Intelligent English word learning and translating system and method

By obtaining and analyzing the user's English word learning data, conducting context classification and learning trend analysis, the problem of insufficient targetedness and accuracy of auxiliary translation in the existing technology is solved, and personalized and intelligent English word learning is achieved. Translation tips.

CN118940772AActive Publication Date: 2024-11-12SHANDONG SHUNSHI EDUCATION TECH GRP CO LTD

Patent Information

Application Number
CN202411136927.8
Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Filing Date
2024-08-19
Publication Date
2024-11-12
Estimated Expiration
2044-08-19

AI Technical Summary

Technical Problem

The existing English word learning system is not targeted and accurate in the translation process, and it is impossible to provide intelligent translation tips based on the learner's actual level and context.

Method used

By obtaining the target words and context information of the user's current learning, indexing historical learning data, conducting context classification, optimizing dimensionality reduction, learning trend analysis and correcting learning rate, and providing personalized translation tips.

Benefits of technology

It improves the pertinence and accuracy of auxiliary translation, and provides personalized and intelligent English word learning and translation tips.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN118940772A_ABST
    Figure CN118940772A_ABST
Patent Text Reader

Abstract

The invention discloses an English word intelligent learning translation system and method, and relates to the related field of data processing, and the system comprises a historical learning data index module which is used for obtaining a target word, collecting target context information, and indexing historical learning data; the historical context information classification module is used for classifying the historical context information to obtain a historical learning score set; the historical learning score set optimization dimensionality reduction module is used for optimization dimensionality reduction and arrangement according to a time sequence; the learning trend analysis module is used for performing learning trend analysis according to the historical learning score sequence; the correction module is used for matching historical contexts and correcting the learning rate and the degradation rate; and the auxiliary translation prompting module is used for performing auxiliary translation prompting. The technical problem that existing English word learning translation is insufficient in pertinence and accuracy of aided translation is solved, and the technical effect of improving the pertinence and accuracy of aided translation is achieved.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The present application relates to data processing related fields, and in particular to an English word intelligent learning and translation system and method. Background Art

[0002] As globalization deepens, the learning and application of English as an international common language has become particularly important. Traditional English learning methods often focus on vocabulary memorization and grammar learning, but in actual language use, the understanding and application of words are highly dependent on specific contexts. Therefore, how to intelligently learn and translate English words based on context has become the key to improving the efficiency and quality of English learning. Existing English word learning systems mostly use static vocabulary lists or simple examples to assist learning. This method often provides unified learning materials or basic vocabulary replacements during the translation process, and lacks the ability to provide intelligent translation prompts based on the learner's actual level and context, resulting in insufficient auxiliary translation capabilities.

[0003] In the current related technologies, there are technical problems in the auxiliary translation such as insufficient pertinence and accuracy in English word learning and translation. Summary of the invention

[0004] The present application provides an English word intelligent learning and translation system and method, which adopts the following technical means: obtaining the target word currently being learned by the user and its context information, indexing the user's historical learning data of the target word, classifying the historical context information, extracting word meaning information and learning data in different contexts, processing the learning score data by optimizing the dimensionality reduction technology, reducing the dimension and arranging it in time series, analyzing the learning trend according to the historical learning score sequence, obtaining the learning rate and degradation rate, correcting the learning rate and degradation rate according to the deviation between the target context and the matching historical context, making auxiliary translation decisions according to the corrected learning rate and degradation rate, providing personalized translation prompt solutions, etc., to realize personalized and intelligent English word learning and translation prompts, and achieve the technical effect of improving the pertinence and accuracy of auxiliary translation.

[0005] The present application provides an English word intelligent learning and translation system, comprising: A historical learning data indexing module, the historical learning data indexing module is used to obtain the target word that the user is currently learning, and collect the target context information of the target word currently located, and index the historical learning data of the target word within the user's historical time according to the target word; a historical context information classification module, the historical context information classification module is used to classify the historical context information in the historical learning data, obtain multiple classified historical contexts, extract historical word meaning information sets and multiple historical learning data sets under the multiple classified historical contexts, and process to obtain multiple historical learning score sets; a historical learning score set optimization and dimensionality reduction module, the historical learning score set optimization and dimensionality reduction module is used to optimize and reduce the dimensions of the multiple historical learning score sets respectively, and obtain multiple a reduced-dimension historical learning score set, and arrange them in time sequence to obtain multiple historical learning score sequences; a learning trend analysis module, the learning trend analysis module is used to perform learning trend analysis according to the multiple historical learning score sequences to obtain a learning rate and a degradation rate; a correction module, the correction module is used to match the target context information with the multiple classified historical contexts to obtain a matching historical context, and correct the learning rate and the degradation rate according to the deviation between the target context information and the standard matching context information of the matching historical context to obtain a corrected learning rate and a corrected degradation rate; an auxiliary translation prompt module, the auxiliary translation prompt module is used to make an auxiliary translation decision according to the corrected learning rate and the corrected degradation rate, obtain an auxiliary translation prompt scheme, and perform an auxiliary translation prompt.

[0006] In a possible implementation, the target word that the user is currently learning is obtained, and the target context information of the target word is currently collected. According to the target word, the historical learning data of the target word within the user's historical time is indexed, and the following processing is performed: Obtain the target word that the user is currently learning, collect the upper context content and lower context content of the target word as target context information; according to the target word, index the historical learning record of the user to obtain the historical learning data of the target word.

[0007] In a possible implementation, the historical context information in the historical learning data is classified to obtain a plurality of classified historical contexts, a historical word meaning information set and a plurality of historical learning data sets under the plurality of classified historical contexts are extracted, and a plurality of historical learning score sets are obtained by processing, and the following processing is performed: Acquire multiple historical context information in the historical learning data, perform classification processing, and obtain multiple classified historical contexts; extract the standard meanings of target words in the historical learning data under the multiple classified historical contexts to obtain a historical meaning information set, and extract multiple historical translation semantic information sets and multiple historical translation submission time sets translated and submitted by users under the multiple classified historical contexts; classify and obtain multiple historical basic learning score sets according to the deviation ranges between the multiple historical translation semantic information sets and the historical meaning information sets; perform correction calculations on the multiple historical basic learning score sets according to the ratio of a preset submission time threshold and the multiple historical translation submission time sets to obtain multiple historical learning score sets.

[0008] In a possible implementation, the multiple historical learning score sets are optimized and reduced in dimension respectively to obtain multiple reduced-dimension historical learning score sets, and the sets are arranged in time sequence to obtain multiple historical learning score sequences, and the following processing is performed: In the first historical learning score set among the multiple historical learning score sets, a first benchmark learning score is selected to be obtained; according to the difference between other historical learning scores in the first historical learning score set and the first benchmark learning score, a distribution probability is allocated to obtain a first basic probability distribution, wherein the size of the difference is negatively correlated with the size of the distribution probability; according to the first basic probability distribution, the first historical learning score set is optimized and reduced in dimension to obtain a first reduced-dimension learning score set; according to the timestamp information of multiple historical learning scores in the first reduced-dimension historical learning score set, a first historical learning score sequence is obtained; and the other multiple historical learning score sets are optimized and reduced in dimension and arranged in time series to obtain multiple historical learning score sequences.

[0009] In a possible implementation, according to the first basic probability distribution, the first historical learning score set is optimized and reduced in dimension to obtain a first reduced-dimension learning score set, and the following processing is performed: Randomly extract a preset number of historical learning scores for dimensionality reduction in the first historical learning score set to obtain a first dimensionality reduction historical learning score set; assign distribution probabilities according to the difference between the historical learning scores in the first dimensionality reduction historical learning score set and the first benchmark learning scores to obtain a first dimensionality reduction probability distribution; calculate the similarity between the first dimensionality reduction probability distribution and the first basic probability distribution as the first dimensionality reduction fitness; randomly extract a preset number of historical learning scores for dimensionality reduction in the first historical learning score set again to obtain a second dimensionality reduction historical learning score set, and process to obtain the second dimensionality reduction fitness; continue to optimize the dimensionality reduction until convergence, and output the dimensionality reduction historical learning score set with the largest dimensionality reduction fitness as the first dimensionality reduction learning score set.

[0010] In a possible implementation, a learning trend analysis is performed according to the plurality of historical learning score sequences to obtain a learning rate and a degradation rate, and the following processing is performed: According to the sample learning data of multiple users, a set of sample learning score sequences is collected, and a set of sample learning rates and a set of sample degradation rates are obtained according to the learning score change identifiers in each sample learning score sequence; the set of sample learning score sequences is used as classification input, and the set of sample learning rates and the set of sample degradation rates are used as classification output to construct a learning trend analyzer; based on the learning trend analyzer, the multiple historical learning score sequences are classified according to their learning trends to obtain multiple context learning rates and multiple context degradation rates; the similarity between the target context information and the multiple classified historical contexts is analyzed, and the multiple context learning rates and the multiple context degradation rates are weightedly calculated according to the magnitude of the multiple context similarities to obtain the learning rate and the degradation rate.

[0011] In a possible implementation, the target context information is matched with the multiple classified historical contexts to obtain a matching historical context, and the learning rate and the degradation rate are corrected according to a deviation between the target context information and the standard matching context information of the matching historical context to obtain a corrected learning rate and a corrected degradation rate, and the following processing is performed: Select the classified historical context with the greatest similarity as the matching historical context, and obtain the standard matching context information of the matching historical context; set a learning correction coefficient according to the deviation between the target context information and the standard matching context information of the matching historical context; use the learning correction coefficient to correct the learning rate and degradation rate to obtain a corrected learning rate and a corrected degradation rate.

[0012] In a possible implementation, according to the modified learning rate and the modified degradation rate, an auxiliary translation decision is made, an auxiliary translation prompting scheme is obtained, an auxiliary translation prompt is made, and the following processing is performed: A set of sample correction learning rates and a set of sample correction degradation rates are collected, and a sample auxiliary translation prompt scheme is set according to the size of each sample correction learning rate and sample correction degradation rate, so as to obtain a set of sample auxiliary translation prompt schemes, wherein each sample auxiliary translation prompt scheme includes a prompt ratio, and the size of the sample correction learning rate and the sample correction degradation rate is negatively correlated with the size of the prompt ratio; the sample correction learning rate set and the sample correction degradation rate set are used as decision inputs, and the sample auxiliary translation prompt scheme set is used as decision output to construct an auxiliary translation decision maker; the auxiliary translation decision maker is used to make auxiliary translation decisions on the correction learning rate and the correction degradation rate, so as to obtain an auxiliary translation prompt scheme.

[0013] The present application also provides an English word intelligent learning and translation method, comprising: The target word currently being learned by the user is obtained, and the target context information of the target word currently located is collected; according to the target word, the historical learning data of the target word within the user's historical time is indexed; the historical context information in the historical learning data is classified to obtain multiple classified historical contexts, and historical word meaning information sets and multiple historical learning data sets under the multiple classified historical contexts are extracted, and multiple historical learning score sets are obtained by processing; the multiple historical learning score sets are respectively optimized and reduced in dimension to obtain multiple reduced-dimension historical learning score sets, and are arranged in time sequence to obtain multiple historical learning score sequences; learning trend analysis is performed according to the multiple historical learning score sequences to obtain a learning rate and a degradation rate; the target context information is matched with the multiple classified historical contexts to obtain a matching historical context, and the learning rate and the degradation rate are corrected according to the deviation between the target context information and the standard matching context information of the matching historical context to obtain a corrected learning rate and a corrected degradation rate; according to the corrected learning rate and the corrected degradation rate, an auxiliary translation decision is made, an auxiliary translation prompt scheme is obtained, and an auxiliary translation prompt is performed.

[0014] The intelligent learning and translation system and method of English words proposed in the present application obtains the target word currently being learned by the user through the historical learning data indexing module, and collects the target context information of the target word currently located. According to the target word, the historical learning data of the target word in the user's historical time is indexed, and the historical context information in the historical learning data is classified through the historical context information classification module to obtain multiple classified historical contexts, and the historical word meaning information sets and multiple historical learning data sets under the multiple classified historical contexts are extracted, and multiple historical learning score sets are processed to obtain multiple historical learning score sets, and the multiple historical learning score sets are optimized and reduced in dimension respectively through the historical learning score set optimization and dimensionality reduction module to obtain multiple reduced dimension historical contexts. A historical learning score set is obtained, and arranged in chronological order to obtain multiple historical learning score sequences. A learning trend analysis module is used to perform learning trend analysis based on the multiple historical learning score sequences to obtain a learning rate and a degradation rate. A correction module is used to match the target context information with multiple classified historical contexts to obtain a matching historical context. The learning rate and the degradation rate are corrected according to the deviation between the target context information and the standard matching context information of the matching historical context to obtain a corrected learning rate and a corrected degradation rate. An auxiliary translation prompt module is used to make an auxiliary translation decision based on the corrected learning rate and the corrected degradation rate to obtain an auxiliary translation prompt scheme and perform auxiliary translation prompts, thereby achieving a technical effect of improving the pertinence and accuracy of auxiliary translation. BRIEF DESCRIPTION OF THE DRAWINGS

[0015] In order to more clearly illustrate the technical solution of the embodiment of the present invention, the accompanying drawings of the embodiment of the present invention will be briefly introduced below. A flow chart is used in the present application to illustrate the operations performed by the system according to the embodiment of the present application. It should be understood that the preceding or following operations are not necessarily performed accurately in order. On the contrary, various steps can be processed in reverse order or simultaneously as needed. At the same time, other operations can also be added to these processes, or one or more operations can be removed from these processes.

[0016] Figure 1 A schematic diagram of the structure of an English word intelligent learning and translation system provided in an embodiment of the present application.

[0017] Figure 2 A flowchart of an intelligent learning and translation method for English words provided in an embodiment of the present application.

[0018] Explanation of the accompanying drawings: historical learning data indexing module 10, historical context information classification module 20, historical learning score set optimization and dimensionality reduction module 30, learning trend analysis module 40, correction module 50, auxiliary translation prompt module 60. DETAILED DESCRIPTION

[0019] The above description is only an overview of the technical solution of the present application. In order to more clearly understand the technical means of the present application, it can be implemented in accordance with the contents of the specification. In order to make the above and other purposes, features and advantages of the present application more obvious and easy to understand, the specific implementation methods of the present application are listed below.

[0020] In order to make the purpose, technical solutions and advantages of the present application clearer, the present application will be further described in detail below in conjunction with the accompanying drawings. The described embodiments should not be regarded as limiting the present application. All other embodiments obtained by ordinary technicians in the field without making creative work are within the scope of protection of this application.

[0021] In the following description, reference is made to "some embodiments", which describe a subset of all possible embodiments, but it is understood that "some embodiments" may be the same subset or different subsets of all possible embodiments, and may be combined with each other without conflict, and the terms "first\second" involved are merely to distinguish similar objects and do not represent a specific ordering of objects. The terms "including" and "having" and any variations are intended to cover non-exclusive inclusions, for example, a process, method, system, product, or server that includes a series of steps or units is not necessarily limited to those steps or units clearly listed, but may include other steps or modules that are not clearly listed or inherent to these processes, methods, products, or devices. Unless otherwise defined, all technical and scientific terms used herein have the same meaning as those generally understood by technicians in the technical field of this application. The terms used herein are for the purpose of describing the embodiments of the present application only.

[0022] The present application embodiment provides an English word intelligent learning and translation system, such as Figure 1 As shown, the system comprises: The historical learning data indexing module 10 is used to obtain the target word that the user is currently learning, and to collect the target context information of the target word. According to the target word, the historical learning data of the target word in the user's historical time is indexed. Specifically, the user inputs or selects a target word through an interface (such as an input box, clicking a word list, etc.), that is, the target word is the English word that the user is currently learning or searching for, and is the starting point for searching and indexing. The user's current environment is detected or the user is required to manually input the context information of the target word. The target context information is the specific context of the target word, which can be a sentence, paragraph, article or any text fragment containing the target word, which is used to understand the specific meaning and usage of the target word. The collected target context information is stored as metadata associated with the target word.

[0023] According to the target word, a query is initiated in the user's personal learning database, which contains all the user's previous learning records of the target word (and other words). The query request may include strategies such as exact matching of words, fuzzy matching (such as considering different forms or spelling errors of words), etc. After all historical learning records related to the target word are retrieved in the personal learning database, these records are indexed. The index can be based on multiple dimensions, such as the user's historical translation results (such as correct / wrong, score, etc.), the time taken for translation, and the historical context (such as reading, listening, writing practice, etc.). The most relevant or useful historical learning data for the target word can be quickly extracted and sorted through the index.

[0024] In one possible implementation, a target word that the user is currently learning is obtained, and target context information of the target word is currently collected, and based on the target word, historical learning data of the target word within the user's historical time is indexed, including: obtaining the target word that the user is currently learning, collecting the upper context content and the lower context content of the target word as target context information; based on the target word, indexing the user's historical learning record to obtain the historical learning data of the target word.

[0025] Specifically, the user selects or starts to learn a specific word on the learning platform or application. The system captures this selection through interface interaction and records the word as the target word for current learning. Analyze the context in which the target word appears in the user's current learning environment, extract the text content before the target word as the upper context (used to understand the pre-information and meaning of the target word in the sentence or paragraph), extract the text content after the target word as the lower context (used to understand the subsequent information and contextual background of the target word), and merge these two parts of content to form the complete target context information of the target word. Using the target word as a keyword, search in the user's historical learning database to retrieve all historical learning records containing the target word (records of all learning activities generated by the user on the learning platform in the past), including learning time, number of learning times, learning results (such as test scores), usage scenarios and other information. This implementation method helps the system understand the user's current learning needs more accurately by collecting the context information of the target word, and achieves the technical effect of improving the accuracy of subsequent context matching.

[0026] The historical context information classification module 20 is used to classify the historical context information in the historical learning data, obtain multiple classified historical contexts, extract historical word meaning information sets and multiple historical learning data sets under the multiple classified historical contexts, and process to obtain multiple historical learning score sets. Specifically, the context information in each record is extracted from the historical learning data. These context information are sentences, paragraphs or article fragments containing the target word. The extracted context information is pre-processed, including removing irrelevant characters, standardizing punctuation marks, restoring word forms (such as restoring different tenses and voices of words to basic forms), etc., and applying natural language processing (NLP) technology, such as text similarity algorithms (such as cosine similarity, Jaccard similarity, etc.), to evaluate the similarity between different context information. According to the evaluation results of contextual similarity, the historical context information is classified into multiple classified historical contexts, where the classification standard can be hard clustering based on similarity threshold or soft clustering based on probability distribution. Each classified historical context represents a group of contextual information that is similar in content, theme or emotional color.

[0027] For each classified historical context, all word meaning information of the target word is extracted with the help of dictionary resources or context-based word sense disambiguation technology, and the word meaning information in each classified historical context is organized into a historical word meaning information set. In addition to word meaning information, relevant historical learning data, such as translation results, translation time, etc., are extracted from each classified historical context, and these data are organized into multiple historical learning data sets, each set corresponding to a classified historical context. Based on the historical learning data set in each classified historical context, the learning score is calculated. The learning score is a quantitative indicator to measure the user's learning and translation effectiveness of the target word in a specific context. It can be calculated based on multiple factors, such as translation accuracy, translation time, etc. The calculated learning score is organized into multiple historical learning score sets, each set represents the learning and translation effectiveness in a classified historical context.

[0028] In a possible implementation, the historical context information in the historical learning data is classified to obtain multiple classified historical contexts, historical word meaning information sets and multiple historical learning data sets under the multiple classified historical contexts are extracted, and multiple historical learning score sets are processed to obtain multiple historical learning score sets, including: obtaining multiple historical context information in the historical learning data, classifying them, and obtaining multiple classified historical contexts; extracting the standard meanings of target words in the historical learning data under the multiple classified historical contexts to obtain a historical word meaning information set, and extracting multiple historical translation semantic information sets and multiple historical translation submission time sets translated and submitted by users under the multiple classified historical contexts; classifying and obtaining multiple historical basic learning score sets according to the deviation ranges between the multiple historical translation semantic information sets and the historical word meaning information set; and correcting and calculating the multiple historical basic learning score sets according to the ratio of a preset submission time threshold and the multiple historical translation submission time sets to obtain multiple historical learning score sets.

[0029] Specifically, the context in each historical learning data is parsed to identify the key elements in the context (such as themes, scenes, dialogue objects, etc.). Based on the similarity of the context, clustering algorithms (such as K-means, hierarchical clustering, etc.) are used to classify the historical context information into different groups, each of which represents a classified historical context. For each classified historical context, the standard meaning of the target word in the context is determined, that is, the correct or expected meaning of the target word that is generally considered in a specific context (there may be multiple standard meanings, depending on the complexity of the context), and the standard meanings in each classified historical context are collected to form a historical meaning information set. The semantic information of each translation attempt (the meaning of the word or phrase given during the translation attempt) is extracted from the user's historical translation records, and the submission time of each translation attempt is recorded. The translation semantics and historical translation submission time in the same classified historical context are organized into sets. Calculate the degree of deviation between the semantic information of each translation attempt and the standard word meaning (a similarity algorithm such as cosine similarity can be used), assign a basic learning score to each translation attempt based on the degree of deviation (the smaller the deviation, the higher the score), and organize the basic learning scores under the same classification historical context into a set. Preset a submission time threshold to indicate the time within which the translation result is considered acceptable, calculate the ratio of the preset submission time threshold to the submission time of each translation attempt, and correct the basic learning score based on the time ratio (for example, the smaller the ratio, the longer the translation time, and the lower the corrected score), and update the corrected learning score to the corresponding historical learning score set. This implementation method compares the standard word meaning with the historical translation semantics, reflects the user's learning translation effect on the target word, and corrects the learning translation effect through the translation submission time, which more accurately reflects the user's mastery of the target word, achieving the technical effect of realizing accurate quantitative evaluation of the user's mastery of the target word.

[0030] The historical learning score set optimization and dimensionality reduction module 30 is used to optimize and reduce the dimensions of the multiple historical learning score sets respectively, obtain multiple reduced-dimensional historical learning score sets, and arrange them in time series to obtain multiple historical learning score sequences. Specifically, according to the characteristics of the historical learning score set (such as the size of the data, score distribution, etc.), a specific optimization and dimensionality reduction method is applied, such as principal component analysis (PCA), linear discriminant analysis (LDA), t-SNE (t-distributed Stochastic Neighbor Embedding), etc., through dimensionality reduction processing, the original high-dimensional historical learning score data is converted into a low-dimensional representation, while the intrinsic structure and key features of the data are retained as much as possible, and the data after dimensionality reduction form multiple reduced-dimensional historical learning score sets, and the historical learning score dimension in each set is reduced, but the information density is increased. The historical learning scores in each reduced-dimensional historical learning score set are arranged in chronological order (the data in the set is timestamped and sorted according to the timestamp), and the data after time series arrangement form multiple historical learning score sequences, each of which reflects the changing trend of the user's learning and translation results of the target word in different time periods.

[0031] In a possible implementation, the multiple historical learning score sets are optimized and reduced in dimension respectively to obtain multiple reduced-dimension historical learning score sets, and are arranged in time series to obtain multiple historical learning score sequences, including: selecting a first benchmark learning score from a first historical learning score set within the multiple historical learning score sets; allocating a distribution probability according to the difference between other historical learning scores in the first historical learning score set and the first benchmark learning score to obtain a first basic probability distribution, wherein the size of the difference is negatively correlated with the size of the distribution probability; optimizing and reducing the dimension of the first historical learning score set according to the first basic probability distribution to obtain a first reduced-dimension learning score set; sorting the multiple historical learning scores in the first reduced-dimension learning score set according to timestamp information to obtain a first historical learning score sequence; optimizing and reducing the dimension and arranging the other multiple historical learning score sets in time series to obtain multiple historical learning score sequences.

[0032] Specifically, the first historical learning score set is any one of the multiple historical learning score sets. From the first historical learning score set, a representative score is selected as the first benchmark learning score. For example, the first benchmark learning score can be the average value, median or score calculated based on a certain algorithm (such as cluster center) of the first historical learning score set. Traverse each historical learning score in the first historical learning score set, calculate the difference between it and the first benchmark learning score, and assign a distribution probability according to the size of the difference. The assignment method can be specifically implemented by a Gaussian function, an exponential function, etc., wherein the smaller the difference (that is, the closer the historical learning score is to the benchmark learning score), the higher the assigned distribution probability; the larger the difference, the lower the assigned distribution probability. According to the assigned probability distribution, the first historical learning score set is subjected to dimensionality reduction processing, including retaining historical learning scores with high distribution probabilities, removing or merging historical learning scores with low distribution probabilities, or recalculating historical learning scores using a distribution probability weighted method, etc. Obtain the timestamp information of each historical learning score in the first dimensionality reduction learning score set, sort the historical learning scores in the order of the timestamps, and form a first historical learning score sequence. Repeat the above steps and perform the same processing on other multiple historical learning score sets, and each set will obtain an optimized and reduced historical learning score sequence. This implementation method retains the historical learning scores with high similarity to the benchmark points by allocating distribution probabilities, removes or reduces the weights of historical learning scores with low distribution probabilities, reduces the interference of noise data on the analysis results, and achieves the technical effect of improving the accuracy of dimensionality reduction.

[0033] In one possible implementation, the first historical learning score set is optimized and reduced in dimension according to the first basic probability distribution to obtain a first reduced-dimensionality learning score set, including: randomly extracting a preset number of reduced-dimensionality historical learning scores from the first historical learning score set to obtain a first reduced-dimensionality historical learning score set; assigning distribution probabilities according to the difference between the historical learning scores in the first reduced-dimensionality historical learning score set and the first benchmark learning scores to obtain a first reduced-dimensionality probability distribution; calculating the similarity between the first reduced-dimensionality probability distribution and the first basic probability distribution as a first reduced-dimensionality fitness; randomly extracting a preset number of reduced-dimensionality historical learning scores from the first historical learning score set again to obtain a second reduced-dimensionality historical learning score set, and processing to obtain a second reduced-dimensionality fitness; continuing to optimize and reduce the dimension until convergence, and outputting the reduced-dimensionality historical learning score set with the largest reduced-dimensionality fitness as the first reduced-dimensionality learning score set.

[0034] Specifically, according to actual needs, a preset number of historical learning scores are randomly selected from the first historical learning score set, and the preset number is the number of historical learning scores that are desired to be retained after dimensionality reduction, and the randomly selected historical learning scores are used to form the first dimensionality reduction historical learning score set. The difference between each historical learning score and the first benchmark learning score in the first dimensionality reduction historical learning score set is calculated, and a distribution probability is assigned to each historical learning score according to the size of the difference, ensuring that the smaller the difference, the higher the assigned distribution probability; the larger the difference, the lower the assigned distribution probability. The similarity between the first dimensionality reduction probability distribution and the first basic probability distribution is calculated using a similarity measurement method (such as KL divergence, JS divergence, correlation coefficient, etc.). The similarity value reflects the degree of closeness between the dimensionality reduction historical learning score set and the original historical learning score set in probability distribution. The calculated similarity value is used as the first dimensionality reduction fitness. The higher the dimensionality reduction fitness, the better the dimensionality reduction historical learning score set is in retaining the original data characteristics. Randomly extract the preset number of historical learning scores in the first historical learning score set again to obtain the second set of historical learning scores with reduced dimensionality. Perform the same processing on the second set of historical learning scores, including assigning distribution probability, calculating the similarity with the first basic probability distribution, and obtaining the second dimensionality reduction fitness. Repeat this process multiple times. During the multiple random extractions and optimization processes, record the dimensionality reduction fitness obtained each time. When the dimensionality reduction fitness changes very little or no longer changes significantly for multiple consecutive times, it is considered that the algorithm has converged, and the result with the largest dimensionality reduction fitness is selected as the final first set of reduced dimensionality learning scores. This implementation method evaluates the dimensionality reduction effect by calculating the similarity between the probability distribution after dimensionality reduction and the original probability distribution, ensuring that the historical learning score set after dimensionality reduction does as good as possible in retaining the characteristics of the original data, and achieving the technical effect of improving the accuracy of dimensionality reduction.

[0035] The learning trend analysis module 40 is used to perform learning trend analysis based on the multiple historical learning score sequences to obtain the learning rate and degradation rate. Specifically, the historical learning score sequences are cleaned and formatted to ensure the accuracy and consistency of the data, handle missing values, outliers and other issues, and convert the historical learning score sequences into a form suitable for trend analysis. According to the characteristics of the historical learning score sequence, a trend analysis method is selected, such as time series analysis (such as ARIMA model, exponential smoothing method, etc.), regression analysis, machine learning model, etc., and the selected trend analysis method is applied to multiple historical learning score sequences for learning trend analysis, that is, a systematic analysis is conducted on the changes in the user's learning results over a period of time, aiming to discover the changing rules, influencing factors and potential problems of the learning results. During the analysis process, attention is paid to factors such as the changing trend, periodicity, volatility and other factors of the historical learning scores, as well as how these factors change over time, and the learning rate and degradation rate are calculated. Among them, the learning rate represents the user's learning degree of the target word, and is an indicator of learning speed or efficiency. It refers to the degree of mastery of the target word per unit time. For example, it can be represented by the ratio of the mean of the historical learning scores per unit time to the full score; the degradation rate represents the rate at which the learning results decay over time, reflecting the decay of the user's memory or understanding of the target word over a period of time. For example, it can be represented by the reduction from the highest value of the historical learning score per unit time to the current latest learning score.

[0036] In a possible implementation, a learning trend analysis is performed based on the multiple historical learning score sequences to obtain a learning rate and a degradation rate, including: collecting a set of sample learning score sequences based on sample learning data of multiple users, and obtaining a set of sample learning rates and a set of sample degradation rates based on a learning score change identifier within each sample learning score sequence; using the set of sample learning score sequences as a classification input, and using the set of sample learning rates and the set of sample degradation rates as a classification output to construct a learning trend analyzer; based on the learning trend analyzer, performing learning trend classification on the multiple historical learning score sequences to obtain multiple context learning rates and multiple context degradation rates; analyzing the similarity between the target context information and the multiple classified historical contexts, and performing weighted calculation on the multiple context learning rates and the multiple context degradation rates according to the magnitude of the multiple context similarities to obtain a learning rate and a degradation rate.

[0037] Specifically, a learning score sequence is extracted from the historical learning data of multiple users. The learning score sequence is a series of learning scores arranged in chronological order, reflecting the dynamic changes in the user's learning performance. For each user's sample learning score sequence, the learning rate (the speed at which the learning score increases) and the degradation rate (the speed at which the learning score decreases) are calculated by comparing the changes in the learning scores at adjacent time points. The learning rate refers to the speed at which the learning score improves, and the degradation rate refers to the speed at which the learning score decreases. Using machine learning or statistical learning methods, with the sample learning score sequence as the input feature, and the sample learning rate and sample degradation rate as the output target, a model (such as a classifier or regressor) is trained. This model can predict the learning rate and degradation rate corresponding to the new learning score sequence.

[0038] Using the constructed learning trend analyzer, multiple historical learning score sequences are classified to obtain the learning rate and degradation rate corresponding to each sequence. Since each historical learning score sequence corresponds to a context, multiple context learning rates and multiple context degradation rates are obtained. The context learning rate and context degradation rate reflect the impact of context factors on learning effects. The similarity between the target context information and multiple classified historical contexts is analyzed. The context similarity is used to evaluate the applicability of the historical context to the target context. According to the size of the similarity, multiple context learning rates and context degradation rates are weighted and calculated to obtain the final learning rate and degradation rate. This implementation method comprehensively reflects the dynamic changes of learning effects by considering the historical learning data of multiple users and their performance in different contexts, and achieves the technical effect of improving the accuracy of learning rate and degradation rate calculation.

[0039] A correction module 50 is used to match the target context information with the multiple classified historical contexts to obtain a matching historical context, and to correct the learning rate and the degradation rate according to the deviation between the target context information and the standard matching context information of the matching historical context to obtain a corrected learning rate and a corrected degradation rate. Specifically, the target context information is matched with multiple classified historical contexts, and the similarity between the target context and each classified historical context is calculated, such as using a text similarity algorithm (such as cosine similarity, Jaccard similarity, etc.) or a semantic similarity calculation based on word embedding, and the classified historical context with the highest similarity is selected as the matching historical context. Standard matching context information is extracted from the matching historical context, and the standard matching context information is a set of representative context information with a high similarity to the target context. The difference between the target context information and the standard matching context information is analyzed, and the deviation is calculated. The deviation may include multiple aspects, such as the complexity of the context, the usage of the target word in the context, and the emotional color of the context. Based on the calculated deviation, the learning rate and degradation rate are corrected. The correction can be a simple linear adjustment (such as adding or subtracting the learning rate or degradation rate according to the size of the deviation) or a nonlinear model or machine learning algorithm. Through correction, the learning rate and degradation rate are more in line with the actual situation of the current target context, thereby improving the accuracy and practicality of the analysis.

[0040] In one possible implementation, the target context information is matched with the multiple classified historical contexts to obtain a matching historical context, and the learning rate and the degradation rate are corrected according to the deviation between the target context information and the standard matching context information of the matching historical context to obtain a corrected learning rate and a corrected degradation rate, including: selecting the classified historical context with the greatest similarity as the matching historical context, and obtaining the standard matching context information of the matching historical context; setting a learning correction coefficient according to the deviation between the target context information and the standard matching context information of the matching historical context; using the learning correction coefficient to correct and calculate the learning rate and the degradation rate to obtain a corrected learning rate and a corrected degradation rate.

[0041] Specifically, the similarity between the target context information and each classified historical context is calculated, and among all the calculated similarities, the classified historical context with the highest similarity is selected as the matching historical context. According to the matching historical context, the corresponding standard matching context information is queried, that is, a set of information representing the core features of the classified historical context, including keywords, topic tags, emotional tendencies, etc. The target context information is compared with the standard matching context information of the matching historical context, and the deviation between the two is calculated. The deviation can be the sum of the differences based on specific attributes or a certain comprehensive score. According to the calculated deviation, a learning correction coefficient is set, and the learning correction coefficient is used to adjust the learning rate and degradation rate to cope with the difference between the introduction of new context information and the old context information. The learning correction coefficient is applied to the learning rate and degradation rate respectively. The specific formula for the correction calculation can be designed according to actual needs. Through the correction calculation, the corrected learning rate and degradation rate are obtained. This implementation method accurately identifies the difference between the target context and the known historical context by matching the historical context and calculating the deviation, so as to adjust the learning rate and degradation rate in a targeted manner, thereby achieving the technical effect of improving the targetedness and accuracy of the correction.

[0042] The auxiliary translation prompt module 60 is used to make auxiliary translation decisions according to the modified learning rate and the modified degradation rate, obtain auxiliary translation prompt schemes, and make auxiliary translation prompts. Specifically, the modified learning rate and the modified degradation rate reflect the user's mastery of the target word in the current target context, analyze the user's translation needs, including target context information, translation purpose (such as reading comprehension, writing expression, oral communication, etc.), the user's language level, etc., combine the modified learning rate and the modified degradation rate, evaluate the translation difficulty and challenges encountered by the user at the current stage, and make auxiliary translation decisions accordingly. The decision may include selecting appropriate translation methods (such as literal translation, free translation, phrase replacement, etc.), providing examples or context support, recommending learning resources, etc. The auxiliary translation decision is converted into a specific prompt scheme, including word meaning selection, grammar prompts, translation suggestions, example sentence display, context simulation, learning resource links, etc. The generated auxiliary translation prompt scheme is presented to the user in an appropriate manner, such as displaying on the learning interface, prompting through a pop-up window, sending an email or text message, etc., to assist the user in completing the translation task. The prompt content can point out the user's translation difficulties and provide specific solutions or suggestions. In addition, during the prompting process, user feedback can be collected, and the effect of the auxiliary translation prompt can be evaluated through the user's feedback, including the user's acceptance, the improvement of translation accuracy, etc. According to the evaluation results, the auxiliary translation prompt scheme is adjusted and optimized to better meet the needs of the user. The embodiment of the present application adopts the method of obtaining the target word and its context information currently learned by the user, indexing the historical learning data of the target word by the user, classifying the historical context information, extracting the word meaning information and learning data in different contexts, processing the learning score data by optimizing the dimensionality reduction technology, reducing the dimension and arranging it in time sequence, analyzing the learning trend according to the historical learning score sequence, obtaining the learning rate and degradation rate, correcting the learning rate and degradation rate according to the deviation between the target context and the matching historical context, making auxiliary translation decisions according to the corrected learning rate and degradation rate, providing personalized translation prompt schemes and other technical means, realizing personalized and intelligent English word learning and translation prompts, and achieving the technical effect of improving the pertinence and accuracy of auxiliary translation.

[0043] In a possible implementation, according to the corrected learning rate and the corrected degradation rate, an auxiliary translation decision is made, an auxiliary translation prompt scheme is obtained, and an auxiliary translation prompt is performed, including: collecting a sample corrected learning rate set and a sample corrected degradation rate set, and setting a sample auxiliary translation prompt scheme according to the size of each sample corrected learning rate and the sample corrected degradation rate, and obtaining a sample auxiliary translation prompt scheme set, wherein each sample auxiliary translation prompt scheme includes a prompt ratio, and the size of the sample corrected learning rate and the sample corrected degradation rate is negatively correlated with the size of the prompt ratio; using the sample corrected learning rate set and the sample corrected degradation rate set as decision inputs, using the sample auxiliary translation prompt scheme set as decision outputs, and constructing an auxiliary translation decision maker; using the auxiliary translation decision maker, making an auxiliary translation decision on the corrected learning rate and the corrected degradation rate, and obtaining an auxiliary translation prompt scheme.

[0044] Specifically, data on the modified learning rate and modified degradation rate are collected from historical data of multiple translation tasks, translators or translation systems, and organized into a modified learning rate set containing all samples and a modified degradation rate set containing all samples. A prompt ratio is defined for each sample, which determines how much auxiliary information (such as translation suggestions, vocabulary prompts, etc.) is provided during the translation process. According to the size of the sample modified learning rate and the sample modified degradation rate, an association rule is formulated, that is, the higher the learning rate and the lower the degradation rate, the lower the prompt ratio; conversely, the lower the learning rate and the higher the degradation rate, the higher the prompt ratio. According to the association rule, an auxiliary translation prompt scheme is generated for each sample, and these auxiliary translation prompt schemes are combined into a sample auxiliary translation prompt scheme set. According to the characteristics of the sample modified learning rate set and the sample modified degradation rate set, a machine learning or deep learning algorithm (such as a decision tree, a neural network, etc.) is selected, and the sample modified learning rate set and the sample modified degradation rate set are used as inputs, and the sample auxiliary translation prompt scheme set is used as output to train the auxiliary translation decision maker. The new corrected learning rate and corrected degradation rate are passed as input data to the trained auxiliary translation decision maker, and the auxiliary translation decision maker outputs the corresponding auxiliary translation prompt scheme according to the input data. During the translation process, the user is provided with corresponding auxiliary information according to the obtained auxiliary translation prompt scheme to improve the translation efficiency and accuracy. This implementation method achieves the technical effect of improving the efficiency and accuracy of obtaining the auxiliary translation prompt scheme by constructing an auxiliary translation decision maker.

[0045] In the above, refer to Figure 1 A system for intelligent learning and translation of English words according to an embodiment of the present invention is described in detail. Figure 2 An English word intelligent learning and translation method according to an embodiment of the present invention is described.

[0046] An intelligent learning and translation method for English words according to an embodiment of the present invention is used to solve the technical problem of insufficient pertinence and accuracy of auxiliary translation in existing English word learning and translation, and achieve the technical effect of improving the pertinence and accuracy of auxiliary translation. An intelligent learning and translation method for English words includes: The target word currently being learned by the user is obtained, and the target context information of the target word currently located is collected; according to the target word, the historical learning data of the target word within the user's historical time is indexed; the historical context information in the historical learning data is classified to obtain multiple classified historical contexts, and historical word meaning information sets and multiple historical learning data sets under the multiple classified historical contexts are extracted, and multiple historical learning score sets are obtained by processing; the multiple historical learning score sets are respectively optimized and reduced in dimension to obtain multiple reduced-dimension historical learning score sets, and are arranged in time sequence to obtain multiple historical learning score sequences; learning trend analysis is performed according to the multiple historical learning score sequences to obtain a learning rate and a degradation rate; the target context information is matched with the multiple classified historical contexts to obtain a matching historical context, and the learning rate and the degradation rate are corrected according to the deviation between the target context information and the standard matching context information of the matching historical context to obtain a corrected learning rate and a corrected degradation rate; according to the corrected learning rate and the corrected degradation rate, an auxiliary translation decision is made, an auxiliary translation prompt scheme is obtained, and an auxiliary translation prompt is performed.

[0047] Among them, obtaining the target word that the user is currently learning, and collecting the target context information of the target word, and indexing the historical learning data of the target word within the user's historical time according to the target word, can further include: obtaining the target word that the user is currently learning, collecting the upper context content and the lower context content of the target word as the target context information; according to the target word, indexing the historical learning record of the user to obtain the historical learning data of the target word.

[0048] Among them, classifying the historical context information in the historical learning data to obtain multiple classified historical contexts, extracting historical word meaning information sets and multiple historical learning data sets under the multiple classified historical contexts, and processing to obtain multiple historical learning score sets can further include: obtaining multiple historical context information in the historical learning data, classifying them, and obtaining multiple classified historical contexts; extracting the standard meanings of target words in the historical learning data under the multiple classified historical contexts to obtain a historical word meaning information set, and extracting multiple historical translation semantic information sets and multiple historical translation submission time sets translated and submitted by users under the multiple classified historical contexts; classifying and obtaining multiple historical basic learning score sets according to the deviation ranges between the multiple historical translation semantic information sets and the historical word meaning information sets; and correcting and calculating the multiple historical basic learning score sets according to the ratio of a preset submission time threshold and the multiple historical translation submission time sets to obtain multiple historical learning score sets.

[0049] Among them, respectively optimizing and reducing the dimensions of the multiple historical learning score sets to obtain multiple reduced-dimensional historical learning score sets, and arranging them in time series to obtain multiple historical learning score sequences may further include: selecting a first benchmark learning score from a first historical learning score set within the multiple historical learning score sets; allocating distribution probabilities according to the difference between other historical learning scores in the first historical learning score set and the first benchmark learning score to obtain a first basic probability distribution, wherein the size of the difference is negatively correlated with the size of the distribution probability; optimizing and reducing the dimensions of the first historical learning score set according to the first basic probability distribution to obtain a first reduced-dimensional learning score set; sorting the multiple historical learning scores in the first reduced-dimensional historical learning score set according to the timestamp information to obtain a first historical learning score sequence; optimizing and reducing the dimensions of other multiple historical learning score sets and arranging them in time series to obtain multiple historical learning score sequences.

[0050] Among them, according to the first basic probability distribution, the first historical learning score set is optimized and reduced in dimension to obtain the first reduced dimensionality learning score set, which can further include: randomly extracting a preset number of reduced dimensionality historical learning scores from the first historical learning score set to obtain the first reduced dimensionality historical learning score set; assigning distribution probabilities according to the difference between the historical learning scores in the first reduced dimensionality historical learning score set and the first benchmark learning scores to obtain the first reduced dimensionality probability distribution; calculating the similarity between the first reduced dimensionality probability distribution and the first basic probability distribution as the first reduced dimensionality fitness; randomly extracting a preset number of reduced dimensionality historical learning scores from the first historical learning score set again to obtain a second reduced dimensionality historical learning score set, and processing to obtain the second reduced dimensionality fitness; continuing to optimize the dimensionality reduction until convergence, and outputting the reduced dimensionality historical learning score set with the largest dimensionality reduction fitness as the first reduced dimensionality learning score set.

[0051] Among them, performing learning trend analysis according to the multiple historical learning score sequences to obtain learning rate and degradation rate can further include: collecting a set of sample learning score sequences according to sample learning data of multiple users, and obtaining a set of sample learning rate and a set of sample degradation rate according to the learning score change identifier in each sample learning score sequence; using the set of sample learning score sequences as classification input, and using the set of sample learning rate and the set of sample degradation rate as classification output to construct a learning trend analyzer; based on the learning trend analyzer, performing learning trend classification on the multiple historical learning score sequences to obtain multiple context learning rates and multiple context degradation rates; analyzing the similarity between the target context information and the multiple classified historical contexts, and performing weighted calculation on the multiple context learning rates and the multiple context degradation rates according to the size of the multiple context similarities to obtain the learning rate and degradation rate.

[0052] Among them, matching the target context information with the multiple classified historical contexts to obtain a matching historical context, and correcting the learning rate and the degradation rate according to the deviation between the target context information and the standard matching context information of the matching historical context to obtain a corrected learning rate and a corrected degradation rate, may further include: selecting the classified historical context with the greatest similarity as the matching historical context, and obtaining the standard matching context information of the matching historical context; setting a learning correction coefficient according to the deviation between the target context information and the standard matching context information of the matching historical context; using the learning correction coefficient to correct and calculate the learning rate and the degradation rate to obtain a corrected learning rate and a corrected degradation rate.

[0053] Among them, according to the corrected learning rate and the corrected degradation rate, making an auxiliary translation decision, obtaining an auxiliary translation prompt scheme, and making an auxiliary translation prompt can further include: collecting a sample corrected learning rate set and a sample corrected degradation rate set, and setting a sample auxiliary translation prompt scheme according to the size of each sample corrected learning rate and the sample corrected degradation rate, and obtaining a sample auxiliary translation prompt scheme set, wherein each sample auxiliary translation prompt scheme includes a prompt ratio, and the size of the sample corrected learning rate and the sample corrected degradation rate is negatively correlated with the size of the prompt ratio; using the sample corrected learning rate set and the sample corrected degradation rate set as decision inputs, using the sample auxiliary translation prompt scheme set as decision outputs, and constructing an auxiliary translation decision maker; using the auxiliary translation decision maker, making auxiliary translation decisions on the corrected learning rate and the corrected degradation rate, and obtaining an auxiliary translation prompt scheme.

[0054] An English word intelligent learning and translation system provided by an embodiment of the present invention can execute an English word intelligent learning and translation method provided by any embodiment of the present invention, and has functional modules and beneficial effects corresponding to the execution method.

[0055] Although the present application makes various references to certain modules in the system according to the embodiments of the present application, any number of different modules may be used and run on the user terminal and / or server, and the various units and modules included are only divided according to functional logic, but are not limited to the above division, as long as the corresponding functions can be achieved; in addition, the specific names of the functional units are only for the convenience of distinguishing each other, and are not used to limit the scope of protection of the present invention.

[0056] The above specific implementation manner does not constitute a limitation to the protection scope of the present application. It should be understood by those skilled in the art that various modifications, combinations and substitutions can be made according to design requirements and other factors. Any modifications, equivalent substitutions and improvements made within the spirit and principles of the present application should be included in the protection scope of the present application. In some cases, the actions or steps recorded in the present application can be performed in an order different from that in the embodiment and can still achieve the desired results. In addition, the process depicted in the accompanying drawings does not necessarily require the specific order or continuous order shown to achieve the desired results. In some embodiments, multitasking and parallel processing are also possible or may be advantageous.

Claims

1. An intelligent learning and translation system for English words, characterized in that: The system comprises: A historical learning data indexing module, which is used to obtain a target word that the user is currently learning, and to collect target context information of the target word, and to index the historical learning data of the target word within the user's historical time according to the target word; A historical context information classification module, the historical context information classification module is used to classify the historical context information in the historical learning data, obtain multiple classified historical contexts, extract historical word meaning information sets and multiple historical learning data sets under the multiple classified historical contexts, and process to obtain multiple historical learning score sets; A historical learning score set optimization and dimensionality reduction module, wherein the historical learning score set optimization and dimensionality reduction module is used to optimize and reduce the dimensions of the multiple historical learning score sets respectively to obtain multiple reduced-dimensional historical learning score sets, and arrange them in time sequence to obtain multiple historical learning score sequences; A learning trend analysis module, the learning trend analysis module is used to perform learning trend analysis according to the multiple historical learning score sequences to obtain a learning rate and a degradation rate; a correction module, the correction module being used to match the target context information with the multiple classified historical contexts to obtain a matching historical context, and to correct the learning rate and the degradation rate according to a deviation between the target context information and the standard matching context information of the matching historical context to obtain a corrected learning rate and a corrected degradation rate; The auxiliary translation prompt module is used to make auxiliary translation decisions according to the modified learning rate and the modified degradation rate, obtain an auxiliary translation prompt solution, and perform auxiliary translation prompts.

2. The English word intelligent learning and translation system according to claim 1, characterized in that: Obtaining the target word that the user is currently learning, and collecting the target context information of the target word, and indexing the historical learning data of the target word within the user's historical time according to the target word, including: Acquire a target word that the user is currently learning, and collect the upper context content and the lower context content of the target word as target context information; According to the target word, an index is performed in the historical learning record of the user to obtain the historical learning data of the target word.

3. The English word intelligent learning and translation system according to claim 1, characterized in that: The historical context information in the historical learning data is classified to obtain a plurality of classified historical contexts, a historical word meaning information set and a plurality of historical learning data sets under the plurality of classified historical contexts are extracted, and a plurality of historical learning score sets are obtained by processing, including: Acquire multiple historical context information in the historical learning data, perform classification processing, and obtain multiple classified historical contexts; Extracting the standard meanings of the target words in the historical learning data under the multiple classified historical contexts to obtain a historical meaning information set, and extracting multiple historical translation semantic information sets and multiple historical translation submission time sets submitted by users under the multiple classified historical contexts; According to the deviation extents of the multiple historical translation semantic information sets and the historical word meaning information set, a plurality of historical basis learning score sets are obtained by classification; According to the ratio of the preset submission time threshold and the plurality of historical translation submission time sets, the plurality of historical basic learning score sets are corrected and calculated to obtain a plurality of historical learning score sets.

4. The English word intelligent learning and translation system according to claim 1, characterized in that: The multiple historical learning score sets are optimized and reduced in dimension respectively to obtain multiple reduced-dimensional historical learning score sets, and are arranged in time sequence to obtain multiple historical learning score sequences, including: Selecting to obtain a first benchmark learning score from a first historical learning score set in the plurality of historical learning score sets; According to the difference between other historical learning scores in the first historical learning score set and the first benchmark learning score, a distribution probability is allocated to obtain a first basic probability distribution, wherein the size of the difference is negatively correlated with the size of the distribution probability; According to the first basic probability distribution, optimizing and reducing the dimension of the first historical learning score set to obtain a first reduced-dimensionality learning score set; Sort the multiple historical learning scores in the first dimensionality-reduced historical learning score set according to timestamp information to obtain a first historical learning score sequence; Optimize dimensionality reduction and time series arrangement for other multiple historical learning score sets to obtain multiple historical learning score sequences.

5. The English word intelligent learning and translation system according to claim 4 is characterized in that: According to the first basic probability distribution, optimizing and reducing the dimension of the first historical learning score set to obtain a first reduced-dimensional learning score set includes: Randomly extracting a preset number of reduced-dimensionality historical learning scores from the first historical learning score set to obtain a first reduced-dimensionality historical learning score set; According to the difference between the historical learning scores in the first dimensionality reduction historical learning score set and the first benchmark learning score, a distribution probability is assigned to obtain a first dimensionality reduction probability distribution; Calculating the similarity between the first dimensionality reduction probability distribution and the first basic probability distribution as a first dimensionality reduction fitness; Randomly extracting a preset number of reduced-dimension historical learning scores from the first historical learning score set again to obtain a second reduced-dimension historical learning score set, and processing to obtain a second reduced-dimension fitness; Continue to optimize the dimensionality reduction until convergence, and output the dimensionality reduction history learning score set with the largest dimensionality reduction fitness as the first dimensionality reduction learning score set.

6. The English word intelligent learning and translation system according to claim 1, characterized in that: Performing a learning trend analysis according to the plurality of historical learning score sequences to obtain a learning rate and a degradation rate includes: According to the sample learning data of multiple users, a set of sample learning score sequences is collected, and a set of sample learning rates and a set of sample degradation rates are obtained according to a learning score change identifier in each sample learning score sequence; Using the sample learning score sequence set as classification input, using the sample learning rate set and the sample degradation rate set as classification output, to construct a learning trend analyzer; Based on the learning trend analyzer, the multiple historical learning score sequences are classified into learning trend categories to obtain multiple context learning rates and multiple context degradation rates; The similarity between the target context information and the multiple classified historical contexts is analyzed, and the multiple context learning rates and multiple context degradation rates are weightedly calculated according to the magnitude of the multiple context similarities to obtain the learning rate and the degradation rate.

7. The English word intelligent learning and translation system according to claim 6, characterized in that: Matching the target context information with the multiple classified historical contexts to obtain a matching historical context, and correcting the learning rate and the degradation rate according to a deviation between the target context information and the standard matching context information of the matching historical context to obtain a corrected learning rate and a corrected degradation rate, including: Selecting the classified historical context with the greatest similarity as the matching historical context, and obtaining standard matching context information of the matching historical context; Setting a learning correction coefficient according to a deviation between the target context information and the standard matching context information of the matching historical context; The learning correction coefficient is used to perform correction calculation on the learning rate and the degradation rate to obtain a corrected learning rate and a corrected degradation rate.

8. The English word intelligent learning and translation system according to claim 1, characterized in that: According to the modified learning rate and the modified degradation rate, an auxiliary translation decision is made, an auxiliary translation prompt scheme is obtained, and an auxiliary translation prompt is performed, including: Collecting a set of sample correction learning rates and a set of sample correction degradation rates, and setting a sample-assisted translation prompting scheme according to the size of each sample correction learning rate and sample correction degradation rate, to obtain a set of sample-assisted translation prompting schemes, wherein each sample-assisted translation prompting scheme includes a prompting ratio, and the size of the sample correction learning rate and the sample correction degradation rate is negatively correlated with the size of the prompting ratio; Using the sample correction learning rate set and the sample correction degradation rate set as decision inputs, using the sample auxiliary translation prompt scheme set as decision outputs, and constructing an auxiliary translation decision maker; The auxiliary translation decision maker is used to make auxiliary translation decisions on the modified learning rate and the modified degradation rate to obtain an auxiliary translation prompt solution.

9. An intelligent learning and translation method for English words, characterized in that: The method is used for an English word intelligent learning and translation system according to any one of claims 1 to 8, and the method comprises: Obtain the target word that the user is currently learning, and collect the target context information of the target word, and index the historical learning data of the target word within the historical time according to the target word; Classifying the historical context information in the historical learning data to obtain a plurality of classified historical contexts, extracting a set of historical word meaning information and a plurality of historical learning data sets under the plurality of classified historical contexts, and processing to obtain a plurality of historical learning score sets; Optimizing and reducing the dimensions of the multiple historical learning score sets respectively to obtain multiple reduced-dimensional historical learning score sets, and arranging them in chronological order to obtain multiple historical learning score sequences; Performing a learning trend analysis based on the multiple historical learning score sequences to obtain a learning rate and a degradation rate; Matching the target context information with the multiple classified historical contexts to obtain a matching historical context, and correcting the learning rate and the degradation rate according to a deviation between the target context information and the standard matching context information of the matching historical context to obtain a corrected learning rate and a corrected degradation rate; According to the modified learning rate and the modified degradation rate, an auxiliary translation decision is made, an auxiliary translation prompt scheme is obtained, and an auxiliary translation prompt is performed.

Citation Information

Patent Citations

  • Candidate phrase querying method and aided translation system for computer aided translation

    CN102693309A

  • Language model pre-training method and device, computer storage medium and electronic equipment

    CN114330327A

  • Auxiliary translation method and device, electronic equipment and storage medium

    CN115659998A

  • Online education method and system based on image simulation and storage medium

    CN117056546A

  • Intelligent recommendation answering system based on distributed summary of examination point data

    CN117540108A

Cited By

  • Medical stent additive manufacturing defect detection system based on artificial intelligence technology

    CN120823183A