English training method, system and device and storage medium

Abstract

The invention discloses an English training method, system and device and a storage medium, and the key points of the technical scheme are that the method comprises the steps: initially evaluating the capability of a user; obtaining real-time news, constructing an initial article according to the real-time news and the user capability, and adjusting the initial article according to the user capability to obtain a training article; generating training questions according to the training article and the user ability, and sending the training questions to the user; receiving a user answer fed back by the user, and updating the user capability according to the user answer; and judging whether to continue training according to the updated user capability. According to the method, the timeliness and interestingness of the training content can be kept, and meanwhile, the training content is dynamically optimized, so that the learning process is more accurate and efficient.

Description

Technical Field

[0001] The present invention belongs to the field of automated office technology, and in particular relates to an English training method, system, equipment and storage medium. Background Art

[0002] Many existing English textbooks and courses are updated at a limited rate, which can lead to problems with repetitive and outdated training articles and questions.

[0003] In order to avoid the problem of repetition of old-fashioned training articles and training topics, if real-time news reported by BBC (British Broadcasting Corporation) and CCN (Cable News Network) are used as training articles, due to the professionalism of BBC and CCN, the training articles will be more difficult and difficult to be suitable for users of all levels. Summary of the Invention

[0004] The purpose of the present invention is to provide an English training method, system, device and storage medium, which can maintain the timeliness and interest of the training content, while dynamically optimizing the training content to make the learning process more accurate and efficient.

[0005] A first aspect of the present invention provides an English training method, comprising:

[0006] Initial assessment of user capabilities;

[0007] Acquire real-time news, construct an initial article based on the real-time news and user capabilities, and adjust the initial article based on the user capabilities to obtain a training article;

[0008] Generate training questions based on the training articles and user capabilities, and send the training questions to the user;

[0009] receiving user responses from user feedback, and updating user capabilities based on the user responses;

[0010] Determine whether to continue training based on the updated user capabilities.

[0011] In some embodiments, constructing an initial article based on the real-time news and user capabilities includes:

[0012] Extracting basic information of the real-time news and generating news background segments based on the basic information;

[0013] Performing full text analysis on the real-time news, and generating news analysis segments based on the full text analysis results;

[0014] A first keyword of the real-time news is extracted, a related topic is determined according to the first keyword and user capability, and a news guide segment is generated according to the first keyword and the related topic.

[0015] In some embodiments, extracting basic information of the real-time news and generating a news background segment based on the basic information includes:

[0016] Extracting entities and causal relationships of the real-time news to obtain basic information;

[0017] Match topic categories according to the basic information to obtain news categories corresponding to the real-time news;

[0018] A news background segment is generated according to the basic information and news category.

[0019] In some embodiments, performing full text analysis on the real-time news and generating news analysis segments based on the full text analysis results include:

[0020] Performing opinion analysis on the real-time news to obtain news opinions of various subjects on the news event;

[0021] Performing sentiment analysis on the real-time news to obtain the sentiment inclination of each subject towards the news event, and verifying the news opinion of each subject based on the sentiment inclination of each subject to obtain a verification result;

[0022] Extract all arguments of the real-time news, calculate the weight of each argument based on its proportion and sentiment intensity, and take the argument with the highest weight as the core argument;

[0023] Generate a news analysis segment based on the news viewpoints, verification results and core arguments.

[0024] In some embodiments, the user abilities include vocabulary abilities and grammatical abilities; and adjusting the initial article according to the user abilities to obtain a training article includes:

[0025] identifying elementary vocabulary, intermediate vocabulary, and advanced vocabulary of the initial article, and calculating the intermediate vocabulary ratio and the advanced vocabulary ratio;

[0026] replacing at least one of elementary vocabulary, intermediate vocabulary, and advanced vocabulary according to the vocabulary ability until the proportion of intermediate vocabulary reaches a first vocabulary threshold corresponding to the vocabulary ability and the proportion of advanced vocabulary reaches a second vocabulary threshold corresponding to the vocabulary ability, thereby obtaining an intermediate article;

[0027] identifying simple sentence patterns, intermediate sentence patterns, and high-order sentence patterns of the intermediate article, and calculating the proportion of intermediate sentences and the proportion of high-order sentences;

[0028] At least one of the simple sentence patterns, intermediate sentence patterns and high-order sentence patterns is adjusted according to the grammatical ability until the proportion of intermediate sentences reaches a first grammatical threshold corresponding to the grammatical ability and the proportion of high-order sentences reaches a second grammatical threshold corresponding to the grammatical ability, thereby obtaining a training article.

[0029] In some embodiments, the user abilities include: logical ability, transfer ability, and comprehensive ability; and generating training questions based on the training article and the user abilities includes:

[0030] Extracting the core idea of the training article, selecting a corresponding first question template according to the comprehensive ability, substituting the core idea into the corresponding first question template to obtain a first training question;

[0031] Determine non-core viewpoints based on the core viewpoints, select a corresponding second question template based on the comprehensive ability, substitute the non-core viewpoints into the corresponding second question template to obtain a second training question;

[0032] Extracting a second keyword from the training article, determining a transfer direction and a third question template based on the comprehensive ability, determining a transfer topic based on the second keyword and the transfer direction, and generating a third training question based on the second keyword and the transfer topic;

[0033] The number of the second training questions is adjusted according to the logical ability, and the number of the third training questions is adjusted according to the migration ability. The first training questions, all the second training questions and all the third training questions constitute the training questions of the training article.

[0034] In some embodiments, the user answers include: a first answer, a second answer, and a third answer, the first answer being an answer to the first training question, the second answer being an answer to the second training question, and the third answer being an answer to the third training question, and updating the user capabilities based on the user answers includes:

[0035] Counting a first vocabulary size of all words in the user's answer and a second vocabulary size of all correct words, and dividing the second vocabulary size by the first vocabulary size to obtain a vocabulary score for the user;

[0036] Counting the number of first sentences of all sentences in the user's answer and the number of all grammatically correct second sentences, and dividing the number of the second sentences by the first sentences to obtain a grammar score of the user;

[0037] Generate a corresponding first logical quantity according to each training question, count all logically correct second logical quantities in each answer of the user, and divide the second logical quantity by the corresponding first logical quantity to obtain the user's logic score;

[0038] Counting all first opinion values in the first and second answers, counting all second opinion values in the third answer, and dividing the second opinion values by the first opinion values to obtain a user migration score;

[0039] The vocabulary score, grammar score, logic score and migration score are weighted and calculated to obtain a comprehensive score of the user;

[0040] The user's ability is updated according to the vocabulary score, grammar score, logic score, transfer score and comprehensive score.

[0041] A second aspect of the present invention provides an English training system, comprising:

[0042] Capability assessment module, used for initial assessment of user capabilities;

[0043] An article construction module is used to obtain real-time news, construct an initial article based on the real-time news and user capabilities, and adjust the initial article based on the user capabilities to obtain a training article;

[0044] A training question generating module, configured to generate training questions based on the training articles and user capabilities, and send the training questions to the user;

[0045] A capability updating module, configured to receive user responses to user feedback and update user capabilities based on the user responses;

[0046] The training judgment module is used to determine whether to continue training based on the updated user capabilities.

[0047] A third aspect of the present invention provides a computer device, comprising a memory and a processor, wherein the memory stores a computer program, and the processor implements the steps of the above method when executing the computer program.

[0048] A fourth aspect of the present invention provides a computer-readable storage medium having a computer program stored thereon, wherein the computer program implements the steps of the above method when executed by a processor.

[0049] The technical solution provided by the present invention has the following advantages and effects: by generating training articles and training questions based on real-time news, English proficiency is exercised, allowing users to understand the content of real-time news while maintaining the timeliness and interest of the training content, and updating user capabilities in real time. Training articles and training questions are adjusted according to user capabilities, and the training content is dynamically optimized, making the learning process more accurate and efficient. BRIEF DESCRIPTION OF THE DRAWINGS

[0050] Figure 1 1 is a flow chart of the English training method provided by the present invention;

[0051] Figure 2 It is a structural block diagram of the English training system provided by the present invention;

[0052] Figure 3 It is a diagram of the internal structure of a computer device provided by an embodiment of the present invention. DETAILED DESCRIPTION

[0053] To facilitate understanding of the present invention, specific embodiments of the present invention will be described in more detail below with reference to the accompanying drawings.

[0054] Unless otherwise specified or defined, the "first, second..." used in this article is only used to distinguish names and does not represent a specific quantity or order.

[0055] Unless stated otherwise or defined otherwise, the term "and / or" as used herein includes any and all combinations of one or more of the associated listed items.

[0056] It should be noted that, in this document, “fixed to” or “connected to” may mean directly fixing or connecting to an element, or indirectly fixing or connecting to an element.

[0057] like Figure 1 As shown, this embodiment provides an English training method, including the following steps S1 to S5:

[0058] Step S1: Initially assess user capabilities.

[0059] In practical applications, vocabulary is first graded based on vocabulary difficulty, categorizing it into elementary, intermediate, and advanced vocabulary. This is a rough classification. Then, elementary, intermediate, and advanced vocabulary databases are established, respectively, based on the elementary, intermediate, and advanced vocabulary levels. The vocabulary in the elementary, intermediate, and advanced vocabulary databases is further subdivided based on vocabulary difficulty, and vocabulary spelling exercises are given based on the subdivided levels. These vocabulary spelling exercises are then sent to the user, who then answers the questions and receives a spelling response. The spelling responses are received and the spelling accuracy rate for each coarsely graded group is calculated. The user's vocabulary ability is then determined based on the spelling accuracy rate.

[0060] For example, there are 10 groups of subdivision levels, 4 groups of subdivision levels for the elementary vocabulary database, 3 groups of subdivision levels for the intermediate vocabulary database, and 3 groups of subdivision levels for the advanced vocabulary database. Each subdivision level has 3 questions, for a total of 30 questions. After obtaining the user's spelling accuracy, the spelling accuracy is substituted into the spelling formula to calculate the user's initial vocabulary score, which is: Among them, the C i Expressing each group of coarse classification coefficients, the D iIndicates the spelling accuracy of each group of coarse grades. The user's initial vocabulary ability is determined based on the user's initial vocabulary score.

[0061] Then, the user's grammatical ability is tested, and incomplete sentences are given as grammar questions. The incomplete sentences can be extracted from the preset corpus (based on authoritative grammar books or test question banks, such as TOEFL, IELTS, GRE), or they can be generated based on news, current affairs, and sentences that meet the target grammatical points, or they can be generated by extracting high-frequency error points from past grammatical errors. For incomplete sentences, users need to supplement the missing words / phrases. Grammatical points such as tenses (general, perfect, progressive, etc.), clauses (attributive, adverbial, noun), subjunctive mood, passive voice, and inverted sentences are involved. Users answer grammatical questions and get grammatical answers. The user's grammatical answers are received and the grammatical accuracy of each grammatical question is calculated. The user's grammatical ability is determined based on the grammatical accuracy. For example, if 10 incomplete sentences are given as grammar questions, in a grammar question, 1 point will be deducted for one wrong point, and 10 points will be added for a completely correct answer. For high-difficulty grammar questions (inversion, subjunctive mood, etc.), the accuracy rate is weighted and increased by 10%. The grammar scoring formula is: Among them, the E i Indicates the grammatical accuracy of each question, the F i Represents the weighted score of high-level grammar, which determines the user's initial grammar ability based on the user's initial grammar score.

[0062] Finally, the user's logical and transfer abilities are tested. A short passage and a corresponding question are given as reading questions. The user is asked to answer the reading questions and obtain the user's reading answers. Semantic analysis is then performed on the user's reading answers, and the correct logical answer volume, the number of matching viewpoints with the passage, and the number of associated viewpoints associated with the passage are counted. The logical answer volume is divided by the number of logical points in the passage to obtain the user's initial logical score. This initial logical score is used to obtain the user's initial logical ability. The number of associated viewpoints is divided by the number of matching viewpoints to obtain the user's initial transfer score. This initial transfer ability is then obtained from the user's initial transfer score. The user's initial vocabulary score, grammar score, logic score, and transfer score are weighted and summed to obtain the user's initial comprehensive score. This initial comprehensive score is then used to obtain the user's initial comprehensive ability. The weight coefficients for the vocabulary score, grammar score, logic score, and transfer score are preset default coefficients.

[0063] Step S2: real-time news is obtained, an initial article is constructed according to the real-time news and user capabilities, and the initial article is adjusted according to the user capabilities to obtain a training article.

[0064] In practical applications, real-time news can be obtained from BBC (British Broadcasting Corporation) and CCN (Cable News Network), which are professional and real-time.

[0065] Specifically, constructing an initial article based on the real-time news and user capabilities includes:

[0066] Extracting basic information of the real-time news and generating news background segments based on the basic information;

[0067] Performing full text analysis on the real-time news, and generating news analysis segments based on the full text analysis results;

[0068] A first keyword of the real-time news is extracted, a related topic is determined according to the first keyword and user capability, and a news guide segment is generated according to the first keyword and the related topic.

[0069] In actual applications, the initial article is constructed based on real-time news and user capabilities. It is possible to preliminarily construct an article suitable for the user based on the user's capabilities. The initial article is divided into a news background section, a news analysis section, and a news guide section to ensure that the logic of the article is clear. The news background section, the news analysis section, and the news guide section can all include at least one paragraph. For example, the news analysis section includes 2 to 3 small paragraphs.

[0070] Specifically, extracting basic information of the real-time news and generating a news background segment based on the basic information includes:

[0071] Extracting entities and causal relationships of the real-time news to obtain basic information;

[0072] Match topic categories according to the basic information to obtain news categories corresponding to the real-time news;

[0073] A news background segment is generated according to the basic information and news category.

[0074] In practical applications, entities include people, time, place, and events. People include the names of individuals, organizations, and groups in news (e.g., "A," "United Nations"). Time includes absolute time (e.g., "April 1, 2025") and relative time (e.g., "three weeks later"). Places include geographic locations (e.g., "Beijing," "Pacific Ocean"), administrative divisions, and specific venues (e.g., "Capitol"). Events include political events (e.g., "signing of an agreement") and disasters (e.g., "earthquake"). Pre-trained language models (e.g., BERT) can be used to identify entities and causal relationships in real-time news, obtaining basic information about the news. This is then based on pre-defined rules and patterns. For example, different topic categories (e.g., politics, economy, science and technology, culture, sports, disasters, etc.) have their own common entity types and relationship patterns. For example, in the science and technology category, entities such as technology companies, researchers, and new technology names are often found, as well as behaviors and causal relationships related to R&D and innovation (e.g., technological breakthroughs leading to product upgrades). Once this basic information is matched to a corresponding topic category, that topic category is used as the news category for the real-time news. The extracted basic information is compared and matched with the patterns of these different topic categories, and then the basic information and news categories are input into a trained first text generation model (such as BART, GPT, etc.). The first text generation model takes the basic information and news categories of real-time news as input, and generates news background segments based on its learned language patterns and semantic understanding capabilities.

[0075] Specifically, the full-text analysis of the real-time news and the generation of news analysis segments based on the full-text analysis results include:

[0076] Performing opinion analysis on the real-time news to obtain news opinions of various subjects on the news event;

[0077] Performing sentiment analysis on the real-time news to obtain the sentiment inclination of each subject towards the news event, and verifying the news opinion of each subject based on the sentiment inclination of each subject to obtain a verification result;

[0078] Extract all arguments of the real-time news, calculate the weight of each argument based on its proportion and sentiment intensity, and take the argument with the highest weight as the core argument;

[0079] Generate a news analysis segment based on the news viewpoints, verification results and core arguments.

[0080] In practical applications, a trained first analysis model (such as BERT, RoBERTa, etc.) is used to perform opinion analysis on real-time news to obtain the news opinions of each subject on the news event. For example, if A expresses support for environmental protection policies, then a trained second analysis model is used to perform sentiment analysis on real-time news to obtain the emotional tendencies of each subject towards the news event. For example, if A uses positive emotional words such as "significant improvement" to describe environmental protection policies, it means that the news opinions and emotional tendencies of subject A on environmental protection policies obtained from the analysis are consistent, and subject A's news opinions on environmental protection policies are credible. If A uses negative emotional words such as "radical" to describe environmental protection policies, it means that the news opinions and emotional tendencies of subject A on environmental protection policies obtained from the analysis are inconsistent, and subject A's news opinions on environmental protection policies are uncredible.

[0081] To extract arguments from real-time news, a trained argument extraction model (such as the BERTopic model) is used to extract all arguments from the real-time news. The proportion of each argument is then calculated. Semantic analysis is then used to determine the sentiment intensity of the descriptive vocabulary for each argument. The weight of each argument is then calculated by taking a weighted average of the proportion and sentiment intensity. The argument with the highest weight is designated as the core argument. Based on the verification results, credible news viewpoints are then selected. Specifically, news viewpoints whose verification results indicate that the news viewpoint and sentiment are consistent are considered credible. These credible news viewpoints and core arguments are then fed into a trained second text generation model (such as GPT-4) to produce the news analysis segment.

[0082] In practical applications, a trained keyword extraction model (such as TF, PageRank, LDA, etc.) can be used to extract the first keyword of the real-time news, and the association direction is determined according to the user's ability. The user's comprehensive ability is divided into primary, intermediate and advanced levels. The association direction corresponding to the primary comprehensive ability is similar topics in the same field, the association direction corresponding to the intermediate comprehensive ability is social topics, and the association direction corresponding to the advanced comprehensive ability is multi-field topics. The first keyword is associated according to the association direction to obtain the associated topic. For example, if the first keyword is "unmanned driving", the association direction corresponding to the primary comprehensive ability can be drones, the association direction corresponding to the intermediate comprehensive ability can be economics, which can guide the user to think about the impact of unmanned driving on the economy, and the association direction corresponding to the advanced comprehensive ability can be economics plus education, which can guide the user to extend from "unmanned driving" to educational topics such as "skill learning", and from "skill learning" to economic issues such as "job market matching". A trained third text generation model (such as GPT-4) is used to generate a news guide segment based on the first keyword and the associated topic, so that after reading the end of the article, the user will naturally trigger thinking, guide the transfer of views, and prepare for subsequent expression training.

[0083] Specifically, adjusting the initial article according to the user's ability to obtain a training article includes:

[0084] identifying elementary vocabulary, intermediate vocabulary, and advanced vocabulary of the initial article, and calculating the intermediate vocabulary ratio and the advanced vocabulary ratio;

[0085] replacing at least one of elementary vocabulary, intermediate vocabulary, and advanced vocabulary according to the vocabulary ability until the proportion of intermediate vocabulary reaches a first vocabulary threshold corresponding to the vocabulary ability and the proportion of advanced vocabulary reaches a second vocabulary threshold corresponding to the vocabulary ability, thereby obtaining an intermediate article;

[0086] identifying simple sentence patterns, intermediate sentence patterns, and high-order sentence patterns of the first intermediate article, and calculating the proportion of intermediate sentences and the proportion of high-order sentences;

[0087] At least one of the simple sentence patterns, intermediate sentence patterns and high-order sentence patterns is adjusted according to the grammatical ability until the proportion of intermediate sentences reaches a first grammatical threshold corresponding to the grammatical ability and the proportion of high-order sentences reaches a second grammatical threshold corresponding to the grammatical ability, thereby obtaining a training article.

[0088] In actual application, after identifying the primary, intermediate and advanced vocabulary in the initial article, the number of primary, intermediate and advanced vocabulary in the initial article is counted, the intermediate vocabulary in the initial article is divided by the sum of the number of primary, intermediate and advanced vocabulary to obtain the intermediate vocabulary ratio, and the advanced vocabulary in the initial article is divided by the sum of the number of primary, intermediate and advanced vocabulary to obtain the advanced vocabulary ratio. Then, according to the user's vocabulary ability, at least one of the primary, intermediate and advanced vocabulary is replaced. When the vocabulary ability is primary, the intermediate vocabulary ratio reaches a first vocabulary threshold corresponding to the primary vocabulary ability, and the advanced vocabulary ratio reaches a second vocabulary threshold corresponding to the primary vocabulary ability. When the vocabulary ability is intermediate, the intermediate vocabulary ratio reaches a first vocabulary threshold corresponding to the intermediate vocabulary ability, and the advanced vocabulary ratio reaches a second vocabulary threshold corresponding to the intermediate vocabulary ability. When the vocabulary ability is advanced, the intermediate vocabulary ratio reaches a first vocabulary threshold corresponding to the advanced vocabulary ability, and the advanced vocabulary ratio reaches a second vocabulary threshold corresponding to the advanced vocabulary ability. The pre-set simple sentence patterns are direct sentences with subject, predicate and object (such as "AI changes jobs."), the intermediate sentence patterns are attributive clauses, causal sentences and concession sentences (such as "AI, which is widely adopted, reshapes jobs."), and the high-level sentence patterns are inversion sentences, subjunctive mood and complex sentences (such as "Only by adapting can we survive theAIera."), count the number of simple sentence patterns, intermediate sentence patterns and high-order sentence patterns in the initial article, divide the number of intermediate sentence patterns by the sum of the number of simple sentence patterns, intermediate sentence patterns and high-order sentence patterns to obtain the proportion of intermediate sentences, divide the number of high-order sentence patterns by the sum of the number of simple sentence patterns, intermediate sentence patterns and high-order sentence patterns to obtain the proportion of high-order sentences, and then adjust at least one of the simple sentence patterns, intermediate sentence patterns and high-order sentence patterns according to the user's grammatical ability. When the grammatical ability is elementary, the proportion of intermediate sentences reaches the first grammatical threshold corresponding to the elementary grammatical ability, and the proportion of high-order sentences reaches the second grammatical threshold corresponding to the elementary grammatical ability. When the grammatical ability is intermediate, the proportion of intermediate sentences reaches the first grammatical threshold corresponding to the intermediate grammatical ability, and the proportion of high-order sentences reaches the second grammatical threshold corresponding to the intermediate grammatical ability. When the grammatical ability is advanced, the proportion of intermediate sentences reaches the first grammatical threshold corresponding to the advanced grammatical ability, and the proportion of high-order sentences reaches the second grammatical threshold corresponding to the advanced grammatical ability, so as to obtain training articles, so that the training articles are adjusted according to the user's ability.

[0089] Step S3: Generate training questions based on the training article and user capabilities, and send the training questions to the user.

[0090] In practical applications, the introduction of user ability generation corresponding training questions enables the generated training questions to provide targeted training for users, making the training process more accurate and efficient.

[0091] Specifically, generating training questions based on the training article and user capabilities includes:

[0092] Extracting the core idea of the training article, selecting a corresponding first question template according to the comprehensive ability, substituting the core idea into the corresponding first question template to obtain a first training question;

[0093] Determine non-core viewpoints based on the core viewpoints, select a corresponding second question template based on the comprehensive ability, substitute the non-core viewpoints into the corresponding second question template to obtain a second training question;

[0094] extracting a second keyword from the training article, determining a transfer direction based on the comprehensive ability, determining a transfer topic based on the second keyword and the transfer direction, and generating a third training question based on the second keyword and the transfer topic;

[0095] The number of the second training questions is adjusted according to the logical ability, and the number of the third training questions is adjusted according to the migration ability. The first training questions, all the second training questions and all the third training questions constitute the training questions of the training article.

[0096] In practice, training articles are generated from real-time news. Opinion analysis of real-time news is performed to obtain each subject's opinion on the news event. The proportion of each opinion is calculated, and the opinion with the largest proportion is used as the core opinion of the training article. When the user's comprehensive ability is at the elementary level, the first question template corresponding to elementary comprehensive ability is selected, such as "Why do you think X?". The core opinion is substituted into the corresponding first question template, replacing X with the core opinion. This generates the first training question, which directly prompts the user to confirm and restate the core opinion to ensure basic expression. When the user's comprehensive ability is at the intermediate level, the first question template corresponding to intermediate comprehensive ability is selected, such as "What are X1 of X2 compared to X3?", where X1, X2, and X3 represent the keywords in the core opinion X. The keywords of the core opinion are substituted into the corresponding first question template to generate the first training question. Complex expression is then added to guide the user. When the user's comprehensive ability is at the advanced level, the first question template corresponding to advanced comprehensive ability is selected, such as "How do X2 contribute to X1, and are there any potential drawbacks?". The keywords of the core opinion are substituted into the corresponding first question template to generate the first training question, guiding the user to express from multiple perspectives.

[0097] To determine non-core viewpoints based on core viewpoints, viewpoints other than the core viewpoint in the news can be considered non-core viewpoints. If the news viewpoints only involve the core viewpoint, semantic analysis is used to find opposing viewpoints from news databases (such as BBC, CNN, and Reuters), or the opinions of industry experts or academics are supplemented by using knowledge bases to obtain non-core viewpoints. When the user's comprehensive ability is elementary, the second question template corresponding to elementary comprehensive ability is selected. When the user's comprehensive ability is intermediate, the second question template corresponding to intermediate comprehensive ability is selected. When the user's comprehensive ability is advanced, the second question template corresponding to advanced comprehensive ability is selected. The non-core viewpoint is substituted into the corresponding second question template to obtain the second training question. For example, the second training question corresponding to elementary comprehensive ability is "List thress reasons why people still prefer gasoline cars." Ask the user to briefly list the reasons for choosing non-core viewpoints. The second training question corresponding to intermediate comprehensive ability is "Compare the costs and convenience of electric cars versus gasoline cars." This guides the user to compare the advantages and disadvantages of the core viewpoint and non-core viewpoints. The third training question corresponding to advanced comprehensive ability is "Some people argue that electric cars are more expensive and less reliable. Do you agree? Why or why?" not?", guiding users to refute non-core opinions.

[0098] To extract the second keyword from the training article, a trained keyword extraction model is used to extract keywords from the training article to obtain the second keyword. When the user's comprehensive ability is at the elementary level, the migration direction of the elementary comprehensive ability is related topics in the same field. When the user's comprehensive ability is at the intermediate level, the migration direction of the intermediate comprehensive ability is social topics. When the user's comprehensive ability is at the advanced level, the migration direction of the intermediate comprehensive ability is multi-field topics. The second keyword is migrated according to the migration direction to obtain a migration topic. For example, if the second keyword is "electric cars," the migration directions corresponding to the elementary comprehensive ability can be "solar energy" or "hydrogen fuel," and the migration topic is green technology. The migration directions corresponding to the intermediate comprehensive ability can be "policy incentives" or "economic impact," and the migration topic is social impact. The migration directions corresponding to the advanced comprehensive ability can be "urban planning" or "power grid transformation," and the migration topic is cross-field impact. A trained training question generation model (such as BERT) is used to generate a third training question based on the second keyword and the migration topic. For example, the third training question corresponding to primary comprehensive ability could be “Describe an eco-friendly technology you know, besides electric cars.” The third training question corresponding to intermediate comprehensive ability could be “How can sustainable energy influence other industries, like aviation or construction?” The third training question corresponding to advanced comprehensive ability could be “If electric cars were widely adopted globally, how might it reshape the energy market and urban planning?”

[0099] The number of the second training questions is adjusted by the logical ability, and the number of the third training questions is adjusted according to the migration ability. If the logical ability is lower than the logical lower limit threshold, the number of the second training questions is increased; if the logical ability is lower than the logical upper limit threshold, the number of the second training questions is reduced; if the migration ability is lower than the migration lower limit threshold, the number of the third training questions is increased; if the migration ability is lower than the migration upper limit threshold, the number of the third training questions is reduced. The training direction can be dynamically adjusted according to the user's ability, and training can be strengthened for the user's weaknesses.

[0100] Step S4: Receive user responses from user feedback, and update user capabilities based on the user responses.

[0101] Specifically, the user answers include: a first answer, a second answer, and a third answer, the first answer is an answer to the first training question, the second answer is an answer to the second training question, and the third answer is an answer to the third training question, and updating the user capabilities according to the user answers includes:

[0102] Counting a first vocabulary size of all words in the user's answer and a second vocabulary size of all correct words, and dividing the second vocabulary size by the first vocabulary size to obtain a vocabulary score for the user;

[0103] Counting the number of first sentences of all sentences in the user's answer and the number of all grammatically correct second sentences, and dividing the number of the second sentences by the first sentences to obtain a grammar score of the user;

[0104] Counting the first logical quantities of all logical points in the training article, counting the second logical quantities of all logically correct answers of the user, and dividing the second logical quantity by the first logical quantity to obtain the user's logic score;

[0105] Counting all first opinion values in the first and second answers, counting all second opinion values in the third answer, and dividing the second opinion values by the first opinion values to obtain a user migration score;

[0106] The vocabulary score, grammar score, logic score and migration score are weighted and calculated to obtain a comprehensive score of the user;

[0107] The user's ability is updated according to the vocabulary score, grammar score, logic score, transfer score and comprehensive score.

[0108] In practical applications, the first vocabulary size is calculated by counting all words in the first, second, and third answers, the second vocabulary size is calculated by counting all correct words in the first, second, and third answers, and the second vocabulary size is divided by the first vocabulary size to obtain the user's trained vocabulary score. The first sentence size is calculated by counting all sentences in the first, second, and third answers, and the second sentence size is calculated by counting all grammatically correct sentences in the first, second, and third answers to obtain the user's trained grammar score. The vocabulary score and grammar score can be used to measure a user's basic English expression ability. A trained logical extraction model (such as LDA or BERTopic) is used to extract corresponding logical points from training articles and news databases based on the training questions. The number of extracted logical points is then counted to obtain a first logical quantity. The first, second, and third answers are then input into the logical extraction model to obtain second logical quantities corresponding to the first answer, the second answer, and the third answer. The second logical quantity is divided by the corresponding first logical quantity to obtain the user's logical score, which is used to determine whether the views expressed in the user's answer are complete, clear, and coherent. The trained opinion extraction model is used to extract opinions from the first answer, the second answer, and the third answer. The sum of the opinion volume of the first answer and the opinion volume of the second answer is counted to obtain the first opinion volume. The opinion volume of the third answer is counted to obtain the second opinion volume. The second opinion volume is divided by the first opinion volume to obtain the user's migration score, which is used to examine whether the user can migrate from one topic to another similar topic.

[0109] The user's comprehensive score is obtained by weighting and calculating the vocabulary score, grammar score, logic score and migration score. The comprehensive score calculation formula is:

[0110] W total =α*W word +β*W grammar +γ*W logic +δ*W transfer

[0111] Wherein, the W total Indicates the comprehensive score, W word represents the vocabulary score, W grammar represents the grammatical score, W logic represents the logical score, W transfer The weight coefficients are adjusted based on the user's overall score. As the overall score increases, the weights of α and β gradually decrease, while the weights of γ and δ gradually increase. As the overall score decreases, the weights of α and β gradually increase, while the weights of γ and δ gradually decrease. This dynamically optimizes the training content based on the user's current level, making the learning process more accurate and efficient.

[0112] Step S5: Determine whether to continue training based on the updated user ability.

[0113] In actual application, the updated user ability is compared with the preset learning goal to determine whether the user ability reaches the learning goal. If so, the training is stopped and a training completion message is sent to the user. If not, the user continues to be trained and returns to step S2.

[0114] The English training method of the present invention generates training articles and training questions based on real-time news to improve English proficiency, allowing users to understand the content of real-time news while maintaining the timeliness and interest of the training content. It also updates user abilities in real time, adjusts training articles and training questions based on user abilities, and dynamically optimizes training content, making the learning process more accurate and efficient.

[0115] like Figure 2 As shown, the embodiment of the present invention further provides an English training system, including:

[0116] A capability assessment module 10, for initially assessing user capabilities;

[0117] An article construction module 20 is configured to obtain real-time news, construct an initial article based on the real-time news and user capabilities, and adjust the initial article based on the user capabilities to obtain a training article;

[0118] A training question generating module 30 is used to generate training questions based on the training article and the user's ability, and send the training questions to the user;

[0119] A capability updating module 40 is configured to receive user responses to user feedback and update user capabilities based on the user responses;

[0120] The training judgment module 50 is used to judge whether to continue training based on the updated user ability.

[0121] Each module of the English training system described above can be implemented in whole or in part through software, hardware, or a combination thereof. Each of the modules and units described above can be embedded in or independent of a processor in a computer device in hardware form, or can be stored in a memory in a computer device in software form, so that the processor can call and execute the corresponding operations of each module.

[0122] like Figure 3 As shown, an embodiment of the present invention discloses a computer device, including a memory and a processor, wherein the memory stores a computer program;

[0123] The computer device may be a server, and its internal structure diagram may be as follows: Figure 3As shown. The computer device includes a processor, a memory, a network interface and a database connected via a system bus. The processor of the computer device is used to provide computing and control capabilities. The memory of the computer device includes a non-volatile storage medium and an internal memory. The non-volatile storage medium stores an operating system, a computer program and a database. The internal memory provides an environment for the operation of the operating system and computer program in the non-volatile storage medium. The network interface of the computer device is used to communicate with an external terminal via a network connection. When the computer program is executed by the processor, the English training method described in the above embodiments is implemented.

[0124] Those skilled in the art will understand that Figure 3 The structure shown in the figure is only a block diagram of a part of the structure related to the solution of the present application, and does not constitute a limitation on the computer device to which the solution of the present application is applied. The specific computer device may include more or fewer components than shown in the figure, or combine certain components, or have a different component arrangement.

[0125] An embodiment of the present invention further discloses a computer-readable storage medium storing a computer program, wherein the computer program enables a computer to execute the English training method described in the above embodiments.

[0126] Those skilled in the art will appreciate that all or part of the processes in the above-mentioned embodiments can be implemented by instructing the relevant hardware through a computer program. The computer program can be stored in a non-volatile computer-readable storage medium. When the computer program is executed, it can include the processes of the embodiments of the above-mentioned methods. Among them, any reference to memory, storage, database or other media used in the embodiments provided in this application can include non-volatile and / or volatile memory. Non-volatile memory can include read-only memory (ROM), programmable ROM (PROM), electrically programmable ROM (EPROM), electrically erasable programmable ROM (EEPROM) or flash memory. Volatile memory can include random access memory (RAM) or external cache memory. By way of illustration and not limitation, RAM is available in various forms, such as static RAM (SRAM), dynamic RAM (DRAM), synchronous DRAM (SDRAM), double data rate SDRAM (DDRSDRAM), enhanced SDRAM (ESDRAM), synchronous link (Synchlink) DRAM (SLDRAM), memory bus (Rambus) direct RAM (RDRAM), direct memory bus dynamic RAM (DRDRAM), and memory bus dynamic RAM (RDRAM).

[0127] The technical features of the above embodiments can be combined arbitrarily. To make the description concise, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.

Claims

1. An English training method, characterized in that: include: Initial assessment of user capabilities; Acquire real-time news, construct an initial article based on the real-time news and user capabilities, and adjust the initial article based on the user capabilities to obtain a training article; Generate training questions based on the training articles and user capabilities, and send the training questions to the user; receiving user responses from user feedback, and updating user capabilities based on the user responses; Determine whether to continue training based on the updated user capabilities.

2. The English training method according to claim 1, wherein The initial article is constructed according to the real-time news and user capabilities, including: Extracting basic information of the real-time news and generating news background segments based on the basic information; Performing full text analysis on the real-time news, and generating news analysis segments based on the full text analysis results; A first keyword of the real-time news is extracted, a related topic is determined according to the first keyword and user capability, and a news guide segment is generated according to the first keyword and the related topic.

3. The English training method according to claim 2, wherein: The step of extracting basic information of the real-time news and generating a news background segment based on the basic information includes: Extracting entities and causal relationships of the real-time news to obtain basic information; Match topic categories according to the basic information to obtain news categories corresponding to the real-time news; A news background segment is generated according to the basic information and news category.

4. The English training method according to claim 2, wherein: The full-text analysis of the real-time news and the generation of news analysis segments based on the full-text analysis results include: Performing opinion analysis on the real-time news to obtain news opinions of various subjects on the news event; Performing sentiment analysis on the real-time news to obtain the sentiment inclination of each subject towards the news event, and verifying the news opinion of each subject based on the sentiment inclination of each subject to obtain a verification result; Extract all arguments of the real-time news, calculate the weight of each argument based on its proportion and sentiment intensity, and take the argument with the highest weight as the core argument; Generate a news analysis segment based on the news viewpoints, verification results and core arguments.

5. The English training method according to claim 1, wherein: The user's abilities include vocabulary ability and grammatical ability; and adjusting the initial article according to the user's abilities to obtain a training article includes: identifying elementary vocabulary, intermediate vocabulary, and advanced vocabulary of the initial article, and calculating the intermediate vocabulary ratio and the advanced vocabulary ratio; replacing at least one of elementary vocabulary, intermediate vocabulary, and advanced vocabulary according to the vocabulary ability until the proportion of intermediate vocabulary reaches a first vocabulary threshold corresponding to the vocabulary ability and the proportion of advanced vocabulary reaches a second vocabulary threshold corresponding to the vocabulary ability, thereby obtaining an intermediate article; identifying simple sentence patterns, intermediate sentence patterns, and high-order sentence patterns of the intermediate article, and calculating the proportion of intermediate sentences and the proportion of high-order sentences; At least one of the simple sentence patterns, intermediate sentence patterns and high-order sentence patterns is adjusted according to the grammatical ability until the proportion of intermediate sentences reaches a first grammatical threshold corresponding to the grammatical ability and the proportion of high-order sentences reaches a second grammatical threshold corresponding to the grammatical ability, thereby obtaining a training article.

6. The English training method according to claim 1, wherein: The user abilities include: logical ability, transfer ability and comprehensive ability; the training questions generated according to the training articles and user abilities include: Extracting the core idea of the training article, selecting a corresponding first question template according to the comprehensive ability, substituting the core idea into the corresponding first question template to obtain a first training question; Determine non-core viewpoints based on the core viewpoints, select a corresponding second question template based on the comprehensive ability, substitute the non-core viewpoints into the corresponding second question template to obtain a second training question; Extracting a second keyword from the training article, determining a transfer direction and a third question template based on the comprehensive ability, determining a transfer topic based on the second keyword and the transfer direction, and generating a third training question based on the second keyword and the transfer topic; The number of the second training questions is adjusted according to the logical ability, and the number of the third training questions is adjusted according to the migration ability. The first training questions, all the second training questions and all the third training questions constitute the training questions of the training article.

7. The English training method according to claim 6, wherein: The user answers include: a first answer, a second answer, and a third answer, wherein the first answer is an answer to the first training question, the second answer is an answer to the second training question, and the third answer is an answer to the third training question. Updating the user capability according to the user answers includes: Counting a first vocabulary size of all words in the user's answer and a second vocabulary size of all correct words, and dividing the second vocabulary size by the first vocabulary size to obtain a vocabulary score for the user; Counting the number of first sentences of all sentences in the user's answer and the number of all grammatically correct second sentences, and dividing the number of the second sentences by the first sentences to obtain a grammar score of the user; Generate a corresponding first logical quantity according to each training question, count all logically correct second logical quantities in each answer of the user, and divide the second logical quantity by the corresponding first logical quantity to obtain the user's logic score; Counting all first opinion values in the first and second answers, counting all second opinion values in the third answer, and dividing the second opinion values by the first opinion values to obtain a user migration score; The vocabulary score, grammar score, logic score and migration score are weighted and calculated to obtain a comprehensive score of the user; The user's ability is updated according to the vocabulary score, grammar score, logic score, transfer score and comprehensive score.

8. English training system, characterized in that, include: Capability assessment module, used for initial assessment of user capabilities; An article construction module is used to obtain real-time news, construct an initial article based on the real-time news and user capabilities, and adjust the initial article based on the user capabilities to obtain a training article; A training question generating module, configured to generate training questions based on the training articles and user capabilities, and send the training questions to the user; A capability updating module, configured to receive user responses to user feedback and update user capabilities based on the user responses; The training judgment module is used to determine whether to continue training based on the updated user capabilities.

9. A computer device comprising a memory and a processor, wherein the memory stores a computer program, characterized in that: When the processor executes the computer program, the steps of the method according to any one of claims 1 to 7 are implemented.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the steps of the method according to any one of claims 1 to 7 are implemented.

Images