A method for constructing knowledge graphs based on English vocabulary data

Abstract

The present invention relates to the technical field of English vocabulary data processing, and in particular to a construction method suitable for constructing a knowledge graph based on English vocabulary data. The method comprises the following steps: obtaining an English vocabulary image to be tested; performing image preprocessing on the English vocabulary image to be tested to obtain a preprocessed image to be tested; performing image standardization on the preprocessed image to be tested to generate a standard vocabulary image to be tested; performing text area recognition on the standard vocabulary image to be tested to obtain text area data to be tested; performing area selection on the text area data to be tested to obtain a text area block to be tested. The present invention uses data processing technology, pattern recognition technology, semantic analysis technology and deep learning technology to achieve dynamic optimization of the shape representation structure of handwritten English vocabulary text, and to achieve semantic expression association analysis of the context information of the vocabulary, thereby improving the degree of extraction of handwritten English vocabulary text and accurately presenting English vocabulary expression information.

Description

Technical Field

[0001] The present invention relates to the technical field of English vocabulary data processing, and in particular to a construction method suitable for constructing a knowledge graph based on English vocabulary data. Background Art

[0002] The development of image text extraction technology has gone through manual recognition and automatic extraction. Initially, it was necessary to manually recognize the text in the picture and mainly rely on paper dictionaries to integrate and summarize English vocabulary data. With the introduction of OCR technology (Optical Character Recognition), it can automatically recognize and extract the text in the image and convert it into an editable text format. The OCR tool can recognize various fonts and text arrangements. The English vocabulary data recognized by the OCR tool presents a huge amount of data and is electronic data. Therefore, it is necessary to store the electronic data of English vocabulary and build a knowledge graph to fully present the recognized English vocabulary data. However, due to the irregularity and diversity of handwritten English vocabulary characters in terms of stroke features, shape features, and edge features, the existing English vocabulary image extraction technology has a low degree of extraction of handwritten English vocabulary characters. In addition, the existing English vocabulary image recognition technology fails to perform semantic expression association analysis based on the context information of the vocabulary, making it difficult to accurately present the characteristics of English vocabulary. Summary of the invention

[0003] Based on this, it is necessary to provide a method for constructing a knowledge graph suitable for English vocabulary data to solve at least one of the above technical problems.

[0004] To achieve the above purpose, a method for constructing a knowledge graph based on English vocabulary data is provided, the method comprising the following steps: Step S1: obtaining an English vocabulary image to be tested; performing image preprocessing on the English vocabulary image to be tested to obtain a preprocessed image to be tested; performing image standardization on the preprocessed image to be tested to generate a standard vocabulary image to be tested; Step S2: performing text region recognition on the standard vocabulary image to be tested to obtain text region data to be tested; performing region selection on the text region data to be tested to obtain text region blocks to be tested; performing text vocabulary feature extraction on the text region blocks to be tested to obtain text vocabulary feature data; Step S3: perform vocabulary visual style detection on the text vocabulary feature data to generate vocabulary visual styles; perform vocabulary formation degree perception on the vocabulary visual styles to obtain English vocabulary formation degree; perform dynamic optimization of vocabulary presentation according to the English vocabulary formation degree to obtain English vocabulary optimization data; determine the sentence separation site of the standard vocabulary image to be tested to generate the sentence separation site; perform English vocabulary segmentation on the English vocabulary optimization data based on the sentence separation site to obtain English vocabulary division data; Step S4: Perform regional vocabulary context association on the English vocabulary division data to generate regional vocabulary association data; perform semantic analysis on the regional vocabulary association data to obtain regional vocabulary analysis data; perform English sentence expression meaning matching on the English vocabulary division data based on the regional vocabulary analysis data to obtain sentence expression meaning data; perform vocabulary library construction on the English vocabulary recognition data to obtain an English vocabulary database.

[0005] The present invention obtains an English vocabulary image to be tested and performs image preprocessing to eliminate noise and irrelevant background information in the image, thereby obtaining a preprocessed image to be tested with higher clarity; image standardization further ensures that images from different sources have a unified format and size, which is convenient for subsequent processing; these steps improve the accuracy of image recognition and provide high-quality image data for subsequent text area recognition and feature extraction. Text area recognition can accurately locate the text area in the standard vocabulary image to be tested and extract the text area data to be tested; area selection further refines the position of text vocabulary and provides accurate area blocks for feature extraction; text vocabulary feature extraction can capture the visual features of text vocabulary, and these feature data are the key to identifying and understanding text content; and provide necessary feature information for subsequent vocabulary visual style detection and degree of formation perception. Vocabulary visual style detection generates the visual style of vocabulary based on the extracted text vocabulary feature data, which helps to identify and distinguish different writing styles. The vocabulary formation evaluation generates the English vocabulary formation according to the visual style, and evaluates the writing standardization of each vocabulary; by determining the sentence separation site; it can accurately divide the various words in the sentence and generate the sentence separator site; based on these sites, the English vocabulary optimization data is segmented to obtain the English vocabulary division data, which provides a clear boundary for the further processing of the vocabulary. The English vocabulary division data is associated with the regional vocabulary context to generate the regional vocabulary association data, which can clarify the meaning of the vocabulary in a specific context; semantic analysis is performed based on the regional vocabulary association data, so that the semantic content of the vocabulary is deeply analyzed and understood, and the regional vocabulary analysis data is obtained; finally, the meaning of the English sentence expression can be matched, and the sentence expression meaning data is obtained, so as to more accurately grasp the semantic information of the text; by constructing an English vocabulary database, vocabulary information can be accumulated, and rich data resources are provided for subsequent text recognition and understanding tasks. Therefore, the present invention realizes the dynamic optimization of the shape representation structure of handwritten English vocabulary text through data processing technology, pattern recognition technology, semantic analysis technology and deep learning technology, and realizes the semantic expression association analysis of the context information of the vocabulary; thereby improving the extraction degree of handwritten English vocabulary text and accurately presenting English vocabulary expression information.

[0006] Preferably, step S1 comprises the following steps: Step S11: obtaining an English vocabulary image to be tested; Step S12: performing image preprocessing on the English vocabulary image to be tested to obtain a preprocessed image to be tested; performing image background interference elimination on the preprocessed image to be tested to generate a background denoised image to be tested; Step S13: performing image light source compensation on the background denoised image to be tested to obtain a uniformly illuminated image to be tested; performing image presentation feature enhancement on the uniformly illuminated image to be tested to generate a presentation feature enhanced image; Step S14: grayscale standardization is performed on the image with enhanced features to generate a grayscale standardized image to be tested; size standardization is performed on the pre-processed image to be tested to generate a size standardized image to be tested; Step S15: performing standardized image merging on the grayscale standardized image to be tested and the size standardized image to be tested to obtain a standard vocabulary image to be tested.

[0007] The invention obtains the English vocabulary image to be tested as the starting step of the recognition process, which provides raw data for subsequent image processing and recognition; ensures that the system can receive the English vocabulary image to be recognized, laying a foundation for image preprocessing and feature extraction; performs image preprocessing on the English vocabulary image to be tested, which can effectively improve the image quality and remove irrelevant information in the image; further, by eliminating the image background interference, a background denoised image to be tested is generated, which reduces the interference of background noise on the subsequent recognition process and improves the recognition accuracy; performs image light source compensation on the background denoised image to be tested, eliminates the influence of different lighting conditions, obtains a uniformly illuminated image to be tested, ensures the consistency of the image under different light source environments, and provides a basis for image presentation feature enhancement. Next, the image presentation feature enhancement is performed on the uniformly illuminated image to highlight the key features in the image, providing clearer image data for subsequent image recognition and analysis; the grayscale standardization of the presentation feature enhanced image ensures the consistency of the image in the grayscale level, which is conducive to improving the robustness of the image processing algorithm; at the same time, the size of the pre-processed image to be tested is standardized so that images of different sizes can be unified to the same size standard for subsequent processing; the grayscale standardized image to be tested and the size standardized image to be tested are merged to integrate the grayscale and size standardization processing, and a vocabulary image to be tested that meets the unified standard is generated, which provides high-quality image input for subsequent text area recognition and feature extraction.

[0008] Preferably, step S2 comprises the following steps: Step S21: performing image text marking on the standard vocabulary image to be tested to obtain text marking data; performing text region recognition on the standard vocabulary image to be tested according to the text marking data to obtain text region data to be tested; Step S22: performing region shape recognition on the text region data to be tested to obtain text region shape data; performing region block type classification on the text region shape data to generate region block type data; performing region selection on the text region data to be tested based on the region block type data to obtain a text region block to be tested; Step S23: marking text characters on the text area block to be tested, generating marked text characters to be tested; performing text vocabulary contour recognition on the marked text characters to be tested, obtaining text vocabulary contour features; Step S24: Performing outline structured coding on the text vocabulary outline features to generate vocabulary outline coding data; performing text vocabulary feature mapping on the text area block to be tested according to the vocabulary outline coding data to obtain text vocabulary feature data.

[0009] The present invention performs image text marking on a standard vocabulary image to be tested, and can clarify the position and range of the text in the image; performs text region recognition on the standard vocabulary image to be tested according to text marking data, so that specific text information can be extracted from the image; performs region shape recognition on the text region data to be tested, and can recognize the specific shape features of the text region, providing important geometric information for further processing of the text region; performs region block type division on the text region shape data, so that the text region is classified according to the shape features, which is convenient for subsequent region selection and processing; performs region selection on the text region data to be tested based on the region block type data, ensures the selection precision and accuracy of the text region, and provides accurate region positioning for the recognition and extraction of text characters. By marking the text characters of the text area block to be tested, the individual characters in the text area can be identified, which provides a basis for the recognition of text vocabulary; by recognizing the text vocabulary contours of the marked characters of the text to be tested, the shape and structural features of the text characters can be captured, and detailed contour information can be provided for further analysis of the text vocabulary; by contour structured coding of the text vocabulary contour features, the contour features of the text vocabulary can be converted into processable structured data, which is convenient for computer recognition and processing; by mapping the text vocabulary features of the text area block to be tested according to the vocabulary contour coding data, it is ensured that the contour features of the text vocabulary are associated with the specific text area block, which provides rich feature information for the recognition and understanding of the text vocabulary.

[0010] Preferably, step S3 comprises the following steps: Step S31: performing vocabulary structural decomposition on the text vocabulary feature data to obtain text vocabulary structural features; performing vocabulary visual style detection on the text vocabulary feature data based on the text vocabulary structural features to generate vocabulary visual styles; Step S32: Perform vocabulary formation degree perception on vocabulary visual style to obtain English vocabulary formation degree; Step S33: dynamically optimizing the vocabulary presentation according to the degree of English vocabulary formation to obtain English vocabulary optimization data; Step S34: performing sentence character feature recognition on the standard vocabulary image to be tested to obtain the character features of the sentence to be tested; determining the sentence separation position on the character features of the sentence to be tested to generate the sentence separation position; Step S35: Delineating the word boundaries at the sentence delimiter sites to obtain English word boundaries; performing English word segmentation on the English word optimization data according to the English word boundaries to obtain English word segmentation data.

[0011] The present invention performs vocabulary structural decomposition on text vocabulary feature data, which can decompose complex text vocabulary into structural features that are easier to analyze and identify, thereby improving the accuracy of vocabulary recognition; performs vocabulary visual style detection on text vocabulary feature data based on text vocabulary structural features, ensures the recognition and distinction of texts with different writing styles and fonts, and enhances the adaptability and recognition ability of the system; performs vocabulary formation perception on vocabulary visual style, which can evaluate the writing standardization and integrity of each vocabulary, and provides an important reference basis for subsequent vocabulary optimization; performs dynamic optimization of vocabulary presentation according to English vocabulary formation, which can adjust the display of vocabulary according to the formation of vocabulary to improve its readability and recognition rate, thereby optimizing the presentation effect of vocabulary. Performs sentence character feature recognition on standard vocabulary images to be tested, which can extract the features of each character in the sentence from the image, and provides a basis for further analysis of the sentence; performs sentence separation site determination on the character features of the sentence to be tested, accurately identifies the word boundaries in the sentence, and provides accurate site information for vocabulary segmentation. The lexical boundaries are delineated at the sentence separator sites to clarify the boundaries of each word, providing a basis for accurate word segmentation; English vocabulary optimization data is segmented according to English vocabulary boundaries, and the optimized words are segmented according to the boundaries to obtain accurate English vocabulary division data, providing structured vocabulary units for subsequent text processing and analysis.

[0012] Preferably, step S31 includes the following steps: Step S311: performing vocabulary stroke unit recognition on the text vocabulary feature data to generate vocabulary stroke units; performing English letter decomposition on the vocabulary stroke units to generate vocabulary decomposition letter sets; Step S312: judging the thickness of the word stroke unit to obtain the thickness of the word stroke; determining the word stroke connection mode according to the thickness of the word stroke to generate the stroke connection mode; Step S313: performing single letter shape recognition on the vocabulary decomposed letter set to generate single letter shape features; performing letter geometric attribute extraction on the single letter shape features to obtain letter geometric attributes; Step S314: determining the position of a single letter based on the geometric attributes of the letters to obtain single letter position data; calculating the letter spacing of the vocabulary decomposed letter set based on the single letter position data to generate letter spacing data; Step S315: dividing the vocabulary visual presentation style based on the stroke connection mode, letter geometric attributes and letter spacing data to obtain the vocabulary visual style.

[0013] The present invention performs vocabulary stroke unit recognition on text vocabulary feature data, and can decompose vocabulary into basic stroke units; decomposes vocabulary stroke units into English letters, so that the vocabulary stroke units correspond to specific letters. The stroke thickness judgment of vocabulary stroke units can identify the thickness characteristics of the strokes, providing important geometric information for further processing of the strokes; the vocabulary stroke connection mode is determined according to the thickness of the vocabulary strokes, and the connection relationship between the strokes can be identified, providing key information for the structural analysis of vocabulary; the single letter shape recognition of the vocabulary decomposed letter set can identify the unique shape of each letter, providing a basis for the recognition and classification of the letters; the letter geometric attribute extraction of the single letter shape characteristics is performed, so as to analyze the shape and structural characteristics of the letters and obtain detailed geometric data. By determining the position of individual letters based on their geometric properties, we can determine the specific position of each letter in the vocabulary, and calculate the letter spacing of the vocabulary's letter sets based on the individual letter position data, thereby analyzing the spatial relationship between letters and providing important layout information for the visual presentation of the vocabulary. We divide the visual presentation styles of vocabulary based on the stroke connection method, letter geometric properties, and letter spacing data. By integrating multi-dimensional information such as stroke connection, letter shape, and spacing, we comprehensively analyze and classify the visual styles of vocabulary, providing accurate style information for the visual recognition and presentation of vocabulary.

[0014] Preferably, step S32 includes the following steps: Step S321: determining the stroke connection points of the text vocabulary feature data according to the stroke connection mode to obtain the vocabulary stroke connection points; performing stroke connection clarity detection according to the vocabulary stroke connection points to obtain the stroke connection clarity index; Step S322: calculating the letter aspect ratio of the text vocabulary feature data based on the letter geometric attributes to obtain letter aspect ratio data; evaluating the shape presentation degree of the letter aspect ratio data to obtain a shape presentation evaluation index; Step S323: performing horizontal and vertical spacing identification on the text vocabulary feature data through the letter spacing data to obtain horizontal and vertical spacing features; performing spacing coordination evaluation on the horizontal and vertical spacing features to obtain a letter spacing coordination index; Step S324: quantify the stroke connection clarity index, shape presentation evaluation index and letter spacing coordination index to generate a vocabulary formation level index; perform formation perception evaluation on the vocabulary formation level index to obtain the English vocabulary formation level.

[0015] The present invention determines the stroke connection points of text vocabulary feature data according to the stroke connection mode, and accurately identifies the connection positions of the strokes in the vocabulary; performs stroke connection clarity detection according to the vocabulary stroke connection points, evaluates the recognizability of the stroke connection, and thus improves the accuracy of vocabulary recognition; calculates the letter aspect ratio of the text vocabulary feature data based on the letter geometric attributes, quantifies the shape characteristics of the letters, and provides important geometric parameters for shape presentation evaluation; performs shape presentation evaluation on the letter aspect ratio data, and thus determines whether the shape of the letters conforms to the standard writing specification, and thus improves the accuracy and reliability of letter recognition; performs horizontal and vertical spacing identification on the text vocabulary feature data through letter spacing data, identifies the spatial layout characteristics between letters, and provides a basis for spacing coordination evaluation; performs spacing coordination evaluation on the horizontal and vertical spacing characteristics, and evaluates the consistency and coordination of letter spacing, and thus improves the readability of the text and the accuracy of recognition. The stroke connection clarity index, shape presentation evaluation index and letter spacing coordination index are used to quantify vocabulary formation. This step integrates the evaluation results of multiple dimensions and provides a quantitative grade indicator for vocabulary formation. The formation perception evaluation of the vocabulary formation grade index makes it possible to evaluate the overall formation quality of the vocabulary, which can more accurately evaluate and optimize the visual presentation of the vocabulary and improve the overall performance of vocabulary recognition.

[0016] Preferably, step S33 includes the following steps: Step S331: marking the missing stroke connection points of the text vocabulary feature data according to the degree of English vocabulary formation to obtain the missing stroke connection area; identifying the missing stroke connection points of the missing stroke connection area to generate missing stroke connection point data; and filling the missing stroke connection point data with vocabulary strokes to generate stroke filling data; Step S332: performing letter shape abnormality detection on the text vocabulary feature data according to the English vocabulary formation degree to obtain letter data with abnormal shape presentation; performing letter geometric property optimization on the letter data with abnormal shape presentation to generate letter geometric property optimized data; Step S333: identifying the region with inconsistent letter spacing on the text vocabulary feature data according to the degree of English vocabulary formation to obtain data on the region with inconsistent spacing; adjusting the horizontal and vertical spacing of letters on the data on the region with inconsistent spacing to generate data on letter spacing adjustment; Step S334: dynamically optimize and integrate the stroke filling data, letter geometry attribute optimization data and letter spacing adjustment data for vocabulary presentation to obtain English vocabulary optimization data.

[0017] The present invention marks the missing stroke connection parts of text vocabulary feature data according to the degree of English vocabulary formation, can identify the position of incomplete stroke connection in the vocabulary, and provide accurate positioning for subsequent stroke supplementation; recognize the missing stroke connection points of the stroke connection missing area to clarify the specific position of the missing stroke; fill the vocabulary strokes with the missing stroke connection point data, and improve the integrity and recognizability of the vocabulary by supplementing the missing strokes; perform letter shape abnormality detection on text vocabulary feature data according to the degree of English vocabulary formation, and can identify letters whose shapes do not meet the standard specifications, which provides a basis for the optimization of letter shapes; and detect the abnormal letters with abnormal shapes. The data is optimized for letter geometric properties. By adjusting the geometric properties of letters, the letter shapes are made closer to the standard, thereby improving the recognition accuracy of letters. According to the degree of English vocabulary formation, the text vocabulary feature data is used to identify areas with inconsistent letter spacing. It is able to identify areas where letter spacing does not meet standard specifications, thus providing positioning for the adjustment of letter spacing. The letter spacing is adjusted to make the distance between letters more uniform and coordinated, thereby improving the overall readability of the text. The results of stroke filling, letter geometric property optimization and letter spacing adjustment are combined to comprehensively optimize the visual presentation of vocabulary, thereby improving the overall recognition rate and visual effect of vocabulary.

[0018] Preferably, step S4 comprises the following steps: Step S41: Scanning the English vocabulary division data for vocabulary preambles to obtain preamble scanning data; scanning the English vocabulary division data for vocabulary postambles to obtain postamble scanning data; integrating the preamble scanning data and postamble scanning data into regional vocabulary context to obtain regional context vocabulary data; Step S42: performing context dependency evaluation on the regional context vocabulary data to generate vocabulary context dependency; performing vocabulary context association evaluation on the English vocabulary division data according to the vocabulary context dependency to generate regional vocabulary association data; Step S43: semantically analyze the regional vocabulary association data to obtain regional vocabulary analysis data; match the English sentence expression meaning with the English vocabulary division data based on the regional vocabulary analysis data to obtain sentence expression meaning data; Step S44: constructing an English vocabulary database for the English sentence expression meaning data to obtain an English vocabulary database.

[0019] The present invention performs a vocabulary pre-sentence scanning on the English vocabulary division data, which can capture the sentence information before each vocabulary, and provides context information for understanding the use of vocabulary in the sentence; performs a vocabulary post-sentence scanning on the English vocabulary division data, obtains post-sentence scanning data, and further enriches the context environment of the vocabulary; and constructs a complete context environment for each vocabulary by integrating the pre- and post-sentence information. The regional context vocabulary data is evaluated for context dependency, which can analyze the dependency relationship between the vocabulary and its context, and identify the degree to which the vocabulary meaning is affected by the context; the English vocabulary division data is evaluated for vocabulary context association according to the vocabulary context dependency, and the association strength between the vocabulary and its context is determined, which provides key information for the accurate understanding and use of the vocabulary; the regional vocabulary association data is semantically parsed, which can deeply analyze the semantic relationship between the vocabulary and its context, and extract the specific meaning of the vocabulary; the English vocabulary division data is matched with the English sentence expression meaning based on the regional vocabulary parsing data, and the semantic information of the vocabulary is matched with the sentence structure, so as to accurately understand the overall meaning of the sentence; the English sentence expression meaning data is constructed for the English vocabulary library, and by integrating the sentence expression meaning data into the vocabulary library, a systematic tool is provided for the storage and retrieval of English vocabulary.

[0020] Preferably, step S43 includes the following steps: Step S431: performing semantic unit recognition on the regional vocabulary association data to generate regional vocabulary semantic units; performing semantic association relationship analysis on the regional vocabulary semantic units to obtain vocabulary semantic relationship data; Step S432: performing semantic role assignment on the lexical semantic relationship data to obtain semantic role assignment data; performing lexical function determination on the regional lexical association data based on the semantic role assignment data to obtain English lexical function data; performing lexical function framework analysis on the English lexical function data to obtain regional lexical analysis data; Step S433: matching the regional vocabulary analysis data with the English sentence structure to generate corresponding sentence structure data; matching the sentence expression pattern according to the corresponding sentence structure data to obtain the sentence expression pattern; Step S434: Based on the English vocabulary division data, the sentence expression pattern is mapped to vocabulary meaning to obtain a vocabulary meaning mapping sentence pattern; the vocabulary meaning mapping sentence pattern is presented in an English sentence expression to obtain sentence expression meaning data.

[0021] The present invention identifies semantic units of regional vocabulary association data, and can decompose vocabulary association data into units with clear semantics; performs semantic association relationship analysis on regional vocabulary semantic units, clarifies the connection and dependency between different semantic units, and further reveals the semantic structure between words; performs semantic role allocation on vocabulary semantic relationship data, and can identify the role played by each word in the semantic structure, providing key information for understanding the deep meaning of the sentence; determines vocabulary functions of regional vocabulary association data based on semantic role allocation data, and clarifies the grammatical function and semantic role of each word in the sentence; and further refines the semantic and grammatical features of the vocabulary by parsing the functional framework of the vocabulary; By matching regional vocabulary analysis data with English sentence structures, the vocabulary analysis data can be matched with specific sentence structures, providing a structured framework for sentence understanding; sentence expression pattern matching is performed according to the corresponding sentence structure data, so that common sentence expression patterns can be identified and matched, thereby improving the understanding of sentence structure and function; vocabulary meaning mapping of sentence expression patterns based on English vocabulary division data can map the specific meaning of vocabulary into sentence expression patterns, providing rich semantic information for the semantic understanding of sentences; by integrating vocabulary meaning and sentence structure, complete sentence expression meaning data is generated, providing accurate data support for the final understanding of the intent and meaning of the sentence.

[0022] Preferably, step S44 includes the following steps: Step S441: classifying the English sentence expression meaning data to obtain expression meaning type data; annotating the expression meaning type data with attributes to generate vocabulary attribute annotation items; Step S442: performing English vocabulary association feature recognition based on vocabulary attribute annotation items and expression meaning type data to obtain vocabulary expression association features; constructing an English vocabulary library for English sentence expression meaning data based on the vocabulary expression association features to obtain an English vocabulary database; Step S443: performing English vocabulary entity recognition on the English vocabulary database to obtain English vocabulary entity data; performing entity relationship extraction on the English vocabulary entity data to generate English vocabulary entity relationships; Step S444: performing relationship enhancement processing on English vocabulary entity relationships to obtain vocabulary relationship enhancement data; performing attribute assignment on the vocabulary relationship enhancement data to generate entity assignment attribute data; Step S445: mapping the entity assignment attribute data to a knowledge graph structure to obtain knowledge graph structure data; performing entity linking on the knowledge graph structure data to generate knowledge graph entity links; fusing the knowledge graph entity links with English vocabulary knowledge to obtain knowledge graph fusion content; Step S446: Construct a knowledge graph based on the knowledge graph structure data, knowledge graph entity links and knowledge graph fusion content to generate an English vocabulary knowledge graph.

[0023] The present invention classifies the expression meaning of English sentence expression meaning data, can systematically classify the meaning of sentences, and provides a structured framework for understanding and processing different types of expressions; performs attribute annotation on expression meaning type data, further refines the characteristics of vocabulary, and provides specific attribute information for in-depth analysis and understanding of vocabulary; performs English vocabulary association feature recognition based on vocabulary attribute annotation items and expression meaning type data, can identify the association features between vocabulary, and provides important information for building the connection between vocabulary; constructs an English vocabulary library for English sentence expression meaning data according to the vocabulary expression association features, centrally stores and manages vocabulary information, and provides a basis for subsequent vocabulary retrieval and analysis; performs English vocabulary entity recognition on the English vocabulary database, can identify specific entity information from the database, and provides a basis for extracting entity relationships; performs entity relationship extraction on English vocabulary entity data, and reveals the connection and interaction between entities; Relationship enhancement processing can strengthen the characteristics of entity relationships and improve the accuracy of relationship recognition; attribute assignment is performed on lexical relationship enhancement data to specifically associate attributes with entities; knowledge graph structure mapping is performed on entity assignment attribute data to map entities and attributes to the structure of the knowledge graph, providing a framework for building a knowledge graph; entity linking is performed on knowledge graph structural data to generate knowledge graph entity links, so that scattered information points are connected into a network, enhancing the connection between knowledge; knowledge graph entity links are fused with English vocabulary knowledge to obtain knowledge graph fusion content; knowledge graph is constructed based on knowledge graph structural data, knowledge graph entity links and knowledge graph fusion content, which integrates structural data, entity links and fusion content to construct a comprehensive and interconnected English vocabulary knowledge graph; this knowledge graph not only contains rich vocabulary information, but also covers the relationships and attributes between vocabulary, providing support for in-depth analysis, knowledge retrieval and intelligent applications of English texts.

[0024] The present invention also provides an image text extraction method applicable to English vocabulary data, the method comprising the following steps: Step S1: obtaining an English vocabulary image to be tested; performing image preprocessing on the English vocabulary image to be tested to obtain a preprocessed image to be tested; performing image standardization on the preprocessed image to be tested to generate a standard vocabulary image to be tested; Step S2: performing text region recognition on the standard vocabulary image to be tested to obtain text region data to be tested; performing region selection on the text region data to be tested to obtain text region blocks to be tested; performing text vocabulary feature extraction on the text region blocks to be tested to obtain text vocabulary feature data; Step S3: perform vocabulary visual style detection on the text vocabulary feature data to generate vocabulary visual styles; perform vocabulary formation degree perception on the vocabulary visual styles to obtain English vocabulary formation degree; perform dynamic optimization of vocabulary presentation according to the English vocabulary formation degree to obtain English vocabulary optimization data; determine the sentence separation site of the standard vocabulary image to be tested to generate the sentence separation site; perform English vocabulary segmentation on the English vocabulary optimization data based on the sentence separation site to obtain English vocabulary division data; Step S4: Perform regional vocabulary context association on the English vocabulary division data to generate regional vocabulary association data; perform semantic analysis on the regional vocabulary association data to obtain regional vocabulary analysis data; perform English sentence expression meaning matching on the English vocabulary division data based on the regional vocabulary analysis data to obtain sentence expression meaning data; perform vocabulary library construction on the English vocabulary recognition data to obtain an English vocabulary database. BRIEF DESCRIPTION OF THE DRAWINGS

[0025] Figure 1 A schematic diagram of the steps of a method for constructing a knowledge graph for English vocabulary data; Figure 2 for Figure 1 Detailed implementation steps of step S3 in FIG. Figure 3 for Figure 2 Detailed implementation steps of step S33; The realization of the purpose, functional features and advantages of the present invention will be further explained in conjunction with embodiments and with reference to the accompanying drawings. DETAILED DESCRIPTION

[0026] The technical method of the present invention is described clearly and completely below in conjunction with the accompanying drawings. Obviously, the described embodiments are part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by technicians in this field without creative work are within the scope of protection of the present invention.

[0027] In addition, the accompanying drawings are only schematic illustrations of the present invention and are not necessarily drawn to scale. The same reference numerals in the figures represent the same or similar parts, and their repeated description will be omitted. Some of the block diagrams shown in the accompanying drawings are functional entities and do not necessarily correspond to physically or logically independent entities. The functional entities can be implemented in software form, or implemented in one or more hardware modules or integrated circuits, or implemented in different networks and / or processor methods and / or microcontroller methods.

[0028] It should be understood that, although the terms "first", "second", etc. may be used herein to describe various units, these units should not be limited by these terms. These terms are used only to distinguish one unit from another unit. For example, without departing from the scope of the exemplary embodiments, the first unit may be referred to as the second unit, and similarly the second unit may be referred to as the first unit. The term "and / or" used herein includes any and all combinations of one or more of the listed associated items.

[0029] To achieve this, please refer to Figures 1 to 3 , a method for constructing a knowledge graph suitable for English vocabulary data, the method comprising the following steps: Step S1: obtaining an English vocabulary image to be tested; performing image preprocessing on the English vocabulary image to be tested to obtain a preprocessed image to be tested; performing image standardization on the preprocessed image to be tested to generate a standard vocabulary image to be tested; Step S2: performing text region recognition on the standard vocabulary image to be tested to obtain text region data to be tested; performing region selection on the text region data to be tested to obtain text region blocks to be tested; performing text vocabulary feature extraction on the text region blocks to be tested to obtain text vocabulary feature data; Step S3: perform vocabulary visual style detection on the text vocabulary feature data to generate vocabulary visual styles; perform vocabulary formation degree perception on the vocabulary visual styles to obtain English vocabulary formation degree; perform dynamic optimization of vocabulary presentation according to the English vocabulary formation degree to obtain English vocabulary optimization data; determine the sentence separation site of the standard vocabulary image to be tested to generate the sentence separation site; perform English vocabulary segmentation on the English vocabulary optimization data based on the sentence separation site to obtain English vocabulary division data; Step S4: Perform regional vocabulary context association on the English vocabulary division data to generate regional vocabulary association data; perform semantic analysis on the regional vocabulary association data to obtain regional vocabulary analysis data; perform English sentence expression meaning matching on the English vocabulary division data based on the regional vocabulary analysis data to obtain sentence expression meaning data; perform vocabulary library construction on the English vocabulary recognition data to obtain an English vocabulary database.

[0030] The present invention obtains an English vocabulary image to be tested and performs image preprocessing to eliminate noise and irrelevant background information in the image, thereby obtaining a preprocessed image to be tested with higher clarity; image standardization further ensures that images from different sources have a unified format and size, which is convenient for subsequent processing; these steps improve the accuracy of image recognition and provide high-quality image data for subsequent text area recognition and feature extraction. Text area recognition can accurately locate the text area in the standard vocabulary image to be tested and extract the text area data to be tested; area selection further refines the position of text vocabulary and provides accurate area blocks for feature extraction; text vocabulary feature extraction can capture the visual features of text vocabulary, and these feature data are the key to identifying and understanding text content; and provide necessary feature information for subsequent vocabulary visual style detection and degree of formation perception. Vocabulary visual style detection generates the visual style of vocabulary based on the extracted text vocabulary feature data, which helps to identify and distinguish different writing styles. The vocabulary formation evaluation generates the English vocabulary formation according to the visual style, and evaluates the writing standardization of each vocabulary; by determining the sentence separation site; it can accurately divide the various words in the sentence and generate the sentence separator site; based on these sites, the English vocabulary optimization data is segmented to obtain the English vocabulary division data, which provides a clear boundary for the further processing of the vocabulary. The English vocabulary division data is associated with the regional vocabulary context to generate the regional vocabulary association data, which can clarify the meaning of the vocabulary in a specific context; semantic analysis is performed based on the regional vocabulary association data, so that the semantic content of the vocabulary is deeply analyzed and understood, and the regional vocabulary analysis data is obtained; finally, the meaning of the English sentence expression can be matched, and the sentence expression meaning data is obtained, so as to more accurately grasp the semantic information of the text; by constructing an English vocabulary database, vocabulary information can be accumulated, and rich data resources are provided for subsequent text recognition and understanding tasks. Therefore, the present invention realizes the dynamic optimization of the shape representation structure of handwritten English vocabulary text through data processing technology, pattern recognition technology, semantic analysis technology and deep learning technology, and realizes the semantic expression association analysis of the context information of the vocabulary; thereby improving the extraction degree of handwritten English vocabulary text and accurately presenting English vocabulary expression information.

[0031] In the embodiment of the present invention, reference Figure 1 As shown, it is a schematic diagram of the steps of a construction method for constructing a knowledge graph based on English vocabulary data of the present invention. In this example, the construction method for constructing a knowledge graph based on English vocabulary data includes the following steps: Step S1: obtaining an English vocabulary image to be tested; performing image preprocessing on the English vocabulary image to be tested to obtain a preprocessed image to be tested; performing image standardization on the preprocessed image to be tested to generate a standard vocabulary image to be tested; In an embodiment of the present invention, an image acquisition device, such as a camera or a scanner, is used to photograph or scan a paper containing English words to obtain raw image data; specifically, the English words on the paper are converted into a digital format for subsequent processing. The raw image data is subjected to denoising, and the operation process is as follows: a template size is selected, such as a 3x3 template, wherein the template includes a central pixel and 8 neighboring pixels around it; for each pixel in the image, the average value of all pixel values ​​in the template is calculated, and then the average value is assigned to the central pixel, thereby replacing its original value; this operation is repeated on each pixel of the image until the entire image is processed. The denoised image is subjected to contrast enhancement, specifically, the histogram of the image is first calculated, that is, the number of times each gray level appears in the image is counted; the cumulative distribution function is calculated based on the histogram, and the function represents the proportion of pixels in the image whose brightness value is less than or equal to a specific value; then the cumulative distribution function is normalized so that its value range is between 0 and 1; finally, the normalized cumulative distribution function is used as a mapping to map the original gray value of each pixel to a new gray value. Extract geometric features from the image after contrast enhancement. Specifically, use edge detection algorithms, such as the Canny algorithm, to identify edges in the image by setting high and low thresholds. First, calculate the gradient amplitude and direction of the image, and then perform non-maximum suppression on the gradient amplitude according to the gradient direction to refine the edge. Then, the algorithm determines strong and weak edges according to the set threshold, and connects edge fragments through dual threshold detection and edge tracking processes, and finally extracts edge features of the image. Match the extracted geometric features with the features in the standard image. Specifically, extract key points and descriptors from the two images, and then compare the similarity of the descriptors to find matching feature point pairs. According to the result of feature matching, estimate the geometric transformation relationship between the input image and the standard image. Use the matching feature point pairs to estimate a transformation matrix that describes the transformation from the input image to the standard image. Use the above transformation matrix to perform perspective transformation on the input image. The operation process is to apply the perspective transformation matrix to transform each pixel in the image, and calculate the position of each pixel in the standard image according to the transformation matrix. This process involves interpolation to determine the pixel value at the transformed position, and finally complete the image standardization.

[0032] Step S2: performing text region recognition on the standard vocabulary image to be tested to obtain text region data to be tested; performing region selection on the text region data to be tested to obtain text region blocks to be tested; performing text vocabulary feature extraction on the text region blocks to be tested to obtain text vocabulary feature data; In the embodiment of the present invention, the image segmentation technology is used to process the standard vocabulary image to be tested to identify the text area in the image; the specific operation includes converting the image to a grayscale image, and then applying threshold processing or edge detection algorithm to distinguish text and background features; such as automatically determining the threshold by Otsu's method, converting the image to a binary image, the text area is usually displayed as a highlight part, and the non-text area (such as the background) is a dark part; through this method, a binary image containing text is obtained, in which the text area is clearly marked. After the text area is identified, the region selection technology is further used, such as a method based on connected domain analysis, to determine the specific text area block; specifically, the connected area in the binary image is marked, and then the area containing the text is screened out according to the size, shape or position of the area; such as setting an area threshold, excluding the noise area that is too small, and retaining the area with a larger area and approximately rectangular, which is identified as the text area block to be tested. Feature extraction is performed on the selected text area blocks to obtain feature data of text vocabulary; the text in the image is converted into machine-readable text data through text recognition (OCR) technology; specifically, the text in the image is recognized using the Tesseract-OCR engine to obtain the text content; then, text vectorization technology, such as the Bag of Words model, is used to convert the text into vector form for easy processing by the machine learning model; in the Bag of Words model, each word is regarded as a feature, and the frequency of the word is used as the feature value, and finally the text vocabulary feature data is obtained.

[0033] Step S3: perform vocabulary visual style detection on the text vocabulary feature data to generate vocabulary visual styles; perform vocabulary formation degree perception on the vocabulary visual styles to obtain English vocabulary formation degree; perform dynamic optimization of vocabulary presentation according to the English vocabulary formation degree to obtain English vocabulary optimization data; determine the sentence separation site of the standard vocabulary image to be tested to generate the sentence separation site; perform English vocabulary segmentation on the English vocabulary optimization data based on the sentence separation site to obtain English vocabulary division data; In the embodiment of the present invention, the vocabulary visual style detection process starts with the analysis of the text vocabulary feature data; first, the text in the image is detected by edge detection using image processing technology to identify the outline of the text; for example, by calculating the gradient value of each pixel in the image, the target pixel reflecting the text area in the image is screened out; then, based on the gradient values ​​of these pixels, a clustering algorithm is used to group them to form the vocabulary visual style. Vocabulary formation perception is achieved by evaluating the completeness and clarity of the vocabulary visual style; a deep learning model, such as a convolutional neural network (CNN), is used to analyze the vocabulary visual style and output a formation score; the model can identify the completeness and clarity of the text by learning a large amount of sample data, thereby evaluating the formation of the vocabulary; the specific operation includes inputting the vocabulary visual style into the trained CNN model, the model outputs the formation score of each vocabulary, and obtains the English vocabulary formation. According to the results of vocabulary formation, the dynamic optimization of vocabulary presentation is carried out. In this step, if the vocabulary formation score is low, the image enhancement technology, such as contrast enhancement or sharpening filtering, is used to improve the readability and visual effect of the text; specifically, by adjusting the contrast and brightness of the image, the clarity of the text is improved so that it can maintain good readability in different display environments. The determination of sentence separation sites involves natural language processing technology, especially text segmentation technology, which determines the boundaries of sentences by identifying separators such as spaces and punctuation marks in the text; such as using regular expressions to match punctuation marks such as periods and commas in the text to determine the sentence separation sites; English vocabulary optimization data is segmented based on sentence separator sites, specifically using text processing technology, especially word segmentation technology, by segmenting the text into separate words, and further analyzing and processing; such as using spaces and punctuation marks as separators to segment sentences into separate words to form a vocabulary list.

[0034] Step S4: Perform regional vocabulary context association on the English vocabulary division data to generate regional vocabulary association data; perform semantic analysis on the regional vocabulary association data to obtain regional vocabulary analysis data; perform English sentence expression meaning matching on the English vocabulary division data based on the regional vocabulary analysis data to obtain sentence expression meaning data; perform vocabulary library construction on the English vocabulary recognition data to obtain an English vocabulary database.

[0035] In the embodiment of the present invention, regional vocabulary context association is performed on English vocabulary division data to generate regional vocabulary association data; natural language processing technology, especially context word embedding technology, is used to capture the semantic changes of vocabulary by combining vocabulary with its context information; for example, the context-aware model BERT is used to encode each vocabulary and its context into a high-dimensional vector, and then the similarity or correlation between these vectors is calculated to construct the association data between the vocabulary and its context. Semantic analysis is performed on the regional vocabulary association data, using semantic analysis technology in natural language processing, such as dependency syntactic analysis and semantic role labeling; by analyzing the dependency relationship and semantic role between vocabulary, the deep semantic information of vocabulary is extracted; and natural language processing tools such as SpaCy or NLTK are used to perform dependency syntactic analysis on the text to identify the subject, predicate and object components in the sentence, and then parse the semantics of the vocabulary. Based on regional vocabulary analysis data, the English vocabulary division data is matched with the meaning of English sentence expressions. By using machine learning methods, especially text matching algorithms, such as the SimCSE model, the semantic similarity of sentences is matched and compared; specifically, the sentences are encoded into vector representations, and then the similarity between vectors is calculated to identify sentence expressions with the same or similar meanings. The vocabulary library of English vocabulary recognition data is constructed, specifically using the vocabulary organization and database construction technology in data science, by collecting and organizing a large amount of English vocabulary data, a database containing vocabulary and related information (such as part of speech, definition, example sentences, etc.) is constructed; for example, the data science tool Pandas is used for data organization, and the organized data is stored in the MySQL database.

[0036] For example, the construction method of the knowledge graph for English vocabulary data of the present invention can also be an image text extraction method for English vocabulary data, which can be used to solve problems such as image-text conversion, medical diagnosis records, etc., to extract text from image texts, so as to solve the problem of lack of database in the current medical industry.

[0037] Preferably, step S1 comprises the following steps: Step S11: obtaining an English vocabulary image to be tested; Step S12: performing image preprocessing on the English vocabulary image to be tested to obtain a preprocessed image to be tested; performing image background interference elimination on the preprocessed image to be tested to generate a background denoised image to be tested; Step S13: performing image light source compensation on the background denoised image to be tested to obtain a uniformly illuminated image to be tested; performing image presentation feature enhancement on the uniformly illuminated image to be tested to generate a presentation feature enhanced image; Step S14: grayscale standardization is performed on the image with enhanced features to generate a grayscale standardized image to be tested; size standardization is performed on the pre-processed image to be tested to generate a size standardized image to be tested; Step S15: performing standardized image merging on the grayscale standardized image to be tested and the size standardized image to be tested to obtain a standard vocabulary image to be tested.

[0038] In an embodiment of the present invention, a paper containing English vocabulary is photographed or scanned by an image acquisition device, such as a camera or a scanner, and the English vocabulary on the paper is converted into a digital image format, thereby obtaining an image of the English vocabulary to be tested. The original image is preprocessed, including denoising and contrast enhancement; denoising is achieved by Gaussian filtering, specifically, after adding Gaussian noise to the image, a Gaussian filter is used for smoothing to reduce the impact of noise; contrast enhancement is achieved by histogram equalization, specifically, the histogram of the image is calculated and the pixel values ​​are redistributed; the GrabCut algorithm is used to eliminate background interference from the preprocessed image, specifically, the target area is circled in the image with a mouse, and then the algorithm automatically separates the circled area from the background, and finally an image with background interference removed is obtained through iterative optimization; light source compensation is performed on the image after background denoising, using histogram equalization; histogram equalization improves the uneven illumination problem of the image by adjusting the histogram distribution of the image, so as to achieve a better image quality. Implement illumination compensation; perform feature enhancement on the image after uniform illumination, use logarithmic transformation technology, and enhance the local contrast of the image by taking the logarithmic transformation of the image as the new pixel value; perform grayscale standardization on the image after feature enhancement, that is, convert the image into a grayscale image so that each pixel value is in the range of 0 to 255; specifically, it is achieved through linear transformation, and each channel of the RGB image is converted into a grayscale value; perform size standardization on the pre-processed image to be tested, that is, scale the image to a uniform size; specifically, it is achieved through an interpolation algorithm, such as bilinear interpolation, to maintain the quality of the image during the scaling process; through image fusion technology, the grayscale standardized image to be tested and the size standardized image to be tested are merged to obtain a standard vocabulary image to be tested.

[0039] Preferably, step S2 comprises the following steps: Step S21: performing image text marking on the standard vocabulary image to be tested to obtain text marking data; performing text region recognition on the standard vocabulary image to be tested according to the text marking data to obtain text region data to be tested; Step S22: performing region shape recognition on the text region data to be tested to obtain text region shape data; performing region block type classification on the text region shape data to generate region block type data; performing region selection on the text region data to be tested based on the region block type data to obtain a text region block to be tested; Step S23: marking text characters on the text area block to be tested, generating marked text characters to be tested; performing text vocabulary contour recognition on the marked text characters to be tested, obtaining text vocabulary contour features; Step S24: Performing outline structured coding on the text vocabulary outline features to generate vocabulary outline coding data; performing text vocabulary feature mapping on the text area block to be tested according to the vocabulary outline coding data to obtain text vocabulary feature data.

[0040] In the embodiment of the present invention, the text in the standard vocabulary image to be tested is marked by using image processing technology, specifically using edge detection algorithm, such as Canny algorithm, to identify the text edge in the image; and calculating the gradient value of each pixel in the image, identifying the edge pixels with high gradient value, and these pixels constitute the outline of the text, thereby obtaining text marking data; according to the text marking data, the text area recognition of the standard vocabulary image to be tested is implemented by the maximum stable extreme region (MSER) ​​algorithm, wherein the algorithm can identify the connected area in the image, thereby obtaining the text area data to be tested. The area shape recognition of the text area data to be tested is performed, and the shape features of the text area are identified by calculating the geometric features of the area, such as the roundness and convexity of the area; the text area shape data is divided into area block types, and the text area is classified into different block types, such as single-line text, multi-line text, etc., by analyzing the shape and size features of the text area. Based on the region block type data, the region of the text region data to be tested is selected, and the text region blocks that meet the specific type are screened out by setting specific shape and size thresholds; the text region blocks to be tested are marked with text characters, and the text region blocks are segmented into individual characters through character segmentation algorithms, such as projection analysis or connected component analysis; the text vocabulary contours are recognized for the marked characters of the text to be tested, and the contour features of the characters are recognized and extracted through contour tracking algorithms, such as gradient-based contour tracking; the text vocabulary contour features are structuredly encoded in the contour, and the specific operations are as follows: first, the extracted text contours are discretized to convert the continuous contours into a series of discrete point sets; then, these discrete points are sorted based on the coordinate values ​​of the x-axis or y-axis to determine the order of the contour points; then, the sorted contour points are encoded using a chain code algorithm, and the position change of each point relative to the previous point is represented by a digital code to form a chain code sequence; finally, the chain code sequence is used as the encoding data of the vocabulary contour. Text vocabulary feature mapping is performed on the text area block to be tested according to the vocabulary contour coding data, by mapping the encoded contour features back to the text area of ​​the original image. Specifically, the contour coding data is used as a query and matched with the standard contour coding stored in the database to find the most similar contour coding; the text features corresponding to the matched standard contour coding are mapped to the corresponding area of ​​the original image, thereby obtaining the text vocabulary feature data.

[0041] As an example of the present invention, refer to Figure 2 As shown, in this example, step S3 includes: Step S31: performing vocabulary structural decomposition on the text vocabulary feature data to obtain text vocabulary structural features; performing vocabulary visual style detection on the text vocabulary feature data based on the text vocabulary structural features to generate vocabulary visual styles; Step S32: Perform vocabulary formation degree perception on vocabulary visual style to obtain English vocabulary formation degree; Step S33: dynamically optimizing the vocabulary presentation according to the degree of English vocabulary formation to obtain English vocabulary optimization data; Step S34: performing sentence character feature recognition on the standard vocabulary image to be tested to obtain the character features of the sentence to be tested; determining the sentence separation position on the character features of the sentence to be tested to generate the sentence separation position; Step S35: Delineating the word boundaries at the sentence delimiter sites to obtain English word boundaries; performing English word segmentation on the English word optimization data according to the English word boundaries to obtain English word segmentation data.

[0042] In the embodiment of the present invention, the text vocabulary feature data is structured and decomposed. Specifically, natural language processing technology, especially text structured decomposition technology, is used, and dependency syntax analysis is used to identify the grammatical relationship between words, and the text is decomposed into a structured syntax tree; by analyzing the subject-predicate-object structure in the sentence, the sentence is decomposed into a series of words and phrases. Based on the text vocabulary structure features, the text vocabulary feature data is subjected to vocabulary visual style detection, and specifically image processing technology, especially feature extraction technology, such as SIFT or SURF, is used to extract key points and descriptors from the image for identifying visual patterns and textures in the image; for example, by calculating the gradient direction and size of each pixel in the image, the key feature points in the image are extracted, and then the visual style of the vocabulary is formed. The word formation degree of the word visual style is perceived, and the formation degree of the text in the image is recognized and evaluated using machine learning models such as convolutional neural networks (CNN); the word presentation is dynamically optimized according to the formation degree of English words, and image enhancement techniques such as contrast enhancement or sharpening filtering are applied to improve the readability and visual effect of the text; the clarity of the text is improved by adjusting the contrast and brightness of the image so that it can maintain good readability in different display environments; the sentence character features of the standard vocabulary image to be tested are recognized, and the characters are recognized from the image using OCR technology, especially character recognition technology; for example, each character in the image is recognized by analyzing the edge and shape features in the image. The sentence separation site of the character features of the sentence to be tested is determined, and the text segmentation technology is used to determine the sentence boundary by identifying separators such as spaces and punctuation marks in the text; specifically, regular expressions are used to match punctuation marks such as periods and commas in the text to determine the sentence separation site. The word boundaries are delineated at the sentence separator sites, and the edge detection technology is used to identify the starting and ending positions of the words by analyzing the edge features in the image; English vocabulary optimization data is segmented according to the English vocabulary boundaries, and text processing technology is used to segment the text into separate words for further analysis and processing; for example, spaces and punctuation marks are used as separators to segment sentences into separate words to form a vocabulary list.

[0043] Preferably, step S31 includes the following steps: Step S311: performing vocabulary stroke unit recognition on the text vocabulary feature data to generate vocabulary stroke units; performing English letter decomposition on the vocabulary stroke units to generate vocabulary decomposition letter sets; Step S312: judging the thickness of the word stroke unit to obtain the thickness of the word stroke; determining the word stroke connection mode according to the thickness of the word stroke to generate the stroke connection mode; Step S313: performing single letter shape recognition on the vocabulary decomposed letter set to generate single letter shape features; performing letter geometric attribute extraction on the single letter shape features to obtain letter geometric attributes; Step S314: determining the position of a single letter based on the geometric attributes of the letters to obtain single letter position data; calculating the letter spacing of the vocabulary decomposed letter set based on the single letter position data to generate letter spacing data; Step S315: dividing the vocabulary visual presentation style based on the stroke connection mode, letter geometric attributes and letter spacing data to obtain the vocabulary visual style.

[0044] In an embodiment of the present invention, image recognition technology is used to identify vocabulary stroke units of text vocabulary feature data. Specifically, edge detection algorithms, such as the Canny algorithm, are used to identify text stroke edges in an image, and then continuous edge pixel points are grouped through connected component analysis to form independent stroke units; English letters are decomposed from vocabulary stroke units, and each letter is separated from the vocabulary by identifying the geometric shape and layout of the stroke units; contour detection technology is used to identify the contour of each letter, and the contour is distinguished from other letters by the geometric features of the contour; the thickness of the strokes is determined for the vocabulary stroke units, and the thickness of the strokes is evaluated by calculating the gradient amplitude of the stroke units; the connection mode of the vocabulary strokes is determined according to the thickness of the vocabulary strokes, and the connection relationship between the stroke units is analyzed to determine how the strokes are connected to each other to form letters; the connection mode between the strokes is determined specifically by analyzing the endpoints and directions of the stroke units. Single letter shape recognition is performed on the vocabulary decomposition letter set, and the shape of each letter is identified by analyzing the outline and geometric features of the letter; a machine learning model, such as a support vector machine (SVM), is used to classify the shape features of the letter; the letter geometric attributes are extracted from the single letter shape features, and the geometric features of the letter, such as area, perimeter, aspect ratio, etc., are specifically calculated; and the geometric attributes of the letter are calculated by calculating the coordinates of the boundary points of the letter outline; the single letter position is determined based on the letter geometric attributes, and the spatial coordinates of each letter are determined by analyzing the position of the letter in the image; and the position of the letter is determined by calculating the centroid or bounding box of the letter; the letter spacing of the vocabulary decomposition letter set is measured based on the single letter position data, and the distance between adjacent letters is calculated; specifically, the letter spacing data can be obtained by calculating the Euclidean distance between the centroids of adjacent letters; the visual presentation style of the vocabulary is divided based on the stroke connection method, the geometric attributes of the letter and the letter spacing data, and the vocabulary is classified into different visual styles by comprehensively considering the shape, size and spacing of the letters; and a clustering algorithm, such as K-means, is used to divide the vocabulary into different visual style categories based on the geometric attributes and spacing data of the letters.

[0045] Preferably, step S32 includes the following steps: Step S321: determining the stroke connection points of the text vocabulary feature data according to the stroke connection mode to obtain the vocabulary stroke connection points; performing stroke connection clarity detection according to the vocabulary stroke connection points to obtain the stroke connection clarity index; Step S322: calculating the letter aspect ratio of the text vocabulary feature data based on the letter geometric attributes to obtain letter aspect ratio data; evaluating the shape presentation degree of the letter aspect ratio data to obtain a shape presentation evaluation index; Step S323: performing horizontal and vertical spacing identification on the text vocabulary feature data through the letter spacing data to obtain horizontal and vertical spacing features; performing spacing coordination evaluation on the horizontal and vertical spacing features to obtain a letter spacing coordination index; Step S324: quantify the stroke connection clarity index, shape presentation evaluation index and letter spacing coordination index to generate a vocabulary formation level index; perform formation perception evaluation on the vocabulary formation level index to obtain the English vocabulary formation level.

[0046] In an embodiment of the present invention, an image feature point detection technology is used to determine the stroke connection points of text vocabulary feature data according to the stroke connection method. The specific operation is to use image processing software to identify the stroke connection points in the text, which are the places where the strokes cross or connect; and the stroke connection points are determined by analyzing the endpoints and intersections of the strokes in the image; the stroke connection clarity is detected according to the vocabulary stroke connection points, and whether the strokes are clear and distinguishable is determined by evaluating the clarity of the stroke connection points; for example, the clarity of the stroke connection is evaluated by calculating the pixel contrast at the stroke connection points; the letter aspect ratio is calculated for the text vocabulary feature data based on the letter geometric attributes, and the specific geometric attribute extraction technology, the specific operation includes measuring the bounding box of each letter, calculating its length and width, and then calculating the aspect ratio; for example, the aspect ratio of the letter is obtained by determining the minimum circumscribed rectangle of the letter bounding box. The shape presentation degree of letter aspect ratio data is evaluated, and the shape of the letter is judged as expected by analyzing the aspect ratio of the letter; and the shape presentation degree of the letter is evaluated by comparing the aspect ratio of the letter with the standard aspect ratio; the horizontal and vertical spacing of the text vocabulary feature data are identified by using the spacing measurement technology through the letter spacing data, and the horizontal and vertical distances between the letters are measured; and the horizontal and vertical spacing of the letters are obtained by calculating the Euclidean distance between the centroids of adjacent letters; the spacing coordination of the horizontal and vertical spacing features is evaluated, and whether the spacing is coordinated is judged by evaluating the consistency of the letter spacing; the coordination of the letter spacing is evaluated by comparing the adjacent letter spacing with the average spacing; the stroke connection clarity index, shape presentation evaluation index and letter spacing coordination index are quantified for vocabulary formation, and the above three indicators are combined according to weights through multi-indicator comprehensive evaluation technology to quantify the degree of vocabulary formation; for example, the three indicators are combined into a degree of formation grade index through the weighted method; the degree of formation perception evaluation of the vocabulary formation grade index is performed, and the overall formation quality of the vocabulary is judged by evaluating the vocabulary formation grade index; for example, the overall degree of formation of English vocabulary is evaluated according to the distribution of the vocabulary formation grade index.

[0047] As an example of the present invention, refer to Figure 3 As shown, in this example, step S33 includes: Step S331: marking the missing stroke connection points of the text vocabulary feature data according to the degree of English vocabulary formation to obtain the missing stroke connection area; identifying the missing stroke connection points of the missing stroke connection area to generate missing stroke connection point data; and performing vocabulary stroke filling on the missing stroke connection point data to generate vocabulary stroke filling data; Step S332: performing letter shape abnormality detection on the text vocabulary feature data according to the English vocabulary formation degree to obtain letter data with abnormal shape presentation; performing letter geometric property optimization on the letter data with abnormal shape presentation to generate letter geometric property optimization data; Step S333: identifying the region with inconsistent letter spacing on the text vocabulary feature data according to the degree of English vocabulary formation to obtain data on the region with inconsistent spacing; adjusting the horizontal and vertical spacing of letters on the data on the region with inconsistent spacing to generate data on letter spacing adjustment; Step S334: dynamically optimize and integrate the stroke filling data, letter geometry attribute optimization data and letter spacing adjustment data for vocabulary presentation to obtain English vocabulary optimization data.

[0048] In an embodiment of the present invention, the text vocabulary feature data is marked for stroke connection missing parts according to the degree of English vocabulary formation, and specifically, image processing technology is used to identify the continuity of text strokes, and the integrity of stroke connections is determined by analyzing the endpoints and intersections of the strokes; and the gradient changes of the strokes in the image are analyzed to finally identify the missing areas of stroke connections; the stroke missing connection points are identified for the stroke connection missing areas, and the local features of the stroke connection missing areas are specifically analyzed to identify the specific connection points of the stroke missing; and the stroke missing connection points are determined by calculating the curvature and direction of the stroke endpoints. Using content-based image filling technology, the missing connection point data of strokes are used to fill in vocabulary strokes. Specifically, according to the identified missing connection points of strokes, the image repair algorithm is used to fill in the missing strokes; and the missing strokes are automatically filled by imitating the style and texture of the surrounding strokes; pattern recognition technology, especially anomaly detection algorithms, are used to detect letter shape anomalies in text vocabulary feature data according to the degree of English vocabulary formation. Specifically, machine learning models such as support vector machines (SVM) are used to classify the shape features of letters and identify letters with abnormal shapes; the geometric properties of letters are optimized for letter data with abnormal shapes, and the shape of letters is optimized by adjusting the geometric properties of letters, such as aspect ratio, inclination, etc.; and the geometric properties of letters are adjusted by calculating the geometric center and bounding box of the letters to make them closer to the standard. Accurate shape; according to the degree of English vocabulary formation, the text vocabulary feature data is used to identify the areas with inconsistent letter spacing, and the image analysis technology, especially the spacing measurement technology, is used to specifically measure the horizontal and vertical distances between letters to identify the areas with inconsistent spacing; and the Euclidean distance between the centroids of adjacent letters is calculated to identify the areas with inconsistent spacing; the horizontal and vertical spacing of letters are adjusted for the data in the areas with inconsistent spacing, and the spacing between letters is adjusted to meet the standard spacing; and the horizontal and vertical distances between letters are calculated and adjusted to make the letter spacing more coordinated; data fusion technology, especially multi-source data fusion technology, is used to dynamically optimize and integrate the stroke filling data, letter geometric attribute optimization data and letter spacing adjustment data for vocabulary presentation, and specifically the above three data sources are integrated to generate English vocabulary optimization data.

[0049] Preferably, step S4 comprises the following steps: Step S41: Scanning the English vocabulary division data for vocabulary preambles to obtain preamble scanning data; scanning the English vocabulary division data for vocabulary postambles to obtain postamble scanning data; integrating the preamble scanning data and postamble scanning data into regional vocabulary context to obtain regional context vocabulary data; Step S42: performing context dependency evaluation on the regional context vocabulary data to generate vocabulary context dependency; performing vocabulary context association evaluation on the English vocabulary division data according to the vocabulary context dependency to generate regional vocabulary association data; Step S43: semantically analyze the regional vocabulary association data to obtain regional vocabulary analysis data; match the English sentence expression meaning with the English vocabulary division data based on the regional vocabulary analysis data to obtain sentence expression meaning data; Step S44: constructing an English vocabulary database for the English sentence expression meaning data to obtain an English vocabulary database.

[0050] In the embodiment of the present invention, the text scanning technology is used to scan the pre-sentence of the English vocabulary division data. The specific operation is to use the word segmentation algorithm to divide the text into words and sentences, and then identify the sentence part before each word; and identify and extract the sentence fragment before each word through regular expression matching and sentence boundary detection algorithm; and extract the sentence fragment after each word through similar technical means. The pre-sentence scanning data and the post-sentence scanning data are integrated into the regional vocabulary context by using the data fusion technology, and the pre-sentence and post-sentence data are associated with the target vocabulary to form a vocabulary data set containing context information; the regional context vocabulary data is evaluated for context dependency by using the dependency parsing technology, and the grammatical relationship between the vocabulary and its context is analyzed by using the dependency parsing algorithm; for example, by constructing a dependency tree, the dependency between the vocabulary and its context can be evaluated; the English vocabulary division data is evaluated for vocabulary context association according to the vocabulary context dependency, and natural language processing technology, especially vocabulary association analysis, is used; and the strength of the vocabulary context association is evaluated according to the context dependency evaluation result and by calculating the co-occurrence frequency between the vocabulary and its context. Perform semantic analysis on regional vocabulary-related data, specifically use semantic role labeling and frame recognition technology to analyze the semantic role and semantic frame of vocabulary; and analyze the deep semantic information of vocabulary by labeling the semantic role of vocabulary; match the meaning of English sentence expressions with English vocabulary division data based on regional vocabulary analysis data, and use semantic similarity calculation algorithm to match the semantic expression of sentences; for example, match sentence expressions with the same or similar meanings by calculating the semantic similarity between sentences; construct an English vocabulary library for the English sentence expression meaning data, and specifically use database construction technology to organize the sentence expression meaning data into a structured format and store it in a database; for example, construct an English vocabulary database containing English vocabulary and their sentence expression meanings by designing a database schema and importing data into the database.

[0051] Preferably, step S43 includes the following steps: Step S431: performing semantic unit recognition on the regional vocabulary association data to generate regional vocabulary semantic units; performing semantic association relationship analysis on the regional vocabulary semantic units to obtain vocabulary semantic relationship data; Step S432: performing semantic role assignment on the lexical semantic relationship data to obtain semantic role assignment data; performing lexical function determination on the regional lexical association data based on the semantic role assignment data to obtain English lexical function data; performing lexical function framework analysis on the English lexical function data to obtain regional lexical analysis data; Step S433: matching the regional vocabulary analysis data with the English sentence structure to generate corresponding sentence structure data; matching the sentence expression pattern according to the corresponding sentence structure data to obtain the sentence expression pattern; Step S434: Based on the English vocabulary division data, the sentence expression pattern is mapped to vocabulary meaning to obtain a vocabulary meaning mapping sentence pattern; the vocabulary meaning mapping sentence pattern is presented in an English sentence expression to obtain sentence expression meaning data.

[0052] In an embodiment of the present invention, semantic unit identification is performed on regional vocabulary association data through semantic analysis technology, specifically using dependency syntax analysis and semantic role labeling technology to identify semantic units of vocabulary, and by building a dependency tree, semantic units such as noun phrases and verb phrases in a sentence are identified; semantic association relationship analysis is performed on regional vocabulary semantic units, and the semantic connection between vocabulary is determined by analyzing the dependency relationship and semantic roles between semantic units; for example, by analyzing the action-agent relationship between noun phrases and verb phrases, semantic relationship data between vocabulary is obtained; semantic role assignment is performed on vocabulary semantic relationship data, specifically using semantic role labeling technology, by identifying the semantic role of vocabulary in a sentence (such as agent, patient, etc.), and assigning corresponding semantic roles to the vocabulary; vocabulary function determination is performed on regional vocabulary association data based on the semantic role assignment data, and its grammatical function in a sentence is determined by analyzing the semantic role of vocabulary; for example, by identifying a noun phrase as the subject or object of a sentence, its grammatical function is determined; vocabulary function framework analysis is performed on English vocabulary function data, and by building a vocabulary function framework framework, parsing the grammatical structure and semantic structure of vocabulary in the sentence; for example, by analyzing the functional framework of vocabulary, obtaining detailed grammatical and semantic information of vocabulary in the sentence; matching the regional vocabulary parsing data with the English sentence structure, specifically using syntactic analysis technology, by identifying the sentence structure of the sentence, matching the vocabulary parsing data with the sentence structure; and obtaining the sentence structure corresponding to the vocabulary parsing data by analyzing the subject-predicate-object structure of the sentence; matching the sentence expression pattern according to the corresponding sentence structure data, by comparing the sentence structure data with the known sentence expression pattern, identifying the matching sentence expression pattern; for example, using a pattern matching algorithm, such as a dynamic time warping (DTW) algorithm, matching an expression pattern similar to the input sentence structure; using semantic mapping technology, mapping the sentence expression pattern with vocabulary meaning based on English vocabulary division data, by matching the meaning of the vocabulary with its position in the sentence expression pattern; for example, using a rule-based method or a statistical method, realizing the semantic mapping of vocabulary; converting the vocabulary meaning mapping sentence into a natural language text through natural language generation technology, obtaining sentence expression meaning data.

[0053] Preferably, step S44 includes the following steps: Step S441: classifying the English sentence expression meaning data to obtain expression meaning type data; annotating the expression meaning type data with attributes to generate vocabulary attribute annotation items; Step S442: performing English vocabulary association feature recognition based on vocabulary attribute annotation items and expression meaning type data to obtain vocabulary expression association features; constructing an English vocabulary library for English sentence expression meaning data based on the vocabulary expression association features to obtain an English vocabulary database; Step S443: performing English vocabulary entity recognition on the English vocabulary database to obtain English vocabulary entity data; performing entity relationship extraction on the English vocabulary entity data to generate English vocabulary entity relationships; Step S444: performing relationship enhancement processing on English vocabulary entity relationships to obtain vocabulary relationship enhancement data; performing attribute assignment on the vocabulary relationship enhancement data to generate entity assignment attribute data; Step S445: mapping the entity assignment attribute data to a knowledge graph structure to obtain knowledge graph structure data; performing entity linking on the knowledge graph structure data to generate knowledge graph entity links; fusing the knowledge graph entity links with English vocabulary knowledge to obtain knowledge graph fusion content; Step S446: Construct a knowledge graph based on the knowledge graph structure data, knowledge graph entity links and knowledge graph fusion content to generate an English vocabulary knowledge graph.

[0054] In the embodiment of the present invention, the English sentence expression meaning data is classified by expression meaning, and the text classification technology is specifically used to classify the meaning of the sentence expression; for example, by training a classification model, the sentence is divided into declarative sentences, interrogative sentences and other types; the expression meaning type data is attributed, and the vocabulary attribute annotation technology is used, and the conditional random field (CRF) or deep learning model, such as Bi-LSTM-CRF, is used to annotate the vocabulary attributes; based on the vocabulary attribute annotation items and the expression meaning type data, the English vocabulary association feature is identified, and the feature extraction and feature selection technology is specifically used, and the features related to the sentence expression meaning are extracted from the vocabulary attribute annotation items, and these features are used to identify the association between the words; for example, by analyzing the co-occurrence frequency and context information of the words, the association features between the words are identified. According to the vocabulary expression association features, the English sentence expression meaning data is constructed into an English vocabulary library, and the database construction technology is used to store the vocabulary expression association features in the database, and the relationship between the words is constructed; for example, through the graph database Neo4j, the words and the relationship are stored as nodes and edges to construct an English vocabulary database. Perform English vocabulary entity recognition on the English vocabulary database using entity recognition technology; and use named entity recognition (NER) models, such as the ALBERT+Bi-LSTM model, to identify entities in the text; specifically, train a NER model to identify entities such as names of people and places from the text; perform entity relationship extraction on English vocabulary entity data, and use relationship extraction technology, specifically using deep learning models such as Siamese networks, to identify the relationship between entities; for example, train a relationship extraction model to extract the relationship between entity pairs from the text. Perform relationship enhancement processing on English vocabulary entity relationships, and specifically use machine learning models, such as support vector machines (SVM), to enhance and classify relationships between entities; perform attribute assignment on vocabulary relationship enhancement data, and use attribute assignment technology to assign attributes to entity relationships to enrich the description of entity relationships; for example, by analyzing the characteristics of entity relationships, additional attributes such as relationship strength or confidence are assigned to entity relationships; perform knowledge graph structure mapping on entity assignment attribute data, and specifically map entity relationships to the structure of the knowledge graph to form a framework of the knowledge graph; for example, construct the structure of the knowledge graph by mapping entity relationships to predefined knowledge graph patterns; perform entity linking on knowledge graph structure data, and use entity linking technology to link entities to corresponding entities in the knowledge graph; and link entities to corresponding entities in the knowledge graph by calculating the similarity between entities; perform English vocabulary knowledge fusion on knowledge graph entity links, and specifically fuse knowledge graph contents from different sources together to form a unified knowledge graph; use graph integration technology to construct a knowledge graph based on knowledge graph structure data, knowledge graph entity links, and knowledge graph fusion content to generate an English vocabulary knowledge graph.

[0055] The present invention also provides an image text extraction method applicable to English vocabulary data, the method comprising the following steps: Step S1: obtaining an English vocabulary image to be tested; performing image preprocessing on the English vocabulary image to be tested to obtain a preprocessed image to be tested; performing image standardization on the preprocessed image to be tested to generate a standard vocabulary image to be tested; Step S2: performing text region recognition on the standard vocabulary image to be tested to obtain text region data to be tested; performing region selection on the text region data to be tested to obtain text region blocks to be tested; performing text vocabulary feature extraction on the text region blocks to be tested to obtain text vocabulary feature data; Step S3: perform vocabulary visual style detection on the text vocabulary feature data to generate vocabulary visual styles; perform vocabulary formation degree perception on the vocabulary visual styles to obtain English vocabulary formation degree; perform dynamic optimization of vocabulary presentation according to the English vocabulary formation degree to obtain English vocabulary optimization data; determine the sentence separation site of the standard vocabulary image to be tested to generate the sentence separation site; perform English vocabulary segmentation on the English vocabulary optimization data based on the sentence separation site to obtain English vocabulary division data; Step S4: Perform regional vocabulary context association on the English vocabulary division data to generate regional vocabulary association data; perform semantic analysis on the regional vocabulary association data to obtain regional vocabulary analysis data; perform English sentence expression meaning matching on the English vocabulary division data based on the regional vocabulary analysis data to obtain sentence expression meaning data; perform vocabulary library construction on the English vocabulary recognition data to obtain an English vocabulary database.

[0056] Therefore, the embodiments should be regarded as illustrative and non-restrictive from all points, and the scope of the present invention is limited by the appended claims rather than the above description, and it is therefore intended that all changes falling within the meaning and range of equivalent elements of the application documents are included in the present invention.

[0057] The above description is only a specific embodiment of the present invention, so that those skilled in the art can understand or implement the present invention. Various modifications to these embodiments will be apparent to those skilled in the art, and the general principles defined herein may be implemented in other embodiments without departing from the spirit or scope of the present invention. Therefore, the present invention will not be limited to the embodiments shown herein, but should conform to the widest scope consistent with the principles and novel features invented herein.

Claims

1. A method for constructing a knowledge graph based on English vocabulary data, characterized in that: The following steps are involved: Step S1: obtaining an English vocabulary image to be tested; performing image preprocessing on the English vocabulary image to be tested to obtain a preprocessed image to be tested; performing image standardization on the preprocessed image to be tested to generate a standard vocabulary image to be tested; Step S2: performing text region recognition on the standard vocabulary image to be tested to obtain text region data to be tested; performing region selection on the text region data to be tested to obtain text region blocks to be tested; performing text vocabulary feature extraction on the text region blocks to be tested to obtain text vocabulary feature data; Step S3: Performing vocabulary visual style detection on the text vocabulary feature data to generate vocabulary visual styles; performing vocabulary formation degree perception on the vocabulary visual styles to obtain English vocabulary formation degree; performing dynamic optimization of vocabulary presentation according to the English vocabulary formation degree to obtain English vocabulary optimization data; Determine the sentence separation site of the standard vocabulary image to be tested, and generate the sentence separation site; Performing English vocabulary segmentation on the English vocabulary optimization data based on the sentence separator position to obtain English vocabulary segmentation data; Step S4: performing regional vocabulary context association on the English vocabulary division data to generate regional vocabulary association data; Perform semantic analysis on the regional vocabulary associated data to obtain regional vocabulary analysis data; Based on the regional vocabulary analysis data, the English vocabulary division data is matched with the English sentence expression meaning to obtain the sentence expression meaning data; the vocabulary library is constructed for the English vocabulary recognition data to obtain the English vocabulary database.

2. The method for constructing a knowledge graph based on English vocabulary data according to claim 1, characterized in that: Step S1 includes the following steps: Step S11: obtaining an English vocabulary image to be tested; Step S12: performing image preprocessing on the English vocabulary image to be tested to obtain a preprocessed image to be tested; performing image background interference elimination on the preprocessed image to be tested to generate a background denoised image to be tested; Step S13: performing image light source compensation on the background denoised image to be tested to obtain a uniformly illuminated image to be tested; performing image presentation feature enhancement on the uniformly illuminated image to be tested to generate a presentation feature enhanced image; Step S14: grayscale standardization is performed on the image with enhanced features to generate a grayscale standardized image to be tested; size standardization is performed on the pre-processed image to be tested to generate a size standardized image to be tested; Step S15: performing standardized image merging on the grayscale standardized image to be tested and the size standardized image to be tested to obtain a standard vocabulary image to be tested.

3. The method for constructing a knowledge graph based on English vocabulary data according to claim 1, characterized in that: Step S2 includes the following steps: Step S21: performing image text marking on the standard vocabulary image to be tested to obtain text marking data; performing text region recognition on the standard vocabulary image to be tested according to the text marking data to obtain text region data to be tested; Step S22: performing region shape recognition on the text region data to be tested to obtain text region shape data; performing region block type classification on the text region shape data to generate region block type data; performing region selection on the text region data to be tested based on the region block type data to obtain a text region block to be tested; Step S23: marking text characters on the text area block to be tested, generating marked text characters to be tested; performing text vocabulary contour recognition on the marked text characters to be tested, obtaining text vocabulary contour features; Step S24: Performing outline structured coding on the text vocabulary outline features to generate vocabulary outline coding data; performing text vocabulary feature mapping on the text area block to be tested according to the vocabulary outline coding data to obtain text vocabulary feature data.

4. The method for constructing a knowledge graph based on English vocabulary data according to claim 1, characterized in that: Step S3 includes the following steps: Step S31: performing vocabulary structural decomposition on the text vocabulary feature data to obtain text vocabulary structural features; performing vocabulary visual style detection on the text vocabulary feature data based on the text vocabulary structural features to generate vocabulary visual styles; Step S32: Perform vocabulary formation degree perception on vocabulary visual style to obtain English vocabulary formation degree; Step S33: dynamically optimizing the vocabulary presentation according to the degree of English vocabulary formation to obtain English vocabulary optimization data; Step S34: performing sentence character feature recognition on the standard vocabulary image to be tested to obtain the character features of the sentence to be tested; determining the sentence separation position on the character features of the sentence to be tested to generate the sentence separation position; Step S35: Delineating the word boundaries at the sentence delimiter sites to obtain English word boundaries; performing English word segmentation on the English word optimization data according to the English word boundaries to obtain English word segmentation data.

5. The method for constructing a knowledge graph based on English vocabulary data according to claim 4, characterized in that: Step S31 The following steps are involved: Step S311: performing vocabulary stroke unit recognition on the text vocabulary feature data to generate vocabulary stroke units; performing English letter decomposition on the vocabulary stroke units to generate vocabulary decomposition letter sets; Step S312: judging the thickness of the word stroke unit to obtain the thickness of the word stroke; Determine the word stroke connection mode according to the thickness of the word strokes, and generate the stroke connection mode; Step S313: performing single letter shape recognition on the vocabulary decomposed letter set to generate single letter shape features; performing letter geometric attribute extraction on the single letter shape features to obtain letter geometric attributes; Step S314: determining the position of a single letter based on the geometric attributes of the letters to obtain single letter position data; calculating the letter spacing of the vocabulary decomposed letter set based on the single letter position data to generate letter spacing data; Step S315: dividing the vocabulary visual presentation style based on the stroke connection mode, letter geometric attributes and letter spacing data to obtain the vocabulary visual style.

6. The method for constructing a knowledge graph based on English vocabulary data according to claim 5, characterized in that: Step S32 The following steps are involved: Step S321: determining the stroke connection points of the text vocabulary feature data according to the stroke connection mode to obtain the vocabulary stroke connection points; performing stroke connection clarity detection according to the vocabulary stroke connection points to obtain the stroke connection clarity index; Step S322: calculating the letter aspect ratio of the text vocabulary feature data based on the letter geometric attributes to obtain letter aspect ratio data; evaluating the shape presentation degree of the letter aspect ratio data to obtain a shape presentation evaluation index; Step S323: performing horizontal and vertical spacing recognition on the text vocabulary feature data through the letter spacing data to obtain horizontal and vertical spacing features; The horizontal and vertical spacing features are evaluated for spacing coordination to obtain the letter spacing coordination index; Step S324: quantify the stroke connection clarity index, shape presentation evaluation index and letter spacing coordination index to generate a vocabulary formation level index; perform formation perception evaluation on the vocabulary formation level index to obtain the English vocabulary formation level.

7. The method for constructing a knowledge graph based on English vocabulary data according to claim 6, characterized in that: Step S33 includes the following steps: Step S331: marking the missing stroke connection points of the text vocabulary feature data according to the degree of English vocabulary formation to obtain the missing stroke connection area; identifying the missing stroke connection points of the missing stroke connection area to generate missing stroke connection point data; and filling the missing stroke connection point data with vocabulary strokes to generate stroke filling data; Step S332: performing letter shape abnormality detection on the text vocabulary feature data according to the English vocabulary formation degree to obtain letter data with abnormal shape presentation; performing letter geometric property optimization on the letter data with abnormal shape presentation to generate letter geometric property optimized data; Step S333: identifying the region with inconsistent letter spacing on the text vocabulary feature data according to the degree of English vocabulary formation to obtain data on the region with inconsistent spacing; adjusting the horizontal and vertical spacing of letters on the data on the region with inconsistent spacing to generate data on letter spacing adjustment; Step S334: dynamically optimize and integrate the stroke filling data, letter geometry attribute optimization data and letter spacing adjustment data for vocabulary presentation to obtain English vocabulary optimization data.

8. The method for constructing a knowledge graph based on English vocabulary data according to claim 4, characterized in that: Step S4 includes the following steps: Step S41: Scanning the English vocabulary division data for vocabulary preambles to obtain preamble scanning data; scanning the English vocabulary division data for vocabulary postambles to obtain postamble scanning data; integrating the preamble scanning data and postamble scanning data into regional vocabulary context to obtain regional context vocabulary data; Step S42: performing context dependency evaluation on the regional context vocabulary data to generate vocabulary context dependency; performing vocabulary context association evaluation on the English vocabulary division data according to the vocabulary context dependency to generate regional vocabulary association data; Step S43: semantically analyze the regional vocabulary association data to obtain regional vocabulary analysis data; match the English sentence expression meaning with the English vocabulary division data based on the regional vocabulary analysis data to obtain sentence expression meaning data; Step S44: constructing an English vocabulary database for the English sentence expression meaning data to obtain an English vocabulary database.

9. The method for constructing a knowledge graph based on English vocabulary data according to claim 8, characterized in that: Step S43 includes the following steps: Step S431: performing semantic unit recognition on the regional vocabulary association data to generate regional vocabulary semantic units; performing semantic association relationship analysis on the regional vocabulary semantic units to obtain vocabulary semantic relationship data; Step S432: performing semantic role assignment on the lexical semantic relationship data to obtain semantic role assignment data; performing lexical function determination on the regional lexical association data based on the semantic role assignment data to obtain English lexical function data; performing lexical function framework analysis on the English lexical function data to obtain regional lexical analysis data; Step S433: matching the regional vocabulary analysis data with the English sentence structure to generate corresponding sentence structure data; matching the sentence expression pattern according to the corresponding sentence structure data to obtain the sentence expression pattern; Step S434: Based on the English vocabulary division data, the sentence expression pattern is mapped to vocabulary meaning to obtain a vocabulary meaning mapping sentence pattern; the vocabulary meaning mapping sentence pattern is presented in an English sentence expression to obtain sentence expression meaning data.

10. The method for constructing a knowledge graph based on English vocabulary data according to claim 8, characterized in that: Step S44 includes the following steps: Step S441: classifying the English sentence expression meaning data to obtain expression meaning type data; annotating the expression meaning type data with attributes to generate vocabulary attribute annotation items; Step S442: performing English vocabulary association feature recognition based on vocabulary attribute annotation items and expression meaning type data to obtain vocabulary expression association features; constructing an English vocabulary library for English sentence expression meaning data based on the vocabulary expression association features to obtain an English vocabulary database; Step S443: performing English vocabulary entity recognition on the English vocabulary database to obtain English vocabulary entity data; performing entity relationship extraction on the English vocabulary entity data to generate English vocabulary entity relationships; Step S444: performing relationship enhancement processing on English vocabulary entity relationships to obtain vocabulary relationship enhancement data; performing attribute assignment on the vocabulary relationship enhancement data to generate entity assignment attribute data; Step S445: mapping the entity assignment attribute data to a knowledge graph structure to obtain knowledge graph structure data; performing entity linking on the knowledge graph structure data to generate knowledge graph entity links; fusing the knowledge graph entity links with English vocabulary knowledge to obtain knowledge graph fusion content; Step S446: Construct a knowledge graph based on the knowledge graph structure data, knowledge graph entity links and knowledge graph fusion content to generate an English vocabulary knowledge graph.

Images