English knowledge graph timeliness updating method based on real-time corpus

Abstract

The invention provides an English knowledge graph timeliness updating method based on real-time corpora. The English knowledge graph timeliness updating method comprises the steps of real-time corpus obtaining and screening, corpus analysis and knowledge point association, knowledge graph dynamic updating and question bank linkage updating. By acquiring authoritative current affair corpora in real time, automatically analyzing knowledge points, dynamically updating a knowledge graph and linking and updating a question bank, the problems of existing teaching material lag and corpora and knowledge graph staticization are solved, and tight combination of English teaching content and current affair is realized.

Description

Technical Field

[0001] This invention relates to the field of corpus processing, and in particular to a method for timely updating of English knowledge graphs based on real-time corpora. Background Technology

[0002] In current English teaching and learning support systems, core teaching resources (textbooks, corpora, question banks) and knowledge graphs generally suffer from insufficient timeliness.

[0003] Lagging textbook updates: Traditional English textbooks have a revision cycle of 2-3 years, which makes it impossible to incorporate current events (such as emerging vocabulary "AI-generated content" and current events expression "carbon neutrality summit") in a timely manner, resulting in a disconnect between what students learn and actual language application; Fixed corpus sources: Existing systems rely heavily on historical static databases (such as classic articles and fixed dialogues), lacking real-time corpora from authoritative global platforms (such as news reports and trending topics on social media), making it difficult to cover vivid language usage; Static knowledge graph: Once an English knowledge graph (including phonetics, vocabulary, grammar, pragmatics, etc.) is constructed, the knowledge points and their relationships remain unchanged for a long time and cannot be updated with language evolution (such as the expansion of the part of speech and usage of new words such as "metaverse") and current events (such as pragmatic scenarios related to "global pandemic"); The question bank is disconnected from the knowledge graph: the question bank updates rely on manual screening and cannot be synchronized with the real-time updates of the corpus and knowledge graph, resulting in outdated question content and an inability to specifically train students' understanding and application of current affairs language. In existing technologies, although some solutions attempt to update corpora or knowledge graphs, none of them have formed a closed loop of "real-time corpus acquisition, knowledge point analysis, dynamic updating of knowledge graphs, and linked generation of question banks," and they have not focused on the core requirement of "current affairs relevance," thus failing to fundamentally solve the problem of limited teaching materials. Summary of the Invention

[0004] The purpose of this invention is to address the shortcomings of the prior art by providing a method for timely updating English knowledge graphs based on real-time corpora, solving the problems of outdated teaching materials and static corpora and knowledge graphs, and enabling close integration of English teaching content with current events.

[0005] To achieve the above objectives, the present invention adopts the following technical solution: This invention provides a method for timely updating of English knowledge graphs based on real-time corpora, comprising the following steps: S1. Real-time corpus acquisition and filtering; S2. Corpus analysis and knowledge point association; S3. Knowledge graph is dynamically updated; S4, Question Bank Updated in Sync.

[0006] Furthermore, S1 specifically refers to: S101, Corpus Source Configuration: Preset real-time corpus source library, including: international news platforms, social media hot topics, and authoritative industry reports; S102. Real-time crawling and real-time monitoring: The corpus acquisition module adopts a real-time crawling and real-time triggering mechanism; it crawls new content from preset sources at regular intervals; at the same time, it monitors hot tags from the source platform, and triggers emergency crawling immediately once a hot tag appears; S103. Corpus Screening and Cleaning: A pre-trained text classification model is used to filter the crawled corpus, eliminating slow content, niche dialect expressions, and redundant information with no teaching value. The corpus retains the types of texts that meet the needs of English teaching: current news texts, conversational discussions, and explanatory content in professional fields, and ensures that the language difficulty of the corpus covers the entire CEFRA1-C2 level.

[0007] Furthermore, S2 specifically refers to: S201. Structured Decomposition of Corpus: Using natural language processing, the filtered real-time corpus is decomposed to extract core language elements, including: Vocabulary level: new vocabulary, new meanings of old words, fixed collocations; Syntactic layer: Typical syntactic structures in current events scenarios; Pragmatic level: Communicative function in current affairs dialogue; S202. Relationship between knowledge points and existing knowledge graphs: A knowledge point classification mapping rule base is constructed, and the decomposed language elements are matched with the preset levels of the English knowledge graph. A semantic similarity algorithm is used to calculate the correlation between new knowledge points and existing knowledge points in the knowledge graph. If the correlation is ≥80%, the new knowledge point is used as an extension case of the existing knowledge point; if the correlation is <80%, it is determined to be a new knowledge point.

[0008] Furthermore, S3 specifically refers to: S301. Knowledge Point Update Operation: S3011. Add new knowledge points: Create nodes in the corresponding levels of the knowledge graph and add attributes; S3012, Extended Case Update: Add current affairs case sub-nodes under existing knowledge point nodes and associate them with real-time corpus fragments; S3013 Knowledge Point Association Update: If a newly added knowledge point has a logical association with multiple existing knowledge points, a new association edge is created in the knowledge graph, and the association type is labeled. S302, Version Management and Retrospection: S3021. For each knowledge graph update, record the production version number, update time, newly added or modified knowledge point ID, and corresponding real-time corpus source; S3022 supports version rollback functionality. If an error is found in a batch of corpus, it can be rolled back to the previous knowledge graph version with one click.

[0009] Furthermore, S4 specifically includes: S401, Question Bank Generation Rule Configuration: Pre-defined rules for generating question banks for different types of knowledge points: S4011, Vocabulary Knowledge Points: Generate fill-in-the-blank, vocabulary discrimination, and context-based word guessing questions; the question stems and options are derived from real-time corpus. S4012, Grammar Knowledge Points: Generating sentence correction and grammar fill-in-the-blank questions, with the question stems adapted from typical sentences in real-time corpus; S4013, Pragmatic Knowledge Points: Generating situational dialogue completion and pragmatic function judgment questions, with scenarios based on current events; S402. Difficulty Fitting and Question Verification: The complexity of the questions is automatically adjusted based on the difficulty level of the knowledge points: S4021, Elementary Level: Simplified question stems, significantly different answer choices; S4022, Advanced: The question stem retains complex sentence structures, and the options include distractors; AI models are used to validate the generated questions, eliminating ambiguous and incorrect ones; ensuring that the matching degree between the questions and the knowledge points is ≥90%. S403, Question Bank Updates and Push Notifications: S4031. Add the verified questions to the question bank subset for the corresponding knowledge point and label them with current affairs related tags; S4032. Push questions that match real-time knowledge points to users of the English learning companion system.

[0010] The beneficial effects of this invention are as follows: by acquiring authoritative current affairs corpus in real time, automatically analyzing knowledge points and dynamically updating the knowledge graph, and simultaneously updating the question bank, it solves the problems of outdated textbooks and static corpus and knowledge graph, and realizes the close integration of English teaching content with current affairs. Attached Figure Description

[0011] Figure 1 This is a flowchart of a method for updating the timeliness of English knowledge graphs based on real-time corpora. Detailed Implementation

[0012] To make the objectives, technical solutions, and advantages of this invention clearer, the invention will be further described in detail below with reference to the accompanying drawings. It should be understood that the specific embodiments described herein are merely illustrative and not intended to limit the invention.

[0013] Please see Figure 1 A method for timely updating of English knowledge graphs based on real-time corpora includes the following steps: S1. Real-time corpus acquisition and filtering; S2. Corpus analysis and knowledge point association; S3. Knowledge graph is dynamically updated; S4, Question Bank Updated in Sync.

[0014] Specifically, S1 is: S101, Corpus Source Configuration: Preset real-time corpus source library, including: international news platforms, social media hot topics, and authoritative industry reports; S102. Real-time crawling and real-time monitoring: The corpus acquisition module adopts a real-time crawling and real-time triggering mechanism; it crawls new content from preset sources at regular intervals; at the same time, it monitors hot tags from the source platform, and triggers emergency crawling immediately once a hot tag appears; S103. Corpus Screening and Cleaning: A pre-trained text classification model is used to filter the crawled corpus, eliminating slow content, niche dialect expressions, and redundant information with no teaching value. The corpus retains the types of texts that meet the needs of English teaching: current news texts, conversational discussions, and explanatory content in professional fields, and ensures that the language difficulty of the corpus covers the entire CEFRA1-C2 level.

[0015] Specifically, S2 is: S201. Structured Decomposition of Corpus: Using natural language processing, the filtered real-time corpus is decomposed to extract core language elements, including: Vocabulary level: new vocabulary, new meanings of old words, fixed collocations; Syntactic layer: Typical syntactic structures in current events scenarios; Pragmatic level: Communicative function in current affairs dialogue; S202. Relationship between knowledge points and existing knowledge graphs: A knowledge point classification mapping rule base is constructed, and the decomposed language elements are matched with the preset levels of the English knowledge graph. A semantic similarity algorithm is used to calculate the correlation between new knowledge points and existing knowledge points in the knowledge graph. If the correlation is ≥80%, the new knowledge point is used as an extension case of the existing knowledge point; if the correlation is <80%, it is determined to be a new knowledge point.

[0016] Specifically, S3 is: S301. Knowledge Point Update Operation: S3011. Add new knowledge points: Create nodes in the corresponding levels of the knowledge graph and add attributes; S3012, Extended Case Update: Add current affairs case sub-nodes under existing knowledge point nodes and associate them with real-time corpus fragments; S3013 Knowledge Point Association Update: If a newly added knowledge point has a logical association with multiple existing knowledge points, a new association edge is created in the knowledge graph, and the association type is labeled. S302, Version Management and Retrospection: S3021. For each knowledge graph update, record the production version number, update time, newly added or modified knowledge point ID, and corresponding real-time corpus source; S3022 supports version rollback functionality. If an error is found in a batch of corpus, it can be rolled back to the previous knowledge graph version with one click.

[0017] Specifically, S4 is: S401, Question Bank Generation Rule Configuration: Pre-defined rules for generating question banks for different types of knowledge points: S4011, Vocabulary Knowledge Points: Generate fill-in-the-blank, vocabulary discrimination, and context-based word guessing questions; the question stems and options are derived from real-time corpus. S4012, Grammar Knowledge Points: Generating sentence correction and grammar fill-in-the-blank questions, with the question stems adapted from typical sentences in real-time corpus; S4013, Pragmatic Knowledge Points: Generating situational dialogue completion and pragmatic function judgment questions, with scenarios based on current events; S402. Difficulty Fitting and Question Verification: The complexity of the questions is automatically adjusted based on the difficulty level of the knowledge points: S4021, Elementary Level: Simplified question stems, significantly different answer choices; S4022, Advanced: The question stem retains complex sentence structures, and the options include distractors; AI models are used to validate the generated questions, eliminating ambiguous and incorrect ones; ensuring that the matching degree between the questions and the knowledge points is ≥90%. S403, Question Bank Updates and Push Notifications: S4031. Add the verified questions to the question bank subset for the corresponding knowledge point and label them with current affairs related tags; S4032. Push questions that match real-time knowledge points to users of the English learning companion system.

[0018] Example 1: A method for timely updating of English knowledge graph based on real-time corpus includes the following steps. The overall architecture is divided into four core modules: real-time corpus acquisition module, corpus parsing and knowledge point association module, knowledge graph dynamic update module, and question bank linkage update module. Each module works together to achieve timely updates.

[0019] The following uses the "2024 Global Climate Summit (COP30)" as an example to explain in detail the implementation process of this invention: Real-time corpus acquisition and filtering The corpus acquisition module detected that The New York Times published a news article titled "COP30 Summit: NationsAgree on New Carbon Reduction Targets" (tagged "Breaking News"), and immediately triggered crawling to obtain the full text of the news article and related reader discussions; After the text classification model filters, the news text (excluding advertisements) and 3 high-quality comments (excluding meaningless comments) are retained, and the difficulty of the corpus is determined to cover CEFR B1-C2 (including intermediate and advanced vocabulary such as "carbon reduction targets" and "renewable energy subsidies").

[0020] Corpus Analysis and Knowledge Point Connection Deconstructing the corpus to extract key knowledge points: Vocabulary: "carbon reduction targets", "renewable energy subsidies"; Grammar: The phrase "Nations have agreed on..." in the news (present perfect tense, indicating a recent consensus); Pragmatic: "The summit calls on developed countries to..." (a pragmatic structure of appeal); Knowledge Point Connections: The term "carbon reduction targets" has a 92% correlation with the existing knowledge graph's "vocabulary → environmental vocabulary" section, and is therefore classified as an "extended case". The term “renewable energy subsidies” has no matching node in the existing knowledge graph, so it is determined to be a “new knowledge point” and classified as “vocabulary → cross-vocabulary of economy and environment”.

[0021] Knowledge graph dynamic updates A new node, "renewable energy subsidies," has been added with the following attributes: Difficulty B2, Applicable Scenarios (Current Affairs News / Economic Reports), Example Sentence ("COP30 agreed to increase renewable energy subsidies by 20%"), and Chinese Explanation (Renewable Energy Subsidies). Add a current events example under the "Present Perfect Tense" node: "Nations have agreed on new carbon reduction targets"; Establish a "support relationship" link between "renewable energy subsidies" and "carbon reduction targets" (labeled "renewable energy subsidies are used to achieve carbon reduction targets"). The updated version number V20241101-COP30 was generated, and the corpus was recorded from The New York Times (November 1, 2024).

[0022] Question bank updates Generate a B2-level "cloze test" for "renewable energy subsidies": “To achieve the goals of COP30, many countries plan to increase______ to promote clean energy. (A. fossil fuel taxes B. renewable energysubsidies C. carbon emissions)” Generate a B1-level "sentence completion" question based on the grammatical structure "Nations have agreed on...". “Since COP30 ended last week, participating nations ______ on a new plan to cut carbon. (A. agree B. will agree C. have agreed)” AI verification confirmed that the questions were unambiguous, added them to the "Environmental Vocabulary" and "Present Perfect Tense" question banks, and pushed these questions to users who were learning "Environmental Topics".

[0023] The embodiments described above are merely illustrative of implementation methods of the present invention, and while the descriptions are specific and detailed, they should not be construed as limiting the scope of the present invention. It should be noted that those skilled in the art can make various modifications and improvements without departing from the concept of the present invention, and these modifications and improvements all fall within the scope of protection of the present invention. Therefore, the scope of protection of this patent should be defined by the appended claims.

Claims

1. A real-time corpus-based English knowledge graph timeliness updating method, characterized in that, Comprise the following steps: S1, real-time corpus acquisition and screening; S2, corpus analysis and knowledge point association; S3, dynamic update of knowledge graph; S4, question bank linkage update. 2.The method of claim 1, wherein, The S1 is specifically: S101, corpus source configuration: preset real-time corpus source library, including: international news platform, social media hot topics, industry authoritative reports; S102, real-time crawling and real-time monitoring: the corpus acquisition module uses real-time crawling and real-time triggering mechanism; Timely crawl new content from preset sources; At the same time, monitor the hot tags of the source platform, and immediately trigger emergency crawling as soon as a hot tag appears; S103, corpus screening and purification: Use a pre-trained text classification model to screen the crawled corpus, exclude low-speed content, minority dialect expressions, and redundant information without teaching value; Retain corpus types that meet English teaching needs: current affairs news texts, conversational discussions, and explanatory content in professional fields, and ensure that the language difficulty of the corpus covers CEFRA1-C2 full level. 3.The method of claim 2, wherein, The S2 is specifically: S201, corpus structured disassembly: use natural language to disassemble the filtered real-time corpus, extract core language elements, including: Vocabulary layer: new words, old words with new meanings, fixed collocations; Grammar layer: typical grammar structures in current affairs scenarios; Pragmatic layer: communicative functions in current affairs dialogues; S202, knowledge point and existing knowledge graph relationship: Build a knowledge point classification mapping rule library, match the disassembled language elements with the preset levels of the English knowledge graph; Use semantic similarity algorithm to calculate the correlation degree of new knowledge points and existing knowledge points in the knowledge graph, if the correlation degree is ≥80%, the new knowledge point is considered as an extension case of the existing knowledge point; If the correlation degree is < 80%, it is determined as a new knowledge point. 4.The method of claim 3, wherein, The S3 is specifically: S301, knowledge point update operation: S3011, add new knowledge points: create nodes in the corresponding level of the knowledge graph and supplement attributes; S3012, extension case update: add current affairs case sub-nodes under the existing knowledge point node and associate real-time corpus fragments; S3013, knowledge point association update: if the new knowledge point has logical association with multiple existing knowledge points, a new association edge is established in the knowledge graph, and the association type is marked; S302, version management and rollback: S3021, produce version number for each knowledge graph update, record update time, added or modified knowledge point ID, and corresponding real-time corpus source; S3022, support version rollback function, if it is found that some batches of corpus have errors, roll back to the knowledge graph version before the update.

5. The method of claim 4, wherein the method further comprises: The S4 is specifically: S401, question bank generation rule configuration: For different types of knowledge points, preset question bank generation rules: S4011, vocabulary knowledge points: generate word choice fill-in-the-blank, vocabulary discrimination, and context guessing question types, the stem and options come from real-time corpus; S4012, grammar knowledge points: generate sentence correction and grammar fill-in-the-blank question types, the stem is adapted from typical sentences in real-time corpus; S4013, pragmatic knowledge points: generate scenario dialogue completion and pragmatic function judgment question types, the scenario is based on current affairs; S402, difficulty adaptation and question verification: According to the difficulty level of knowledge points, automatically adjust the complexity of the questions: S4021, primary: simplify the stem, and set obvious differences in options; S4022, advanced: keep complex sentence patterns in the stem, and set interference items in options; Use AI models to check the generated questions to exclude ambiguous and incorrect questions; Ensure that the matching degree of the questions and knowledge points is ≥ 90%; S403, update and push the question bank: S4031, add the checked questions to the subset of the corresponding knowledge point question bank, and mark the time-related tags; S4032, push the questions matching the real-time knowledge points to the users of the English companion system.

Images