawing the content to be displayed on the screen. In some embodiments, processor 901 may also include an AI (Artificial Intelligence) processor, which is used to handle computational operations related to machine learning.

[0105] The memory 902 may include one or more computer-readable storage media, which may be non-transitory. The memory 902 may also include high-speed random access memory and non-volatile memory, such as one or more disk storage devices or flash memory devices. In some embodiments, the non-transitory computer-readable storage media in the memory 902 are used to store a computer program and are configured to be executed by one or more processors to implement the above-described automatic annotation method in the drilling field.

[0106] In some embodiments, the terminal device 900 may also optionally include a peripheral device interface 903 and at least one peripheral device. The processor 901, memory 902, and peripheral device interface 903 can be connected via a bus or signal line. Each peripheral device can be connected to the peripheral device interface 903 via a bus, signal line, or circuit board. Specifically, the peripheral device includes at least one of a radio frequency circuit 904, a display screen 905, an audio circuit 906, and a power supply 907.

[0107] Those skilled in the art will understand that Figure 9 The structure shown does not constitute a limitation on the terminal device 900, and may include more or fewer components than shown, or combine certain components, or use different component arrangements.

[0108] In an exemplary embodiment, a computer-readable storage medium is also provided, wherein a computer program is stored therein, which, when executed by a processor, implements the above-described automatic annotation method in the field of drilling.

[0109] In an exemplary embodiment, a computer program product is also provided, which is loaded and executed by a processor to implement the above-described automatic annotation method in the field of drilling.

[0110] It should be understood that "multiple" as used in this article refers to two or more. "And / or" describes the relationship between related objects, indicating that three relationships can exist. For example, A and / or B can represent: A alone, A and B simultaneously, or B alone. The character " / " generally indicates that the preceding and following related objects have an "or" relationship.

[0111] The above description is merely an exemplary embodiment of this application and is not intended to limit this application. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of this application should be included within the protection scope of this application.

Claims

1. An automatic annotation method in the field of drilling, characterized in that, The method includes: Obtain the annotation results of the first document in the drilling field; The annotation results of the first file are saved to the drilling knowledge graph, which is a knowledge graph in the drilling field; Based on the drilling knowledge graph, an extraction model is trained; The extraction model is used to automatically annotate the file, and the annotation results of the file are obtained.

2. The method according to claim 1, characterized in that, The method of obtaining the annotation results of the first document in the drilling field includes: If the first file is a portable document format PDF file, the elements in the first file are converted into image form to obtain an image layer of the labeled page, wherein the image layer includes at least one image; The annotation page generates a text layer based on the first file. The text layer includes text content obtained from the first file. The text layer is located above the image layer. The annotation page also includes an annotation layer located between the text layer and the image layer. In response to an annotation operation on a first text segment in the text content, an annotation result for the first file is generated in the annotation layer corresponding to the position of the first text segment.

3. The method according to claim 2, characterized in that, Generating the annotation result of the first file at the position corresponding to the first text segment in the annotation layer includes: Based on the annotation operation, determine the first coordinate information of the first text segment on the screen; The first coordinate information is converted into the second coordinate information in the coordinate system corresponding to the first file; Based on the first coordinate information, the annotation result of the first file is generated in the annotation layer.

4. The method according to claim 2, characterized in that, The process of generating the text layer of the labeled page based on the first file includes: Call the first function to retrieve the text of the page displayed by the first file; The text is processed and rendered to obtain the processed and rendered text content; The text content is displayed in the text layer.

5. The method according to claim 1, characterized in that, The step of training the extraction model based on the drilling knowledge graph includes: Obtain at least one entity and the relationship information between the entities from the knowledge graph; The extraction model is trained using the information about at least one entity and the relationships between the entities.

6. The method according to claim 1, characterized in that, The extraction model is an information extraction model with Transformer as the network architecture.

7. An automatic marking device for the drilling field, characterized in that, The device includes: The results acquisition module is used to acquire the annotation results of the first document in the drilling field. The result saving module is used to save the annotation results of the first file to the drilling knowledge graph, which is a knowledge graph in the drilling field. The model training module is used to train the extraction model based on the drilling knowledge graph; The automatic annotation module is used to automatically annotate the file using the extraction model to obtain the annotation results of the file.

8. A terminal device, characterized in that, The terminal device includes a processor and a memory, the memory storing a computer program, which is loaded and executed by the processor to implement the automatic labeling method in the drilling field as described in any one of claims 1 to 6.

9. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores a computer program, which is loaded and executed by a processor to implement the automatic labeling method in the drilling field as described in any one of claims 1 to 6.

10. A computer program product, characterized in that, The computer program product is loaded and executed by a processor to implement the automatic labeling method in the drilling field as described in any one of claims 1 to 6.

Images