Language learning ability evaluation system and method based on artificial intelligence assistance

Abstract

The invention relates to a language learning ability evaluation system and method based on artificial intelligence assistance, belongs to the field of semantic processing, and more particularly belongs to the field of natural language data processing, and the system comprises a content conversion mechanism which is used for converting a current English composition document into corresponding digital conversion content by adopting a targeted digital conversion mode; and the synchronous evaluation mechanism is used for intelligently evaluating each item score value and plagiarism space ratio value of the current English composition document according to each item of basic information including the digital conversion content based on an AI English ability evaluation model. According to the method and the device, aiming at the technical problem that in the prior art, a targeted English composition digital conversion mechanism, an artificial intelligence model with a customized structure design and various basic information which is fully and comprehensively screened can be introduced at the same time in allusion to the technical problem that various item score data and plagiarism space information of an English composition document are difficult to synchronously evaluate; therefore, the technical problem is solved.

Description

TECHNICAL FIELD

[0001] The semantic processing of the present application relates to the field of processing natural language data, and particularly relates to an artificial intelligence assisted language learning ability evaluation system and method. BACKGROUND

[0002] Processing natural language data is a common processing type of semantic processing, which can help people skip complex and tedious manual processing links and directly provide the required natural language data processing results by using machine processing mode. Since the scale of natural language data to be processed is generally large, artificial intelligence assisted mode is usually used to realize machine processing of large-scale natural language data, thereby providing direct, fast and effective electronic ways for people to understand and judge natural language data.

[0003] Currently, composition is an important manifestation of language learning ability, and electronic composition marking is a reliable way to evaluate language learning ability, and is also one of important application fields of processing natural language data, which can realize non-manual machine processing of a large amount of language data corresponding to each examinee respectively, and directly, quickly and effectively give each total score data corresponding to each composition of a large amount of examinees, wherein, artificial intelligence assisted mode is usually used to realize specific electronic composition marking processing, thereby shortening the speed of composition marking and reducing the cost of composition marking. For example, Chinese invention patent publication CN118916475A proposes a Chinese composition scoring method, device, electronic equipment and readable storage medium, the method comprising: obtaining a to-be-scored composition; performing statistical analysis on the to-be-scored composition in the text dimension to obtain a statistical score, the text dimension at least including the number of words, vocabulary and text errors; using a large model to analyze the to-be-scored composition in the semantic dimension to obtain a semantic score, the semantic dimension at least including the central content, rhetorical expression, article structure, emotional expression and topic-text matching; determining the final score of the to-be-scored composition according to the statistical score, the semantic score and the preset scoring rule.

[0004] For example, Chinese invention patent publication CN111914532A proposes a Chinese composition scoring method, the method comprising: obtaining a to-be-scored Chinese composition; analyzing the to-be-scored Chinese composition to obtain a corresponding scoring dimension table; scoring the to-be-scored Chinese composition according to the scoring rule under each dimension to obtain the scoring result of the composition in each dimension; and weighting the scoring result to obtain the final scoring result of the to-be-scored Chinese composition. Compared with the traditional composition scoring method, the scheme of the present application can obtain a more fine-grained composition scoring result, and can fine-tune the total score by customizing the weight of each dimension to the total score, while making the scoring result more transparent and more interpretable to the user.

[0005] However, the technical solutions of the existing electronic composition marking scheme focus on the marking processing of Chinese natural language data, and lack targeted electronic solutions for English composition marking. At the same time, the existing electronic composition marking mode cannot give specific scores from aspects such as emotional sufficiency, logic, coherence, text correctness, and grammatical accuracy for each English composition, that is, it cannot perform multi-aspect sub-item scoring on each English composition, making it difficult for the marker and the marked to fully understand the English composition, and the marker cannot use a weighted method to judge the final total score. In addition, the results of the existing electronic composition marking mode are relatively single, and cannot give the plagiarism length proportion of the composition together with the scoring data. SUMMARY

[0006] In order to solve the technical problems in the prior art, the present application provides a language learning ability evaluation system and method based on artificial intelligence assistance, which can solve the problem of difficulty in English composition digital conversion by using a targeted English composition digital conversion mechanism. Based on the customized structure design of the artificial intelligence model, according to the obtained basic information of the current English composition after full and comprehensive screening, the current English composition sub-item score data and the plagiarism length proportion data are given synchronously, thereby realizing the comprehensive electronic marking of the English composition, and obtaining the plagiarism suspicious information of the English composition, which saves a lot of labor cost, time cost and economic cost for massive English composition marking.

[0007] According to an aspect of the present application, a language learning ability evaluation system based on artificial intelligence assistance is provided, which comprises: A content conversion mechanism is configured to concatenate the ASCII codes of each English character appearing in each sentence of the current English composition document to obtain a binary code stream with a fixed number of bits, and output the binary code stream corresponding to each sentence of the current English composition document as the digital conversion content of the current English composition document. An information analysis mechanism is configured to output the total number of English characters, the proportion of repeated characters, the standard deviation of the number of English characters corresponding to each sentence, the number of English paragraphs, and the number of English characters corresponding to the longest English paragraph in the current English composition document as the multiple related information of the current English composition document. A model building mechanism is configured to perform multiple learning operations on the feedforward neural network to obtain a feedforward neural network after performing the multiple learning operations and output it as an AI English ability evaluation model, and the number of learning operations performed by the feedforward neural network is positively correlated with the number of limited characters in the English composition. The synchronous evaluation mechanism is connected with the content conversion mechanism, the information analysis mechanism and the model establishment mechanism respectively, and is used for intelligently evaluating the various sub-score values and the plagiarism length proportion value of the current English composition document based on the AI English ability evaluation model according to the English composition limit character number, the various sub-score values of the latest past English compositions of the author to which the current English composition document belongs, the multiple related information of the current English composition document and the digital conversion content of the current English composition document.

[0008] According to another aspect of the present application, a language learning ability evaluation method based on artificial intelligence assistance is provided, the method comprising: The ASCII codes of each English character appearing in each sentence of the current English composition document are concatenated to obtain a fixed number of binary code streams, and the respective binary code streams corresponding to each sentence of the current English composition document are output as the digital conversion content of the current English composition document. The English character total number, the repeated character proportion, the standard deviation of the English character number corresponding to each sentence, the English paragraph number and the English character number corresponding to the longest English paragraph of the current English composition document are output as the multiple related information of the current English composition document. The feedforward neural network is executed multiple times to obtain the feedforward neural network after the multiple learning operations are performed and is output as the AI English ability evaluation model, and the number of learning operations performed by the feedforward neural network is positively correlated with the English composition limit character number. The synchronous evaluation mechanism is connected with the content conversion mechanism, the information analysis mechanism and the model establishment mechanism respectively, and is used for intelligently evaluating the various sub-score values and the plagiarism length proportion value of the current English composition document based on the AI English ability evaluation model according to the English composition limit character number, the various sub-score values of the latest past English compositions of the author to which the current English composition document belongs, the multiple related information of the current English composition document and the digital conversion content of the current English composition document.

[0009] Therefore, the present application has at least the following five outstanding substantial features: (1) The targeted English composition digital conversion mechanism solves the problem of difficulty in English composition digital conversion, thereby providing a key basic information for the intelligent evaluation of the subsequent AI English ability evaluation model. Specifically, the ASCII codes of each English character appearing in each sentence of the current English composition document are concatenated to obtain a fixed number of binary code streams, and the respective binary code streams corresponding to each sentence of the current English composition document are output as the digital conversion content of the current English composition document. For each sentence of English text, the tail zero padding and truncation method is adopted to complete the analysis of the fixed number of binary code streams. (2) In the intelligent evaluation result of the AI English ability evaluation model, the synchronous intelligent evaluation of the current English composition document is realized, and the suspicious information of the English composition is obtained, which saves a lot of manual cost, time cost and economic cost for the massive English composition evaluation; (3) In order to perform the synchronous intelligent evaluation of the current English composition document, the AI English ability evaluation model with customized structure is introduced, which is a feedforward neural network after multiple learning operations, and the number of learning operations of the feedforward neural network is positively correlated with the number of English composition limit characters. The customized structure of the above AI English ability evaluation model ensures the stability and effectiveness of the synchronous intelligent evaluation result; (4) In order to perform the synchronous intelligent evaluation of the current English composition document, the basic information is fully and comprehensively screened, which includes the digital conversion content of the current English composition document after targeted digital conversion, the number of English composition limit characters, the current English composition document belonging to the author's latest multiple English composition item score values and multiple related information of the current English composition document, wherein the multiple related information of the current English composition document includes the total number of English characters, the repetition character ratio, the standard deviation of the number of English characters corresponding to each sentence, the number of English paragraphs and the number of English characters corresponding to the longest English paragraph, and the number of the author's latest multiple English compositions is proportional to the number of English composition limit characters. The full and comprehensive screening of the above basic information further ensures the stability and effectiveness of the synchronous intelligent evaluation result; (5) In each learning operation of the feedforward neural network, the known item score values and the known plagiarism length ratio of a certain past English composition document are used as the two output contents of the feedforward neural network, and the number of English composition limit characters, the item score values of the author's latest multiple English compositions of the certain past English composition document, the multiple related information of the certain past English composition document and the digital conversion content of the certain past English composition document are used as the multiple input contents of the feedforward neural network. Complete this learning operation, thereby ensuring the learning effect of each learning operation. BRIEF DESCRIPTION OF DRAWINGS

[0010] The embodiments of the present application will be described below in conjunction with the accompanying drawings, in which: Figure 1 The working scene diagram of the language learning ability evaluation system and method based on artificial intelligence assistance according to the present application.

[0011] Figure 2 An internal structure diagram of the language learning ability evaluation system assisted by artificial intelligence according to a first embodiment of the present application is shown.

[0012] Figure 3 An internal structure diagram of the language learning ability evaluation system assisted by artificial intelligence according to a second embodiment of the present application is shown.

[0013] Figure 4 An internal structure diagram of the language learning ability evaluation system assisted by artificial intelligence according to a third embodiment of the present application is shown.

[0014] Figure 5 An internal structure diagram of the language learning ability evaluation system assisted by artificial intelligence according to a fourth embodiment of the present application is shown.

[0015] Figure 6 An internal structure diagram of the language learning ability evaluation system assisted by artificial intelligence according to a fifth embodiment of the present application is shown.

[0016] Figure 7 A step flow chart of the language learning ability evaluation method assisted by artificial intelligence according to a sixth embodiment of the present application is shown. DETAILED DESCRIPTION

[0017] As shown in Figure 1 , a working scene diagram of the language learning ability evaluation system and method assisted by artificial intelligence according to the present application is shown. The semantic processing of the present application relates to the field of processing natural language data.

[0018] The specific technical process of the present application is as follows: Technical process A: In order to solve the key bottleneck in the prior art, i.e. the technical problem of difficulty in digital conversion of English compositions, a specific English composition digital conversion mechanism is designed, as shown in Figure 1 . Specifically, the specificity, reliability and stability of the English composition digital conversion mechanism of the present application are mainly embodied in the following two aspects of design: Firstly, the ASCII codes of each English character appearing in each sentence of the current English composition document are connected at the beginning and the end to obtain a binary code stream of a fixed number of bits, and each sentence of the current English composition document corresponds to a binary code stream as the digital conversion content of the current English composition document; Secondly, for each sentence of English text, the tail zero padding and tail cutting method is adopted to complete the analysis of the binary code stream of a fixed number of bits; In this way, through the design of the above targeted English composition digital conversion mechanism, a key basic information is provided for the subsequent intelligent evaluation of the AI English ability evaluation model; Technical process B: in order to perform the synchronous intelligent evaluation of the sub-item score value and the plagiarism proportion value of the current English composition document, an AI English ability evaluation model with customized structure design is introduced, as shown in Figure 1 ; Specifically, the customized structure design of the AI English ability evaluation model mainly reflects in the following three aspects: First: the AI English ability evaluation model is a feedforward neural network after performing multiple learning operations; As can be seen, the AI English ability evaluation model adopts a feedforward neural network architecture, which includes a single input layer, a single output layer, and multiple hidden layers, the multiple hidden layers are between the single input layer and the single output layer, and the number of hidden layers is proportional to the number of limited characters of English composition; For example, when the number of limited characters of English composition is 3000 characters, the number of selected hidden layers is 3, when the number of limited characters of English composition is 4000 characters, the number of selected hidden layers is 4, when the number of limited characters of English composition is 5000 characters, the number of selected hidden layers is 5, when the number of limited characters of English composition is 6000 characters, the number of selected hidden layers is 6, and so on; Second: the number of learning operations performed by the feedforward neural network is positively correlated with the number of limited characters of English composition; For example, when the number of limited characters of English composition is 3000 characters, the number of learning operations performed by the feedforward neural network is selected to be 500 times, when the number of limited characters of English composition is 4000 characters, the number of learning operations performed by the feedforward neural network is selected to be 600 times, when the number of limited characters of English composition is 5000 characters, the number of learning operations performed by the feedforward neural network is selected to be 700 times, when the number of limited characters of English composition is 6000 characters, the number of learning operations performed by the feedforward neural network is selected to be 800 times, and so on; Third: in each learning operation performed on the feedforward neural network, the known sub-item score value and the known plagiarism proportion value of a certain past English composition document are taken as two output contents of the feedforward neural network, the number of limited characters of English composition, the sub-item score value of the latest past multiple English compositions of the author to which the certain past English composition document belongs, the multiple associated information of the certain past English composition document, and the digital conversion content of the certain past English composition document are taken as multiple input contents of the feedforward neural network, and the learning operation is completed, thereby ensuring the learning effect of each learning operation; In this way, the customized structural design of the above-mentioned AI English proficiency assessment model ensures the stability and effectiveness of the synchronous intelligent assessment results; Technical Process C: In order to perform the simultaneous intelligent assessment of the score values ​​of each sub-item and the percentage of plagiarized content of the current English composition document, various basic information that has been fully and comprehensively screened is introduced, such as Figure 1 As shown; Specifically, the basic information includes not only the digitally converted content of the current English composition document obtained after targeted digital conversion, but also the character limit of the English composition, the sub-item rating values ​​of the latest and past English compositions of the author of the current English composition document, and multiple related information of the current English composition document; More specifically, multiple pieces of associated information of the current English composition document include the total number of English characters in the current English composition document, the percentage of repeated characters, the standard deviation of the number of English characters corresponding to each English sentence, the number of English paragraphs, and the number of English characters corresponding to the longest English paragraph, as well as the number of the most recent English compositions written by the author of the current English composition document, which is proportional to the character limit of the English composition. In this way, through the full and comprehensive screening of the above basic information, the stability and effectiveness of the synchronous intelligent assessment results are further guaranteed; Technical process D: Based on the targeted English composition digital conversion mechanism designed in technical process A, the AI ​​English proficiency assessment model with customized structure designed in technical process B is adopted. Based on the basic information of the digital conversion content of the current English composition document obtained after the targeted digital conversion, which is fully and comprehensively screened by technical process C, the synchronous intelligent assessment of the sub-item score values ​​and the proportion of plagiarized content of the current English composition document is completed. Figure 1 As shown; This demonstrates that the present invention solves another technical problem in the prior art, namely, the bottleneck of being unable to simultaneously analyze the sub-item scores and plagiarized length data of English composition documents; Technical process E: Based on the score values ​​of each sub-item and the percentage of plagiarized content of the current English composition document obtained by the intelligent assessment of technical process D, the total score of the current English composition document and whether the current English composition document is a plagiarized composition document are determined respectively; For example, a sub-item weighted scoring mode is adopted to calculate the total score value of the current English composition document based on the sub-item score values ​​of the current English composition document, wherein the sub-item score values ​​of the current English composition document are multiplied by their respective score weights to obtain respective products, and the respective products are added to obtain the total score value of the current English composition document, and the respective score weights corresponding to the respective sub-item score values ​​are all values ​​less than 1, and the cumulative value of the respective score weights corresponding to the respective sub-item score values ​​is equal to 1; For example, when the plagiarism proportion of the current English composition document is greater than or equal to a preset proportion threshold, the current English composition document is identified as a plagiarized composition document, otherwise, the current English composition document is identified as a non-plagiarized composition document, wherein the current English composition document identified as a plagiarized composition document is directly set to zero in the total score value; Therefore, the present application solves the two bottleneck technical problems in the prior art, and based on the targeted English composition digital conversion mechanism, the synchronous intelligent evaluation of the item score value and the plagiarism proportion value of the current English composition document is realized, the comprehensive electronic judgment of the English composition is completed, and the suspicious information of the English composition is obtained, thereby saving a large amount of manual cost, time cost and economic cost for the judgment of a large number of English compositions.

[0019] The key points of the present application are: targeted design of the English composition digital conversion mechanism, simultaneous analysis of the item score and plagiarism proportion data of the English composition document, multi-structure design of the AI English ability evaluation model, and sufficient and comprehensive screening of the basic information of the digital conversion content of the current English composition document obtained after targeted digital conversion.

[0020] In the following, the language learning ability evaluation system and method based on artificial intelligence assistance according to the present application will be described in detail in the form of embodiments. Embodiment

[0021] Figure 2 The internal structure diagram of the language learning ability evaluation system based on artificial intelligence assistance according to the first embodiment of the present application is shown.

[0022] As shown in Figure 2 The language learning ability evaluation system based on artificial intelligence assistance includes the following components: The content conversion mechanism is used for concatenating the ASCII codes of each English character appearing in each sentence of the current English composition document to obtain a binary code stream of a fixed number of bits, and outputting each binary code stream corresponding to each sentence of the current English composition document as the digital conversion content of the current English composition document. Specifically, since the ASCII code of each English character is actually a binary code stream of a fixed number of bits, the item-by-item input content and the item-by-item output content of the subsequent artificial intelligence model for intelligent evaluation, i.e., the AI English ability evaluation model, are formatted as binary values, thereby completing the digitization of the artificial intelligence model for intelligent evaluation and implementation. The information analysis mechanism is configured to output, as the multiple related information of the current English composition document, the total number of English characters in the current English composition document, the proportion of repeated characters, the standard deviation of the number of English characters corresponding to each English sentence, the number of English paragraph, and the number of English characters corresponding to the longest English paragraph. Specifically, the standard deviation of the number of English characters corresponding to each English sentence represents the difference between each English sentence. The greater the standard deviation of the number of English characters corresponding to each English sentence, the greater the difference between each English sentence. Therefore, the standard deviation of the number of English characters corresponding to each English sentence is one of the key basic data for subsequent intelligent evaluation. The model building mechanism is configured to perform multiple learning operations on the feedforward neural network to obtain a feedforward neural network after performing the multiple learning operations and output the feedforward neural network as an AI English ability evaluation model. The number of learning operations performed by the feedforward neural network is positively correlated with the number of limited characters of the English composition. For example, when the number of limited characters of the English composition is 3000 characters, the number of learning operations performed by the feedforward neural network is selected to be 500 times. When the number of limited characters of the English composition is 4000 characters, the number of learning operations performed by the feedforward neural network is selected to be 600 times. When the number of limited characters of the English composition is 5000 characters, the number of learning operations performed by the feedforward neural network is selected to be 700 times. When the number of limited characters of the English composition is 6000 characters, the number of learning operations performed by the feedforward neural network is selected to be 800 times, and so on. Specifically, the AI English ability evaluation model adopts a feedforward neural network architecture, and the feedforward neural network includes a single input layer, a single output layer, and multiple hidden layers. The multiple hidden layers are between the single input layer and the single output layer, and the number of hidden layers is proportional to the number of limited characters of the English composition. For example, when the number of limited characters of the English composition is 3000 characters, the number of selected hidden layers is 3. When the number of limited characters of the English composition is 4000 characters, the number of selected hidden layers is 4. When the number of limited characters of the English composition is 5000 characters, the number of selected hidden layers is 5. When the number of limited characters of the English composition is 6000 characters, the number of selected hidden layers is 6, and so on. The synchronous evaluation mechanism is connected with the content conversion mechanism, the information analysis mechanism and the model establishment mechanism respectively, and is used for intelligently evaluating the various sub-item score values and the plagiarism length proportion value of the current English composition document based on the AI English ability evaluation model according to the limited character number of the English composition, the various sub-item score values of the latest past English compositions of the author to which the current English composition document belongs, the multiple related information of the current English composition document and the digital conversion content of the current English composition document; In this way, the present application simultaneously analyzes the various sub-item scores and the plagiarism length data of the current English composition document, and provides important information for the total score evaluation and the plagiarism evaluation of the current English composition document in the future; The AI English ability evaluation model intelligently evaluates the various sub-item score values and the plagiarism length proportion value of the current English composition document according to the limited character number of the English composition, the various sub-item score values of the latest past English compositions of the author to which the current English composition document belongs, the multiple related information of the current English composition document and the digital conversion content of the current English composition document, and includes that the various sub-item score values are five score values corresponding to emotional sufficiency, logic, coherence, text correctness and grammatical accuracy respectively. The fixed number of values is greater than or equal to a set value limit, and when the number of bits of the binary code stream obtained by concatenating the ASCII codes of the English characters appearing in a certain English text at the beginning and the end is less than the fixed number of values, a tail zero padding operation is performed on the binary code stream obtained by concatenating the ASCII codes of the English characters appearing in the certain English text at the beginning and the end to obtain a binary code stream with a fixed number of bits. The fixed number of values is greater than or equal to a set value limit, and when the number of bits of the binary code stream obtained by concatenating the ASCII codes of the English characters appearing in a certain English text at the beginning and the end is less than the fixed number of values, a tail zero padding operation is performed on the binary code stream obtained by concatenating the ASCII codes of the English characters appearing in the certain English text at the beginning and the end to obtain a binary code stream with a fixed number of bits. The AI-based English ability evaluation model intelligently evaluates the sub-item score values and the plagiarism proportion value of the current English composition document based on the English composition limit character number, the sub-item score values of the latest multiple English composition documents of the author to whom the current English composition document belongs, the multiple associated information of the current English composition document, and the digital conversion content of the current English composition document. The multiple English composition documents of the author to whom the current English composition document belongs are multiple English composition documents completed by the author to whom the current English composition document belongs immediately before the current English composition document, and the number of the multiple English composition documents is proportional to the English composition limit character number. For example, when the English composition limit character number is 3000 characters, the number of the multiple English composition documents is 6; when the English composition limit character number is 4000 characters, the number of the multiple English composition documents is 8; when the English composition limit character number is 5000 characters, the number of the multiple English composition documents is 10; when the English composition limit character number is 6000 characters, the number of the multiple English composition documents is 12; and so on. In each learning operation performed on the feedforward neural network, the known sub-item score values and the known plagiarism proportion value of a certain past English composition document are taken as two output contents of the feedforward neural network, and the English composition limit character number, the sub-item score values of the latest multiple English composition documents of the author to whom the certain past English composition document belongs, the multiple associated information of the certain past English composition document, and the digital conversion content of the certain past English composition document are taken as multiple input contents of the feedforward neural network, and the learning operation is completed. Embodiment

[0023] Figure 3 An internal structure diagram of an AI-assisted language learning ability evaluation system according to a second embodiment of the present application is shown.

[0024] As shown in Figure 3 , compared with Figure 2 , the AI-assisted language learning ability evaluation system further comprises: a plagiarism identification mechanism, connected with the synchronous evaluation mechanism, for identifying the current English composition document as a plagiarized composition document when the plagiarism proportion value of the current English composition document is greater than or equal to a preset proportion threshold, and otherwise identifying the current English composition document as a non-plagiarized composition document; For example, when the plagiarism proportion value of the current English composition document is greater than or equal to a preset proportion threshold, the current English composition document is identified as a plagiarized composition document, and otherwise the current English composition document is identified as a non-plagiarized composition document, which includes that the specific value of the plagiarism proportion value is greater than or equal to one-third. Wherein, when the plagiarism proportion of the current English composition document is greater than or equal to the preset proportion threshold, the current English composition document is identified as a plagiarized composition document, otherwise, the current English composition document is identified as a non-plagiarized composition document, comprising: directly setting the total score value of the current English composition document identified as a plagiarized composition document to zero. Embodiment

[0025] Figure 4 An internal structure diagram of the language learning ability evaluation system assisted by artificial intelligence according to a third embodiment of the present application is shown.

[0026] As shown in Figure 4 , compared with Figure 3 , the language learning ability evaluation system assisted by artificial intelligence further comprises: A total score evaluation mechanism connected with the synchronous evaluation mechanism, configured to calculate the total score value of the current English composition document based on the item score values of the current English composition document in an item weighting scoring mode; For example, the total score evaluation mechanism connected with the synchronous evaluation mechanism, configured to calculate the total score value of the current English composition document based on the item score values of the current English composition document in an item weighting scoring mode, comprises: selecting an FPGA chip to realize the total score evaluation mechanism connected with the synchronous evaluation mechanism, configured to calculate the total score value of the current English composition document based on the item score values of the current English composition document in an item weighting scoring mode; Wherein, the total score value of the current English composition document is calculated based on the item score values of the current English composition document in an item weighting scoring mode, comprising: multiplying each item score value of the current English composition document by a respective score weight to obtain a plurality of products, and adding the plurality of products to obtain the total score value of the current English composition document; Wherein, the total score value of the current English composition document is calculated based on the item score values of the current English composition document in an item weighting scoring mode, comprising: each item score value corresponds to a score weight which is a value less than 1, and the cumulative value of the score weights corresponding to each item score value is equal to 1. Embodiment

[0027] Figure 5 An internal structure diagram of the language learning ability evaluation system assisted by artificial intelligence according to a fourth embodiment of the present application is shown.

[0028] As shown in Figure 5 , compared with Figure 4 , the language learning ability evaluation system assisted by artificial intelligence further comprises: The wireless transmission mechanism is connected with the synchronous evaluation mechanism, and is configured to receive the sub-score values and the plagiarism proportion values of the current English composition document, and to input the sub-score values and the plagiarism proportion values of the current English composition document into a network data packet for wireless transmission to a remote evaluation management server. Specifically, the receiving the sub-score values and the plagiarism proportion values of the current English composition document and inputting the sub-score values and the plagiarism proportion values of the current English composition document into a network data packet for wireless transmission to a remote evaluation management server comprises: the network data packet is an IP data packet. Embodiments

[0029] Figure 6 An internal structure diagram of the language learning ability evaluation system based on artificial intelligence assistance according to a fifth embodiment of the present application is shown.

[0030] As shown in Figure 6 Compared with Figure 5 , the language learning ability evaluation system based on artificial intelligence assistance further comprises: The instant display mechanism is connected with the synchronous evaluation mechanism, and is configured to receive the sub-score values and the plagiarism proportion values of the current English composition document, and to synchronously display the sub-score values and the plagiarism proportion values of the current English composition document. For example, a liquid crystal display screen or an LCD display array can be selected to realize the instant display mechanism connected with the synchronous evaluation mechanism, and configured to receive the sub-score values and the plagiarism proportion values of the current English composition document, and to synchronously display the sub-score values and the plagiarism proportion values of the current English composition document.

[0031] Next, the various embodiments of the present application will be further described.

[0032] In the above various embodiments, optionally, in the language learning ability evaluation system based on artificial intelligence assistance: The AI English ability evaluation model intelligently evaluates the sub-score values and the plagiarism proportion values of the current English composition document according to the limited character number of the English composition, the sub-score values of the latest multiple English compositions of the author to whom the current English composition document belongs, the multiple associated information of the current English composition document, and the digital conversion content of the current English composition document, and the method comprises: Specifically, before being input into the AI English ability evaluation model in parallel, the input content that has not been subjected to binary numerical conversion processing needs to be subjected to binary numerical conversion processing, for example, binary numerical conversion processing is performed on the number of limited characters in the English composition, the various sub-item score values of the latest multiple English compositions completed by the author of the current English composition document, and the multiple associated information of the current English composition document, respectively; Among them, based on the AI English ability evaluation model, according to the number of limited characters in the English composition, the various sub-item score values of the latest multiple English compositions completed by the author of the current English composition document, the multiple associated information of the current English composition document, and the digital conversion content of the current English composition document, the intelligent evaluation of the various sub-item score values and the plagiarism length proportion value of the current English composition document includes: executing the AI English ability evaluation model to obtain the various sub-item score values and the plagiarism length proportion value of the current English composition document output by the AI English ability evaluation model; And wherein the number of limited characters in the English composition, the various sub-item score values of the latest multiple English compositions completed by the author of the current English composition document, the multiple associated information of the current English composition document, and the digital conversion content of the current English composition document are input into the AI English ability evaluation model in parallel includes: using a programmable logic device designed in VHDL language to input the number of limited characters in the English composition, the various sub-item score values of the latest multiple English compositions completed by the author of the current English composition document, the multiple associated information of the current English composition document, and the digital conversion content of the current English composition document into the AI English ability evaluation model in parallel.

[0033] And within each of the above embodiments, optionally, in the language learning ability evaluation system assisted by artificial intelligence: The latest multiple English compositions completed by the author of the current English composition document before the current English composition document are multiple English compositions completed by the author of the current English composition document immediately before the current English composition document, and the number of multiple English compositions is directly proportional to the number of limited characters in the English composition includes: using a numerical transformation formula to represent the numerical transformation relationship between the number of multiple English compositions and the number of limited characters in the English composition; For example, the simulation and simulation of the data conversion process of the numerical transformation relationship between the number of multiple English compositions and the number of limited characters in the English composition represented by the numerical transformation formula can be implemented by using a numerical simulation mode; and wherein the ASCII codes of each English character appearing in each sentence of the current English composition document are concatenated to obtain a binary code stream of a fixed number of bits, and each binary code stream corresponding to each sentence of the current English composition document is output as the digital conversion content of the current English composition document; and when the binary code stream obtained by concatenating the ASCII codes of each English character appearing in a certain sentence of the current English composition document has a number of bits equal to the fixed number of bits, the binary code stream obtained by concatenating the ASCII codes of each English character appearing in the certain sentence is directly output as the binary code stream of the fixed number of bits corresponding to the certain sentence. Embodiments

[0034] Figure 7 A step flowchart of an artificial intelligence assisted language learning ability assessment method according to a sixth embodiment of the present application.

[0035] As shown in Figure 7 the artificial intelligence assisted language learning ability assessment method includes the following steps: Step S1: The ASCII codes of each English character appearing in each sentence of the current English composition document are concatenated to obtain a binary code stream of a fixed number of bits, and each binary code stream corresponding to each sentence of the current English composition document is output as the digital conversion content of the current English composition document. Specifically, since the ASCII code of each English character is actually a binary code stream of a fixed number of bits, the item-by-item input content and the item-by-item output content of the subsequent artificial intelligence model for intelligent assessment, i.e., the AI English ability assessment model, are formatted as binary values, thereby completing the digitization of the artificial intelligence model for intelligent assessment and implementation. Step S2: The total number of English characters in the current English composition document, the proportion of repeated characters, the standard deviation of the number of English characters corresponding to each sentence, the number of English paragraph, and the number of English characters corresponding to the longest English paragraph are output as the multi-item related information of the current English composition document. Specifically, the standard deviation of the number of English characters corresponding to each sentence represents the degree of difference between the sentences, and the greater the standard deviation of the number of English characters corresponding to each sentence, the greater the degree of difference between the sentences. Therefore, the standard deviation of the number of English characters corresponding to each sentence is one of the key basic data for subsequent intelligent assessment. Step S3: performing a plurality of learning operations on the feedforward neural network to obtain the feedforward neural network after performing the plurality of learning operations and output as an AI English ability evaluation model, the number of learning operations performed by the feedforward neural network is positively correlated with the number of limited characters of the English composition; For example, the number of learning operations performed by the feedforward neural network is positively correlated with the number of limited characters of the English composition, including: when the number of limited characters of the English composition is 3000 characters, the number of learning operations performed by the feedforward neural network is selected to be 500 times, when the number of limited characters of the English composition is 4000 characters, the number of learning operations performed by the feedforward neural network is selected to be 600 times, when the number of limited characters of the English composition is 5000 characters, the number of learning operations performed by the feedforward neural network is selected to be 700 times, when the number of limited characters of the English composition is 6000 characters, the number of learning operations performed by the feedforward neural network is selected to be 800 times, and so on. Specifically, the AI English ability evaluation model adopts a feedforward neural network architecture, and the feedforward neural network includes a single input layer, a single output layer, and a plurality of hidden layers, the plurality of hidden layers are between the single input layer and the single output layer, and the number of hidden layers is proportional to the number of limited characters of the English composition; For example, the number of hidden layers is proportional to the number of limited characters of the English composition, including: when the number of limited characters of the English composition is 3000 characters, the number of selected hidden layers is 3, when the number of limited characters of the English composition is 4000 characters, the number of selected hidden layers is 4, when the number of limited characters of the English composition is 5000 characters, the number of selected hidden layers is 5, when the number of limited characters of the English composition is 6000 characters, the number of selected hidden layers is 6, and so on. Step S4: intelligently evaluating the plurality of sub-score values and the plagiarism length ratio value of the current English composition document based on the AI English ability evaluation model according to the number of limited characters of the English composition, the plurality of sub-score values of the latest past English compositions of the author to which the current English composition document belongs, the plurality of associated information of the current English composition document, and the digital conversion content of the current English composition document; In this way, the present application simultaneously analyzes the plurality of sub-score values and the plagiarism length data of the current English composition document, providing important information for subsequent total score evaluation and plagiarism evaluation of the current English composition document; Among them, intelligently evaluating the plurality of sub-score values and the plagiarism length ratio value of the current English composition document based on the AI English ability evaluation model according to the number of limited characters of the English composition, the plurality of sub-score values of the latest past English compositions of the author to which the current English composition document belongs, the plurality of associated information of the current English composition document, and the digital conversion content of the current English composition document includes: the plurality of sub-score values are respectively five score values corresponding to emotional sufficiency, logic, coherence, text correctness, and grammatical accuracy. wherein, the ASCII codes of each English character appearing in each sentence of the current English composition document are concatenated to obtain a binary code stream of a fixed number of bits, and the binary code stream of a fixed number of bits of each sentence of the current English composition document is output as the digital conversion content of the current English composition document, and when the number of bits of the binary code stream obtained by concatenating the ASCII codes of each English character appearing in a certain sentence of the current English composition document is less than the fixed number of bits, a tail zero padding operation is performed on the binary code stream obtained by concatenating the ASCII codes of each English character appearing in the certain sentence to obtain a binary code stream of a fixed number of bits; wherein, the ASCII codes of each English character appearing in each sentence of the current English composition document are concatenated to obtain a binary code stream of a fixed number of bits, and the binary code stream of a fixed number of bits of each sentence of the current English composition document is output as the digital conversion content of the current English composition document, and when the number of bits of the binary code stream obtained by concatenating the ASCII codes of each English character appearing in a certain sentence of the current English composition document is greater than the fixed number of bits, a tail truncation operation is performed on the binary code stream obtained by concatenating the ASCII codes of each English character appearing in the certain sentence to obtain a binary code stream of a fixed number of bits; wherein, the AI English ability evaluation model is used to intelligently evaluate the sub-item score values and the plagiarism length proportion value of the current English composition document based on the English composition limit character number, the latest past multiple English composition sub-item score values of the author to whom the current English composition document belongs, the multiple associated information of the current English composition document, and the digital conversion content of the current English composition document, and the multiple English compositions of the author to whom the current English composition document belongs are the multiple English compositions completed by the author to whom the current English composition document belongs immediately before the current English composition document, and the number of the multiple English compositions is proportional to the English composition limit character number; For example, when the English composition limit character number is 3000 characters, the number of the multiple English compositions is 6, when the English composition limit character number is 4000 characters, the number of the multiple English compositions is 8, when the English composition limit character number is 5000 characters, the number of the multiple English compositions is 10, when the English composition limit character number is 6000 characters, the number of the multiple English compositions is 12, and so on. and wherein, in each learning operation performed on the feedforward neural network, the known sub-item score value and the known plagiarism length ratio value of the past certain English composition document are taken as two output contents of the feedforward neural network, the English composition limit character number, the sub-item score values of the past certain English composition document belonging to the latest past multiple English composition documents of the author, the multiple correlation information of the past certain English composition document, and the digital conversion content of the past certain English composition document are taken as multiple input contents of the feedforward neural network, and the learning operation is completed.

[0036] In addition, in the artificial intelligence assisted language learning ability evaluation system and method according to the present application: The multiple learning operations are performed on the feedforward neural network to obtain the feedforward neural network after the multiple learning operations are performed and output as the AI English ability evaluation model, and the number of learning operations performed on the feedforward neural network is positively correlated with the English composition limit character number, which includes using an information mapping function to represent the information mapping relationship between the number of learning operations performed on the feedforward neural network and the English composition limit character number; For example, using the information mapping function to represent the information mapping relationship between the number of learning operations performed on the feedforward neural network and the English composition limit character number includes selecting a MATLAB toolbox to complete the simulation and test of the information mapping function implementation process; In the information mapping function, the English composition limit character number is the input information of the information mapping function. In the information mapping function, the number of learning operations performed on the feedforward neural network positively correlated with the English composition limit character number is the output information of the information mapping function.

[0037] The above is only a specific implementation of the embodiments of the present application, but the protection scope of the embodiments of the present application is not limited thereto, any change or replacement within the technical scope disclosed in the embodiments of the present application should be covered in the protection scope of the embodiments of the present application. Therefore, the protection scope of the embodiments of the present application should be subject to the protection scope of the claims.

Claims

1. A language learning ability assessment system based on artificial intelligence, characterized in that: The system comprises: a content conversion mechanism for concatenating the ASCII codes of the English characters appearing in each English text in the current English composition document to obtain a binary code stream with a fixed number of bits, and outputting the binary code streams corresponding to the English texts in the current English composition document as digitally converted content of the current English composition document; An information parsing unit is configured to output, as multiple pieces of related information of the current English composition document, the total number of English characters, the percentage of repeated characters, the standard deviation of the number of English characters corresponding to each English sentence, the number of English paragraphs, and the number of English characters corresponding to the longest English paragraph; A model building mechanism for performing multiple learning operations on a feedforward neural network to obtain a feedforward neural network after performing the multiple learning operations and outputting the feedforward neural network as an AI English proficiency assessment model, wherein the number of learning operations performed by the feedforward neural network is positively correlated with the number of characters limited to an English composition; The synchronous assessment agency is connected to the content conversion agency, the information analysis agency and the model building agency respectively, and is used to intelligently assess the sub-item score values ​​and the proportion of plagiarized content of the current English composition document based on the AI ​​English proficiency assessment model according to the limited number of characters in the English composition, the sub-item score values ​​of the latest and past English compositions of the author of the current English composition document, multiple related information of the current English composition document, and the digital conversion content of the current English composition document.

2. The artificial intelligence-assisted language learning ability assessment system according to claim 1, characterized in that: Based on the AI ​​English proficiency assessment model, the current English essay document is intelligently assessed based on the character limit of the English essay, the sub-item scores of the author's latest and past English essays, multiple related information of the current English essay document, and the digital conversion content of the current English essay document. The sub-item scores and the proportion of plagiarized content are: the sub-item scores are divided into five scores corresponding to emotional adequacy, logic, coherence, text correctness, and grammatical accuracy; The steps of: concatenating the ASCII codes of each English character appearing in each English text in the current English composition document end to end to obtain a binary code stream of a fixed number of bits; and outputting each binary code stream corresponding to each English text in the current English composition document as the digital conversion content of the current English composition document, including: when the value of the fixed number of bits is greater than or equal to a set value limit, and when the number of bits of the binary code stream obtained by concatenating the ASCII codes of each English character appearing in a certain English text end to end is less than the value of the fixed number of bits, performing a tail zero padding operation on the binary code stream obtained by concatenating the ASCII codes of each English character appearing in the English text end to end to obtain a binary code stream of a fixed number of bits; Among them, the ASCII codes of each English character appearing in each English text in the current English composition document are connected head to tail to obtain a binary code stream with a fixed number of bits, and the binary code streams corresponding to each English text in the current English composition document are output as the digital conversion content of the current English composition document. It also includes: when the number of bits of the binary code stream obtained by connecting the ASCII codes of each English character appearing in a certain English text head to tail is greater than the value of the fixed number of bits, the binary code stream obtained by connecting the ASCII codes of each English character appearing in the English text head to tail is subjected to a tail truncation operation to obtain a binary code stream with a fixed number of bits.

3. The artificial intelligence-assisted language learning ability assessment system according to claim 2, characterized in that: Based on the AI ​​English proficiency assessment model, the current English composition document is intelligently assessed based on the character limit of the English composition, the score values ​​of the latest and previous English compositions of the current English composition document author, multiple related information of the current English composition document, and the digital conversion content of the current English composition document. The score values ​​of the various sub-items and the proportion of plagiarized content of the current English composition document are also included: the latest and previous English compositions of the current English composition document author are multiple English compositions completed by the current English composition document author immediately before the current English composition document, and the number of multiple English compositions is proportional to the character limit of the English composition document; Among them, in each learning operation performed on the feedforward neural network, the known sub-item score values ​​of a past English composition document and the known plagiarized content ratio value are used as the two output contents of the feedforward neural network, and the limited number of characters in the English composition, the sub-item score values ​​of the latest multiple English compositions of the author of the past English composition document, multiple related information of the past English composition document and the digital conversion content of the past English composition document are used as multiple input contents of the feedforward neural network to complete this learning operation.

4. The language learning ability assessment system based on artificial intelligence assistance according to claim 3, characterized in that: The system further comprises: The plagiarism identification agency is connected to the synchronous assessment agency and is used to identify the current English composition document as a plagiarized composition document when the proportion of plagiarized content in the current English composition document is greater than or equal to a preset proportion threshold; otherwise, the current English composition document is identified as a non-plagiarized composition document; Among them, when the proportion of plagiarized content in the current English composition document is greater than or equal to a preset proportion threshold, the current English composition document is identified as a plagiarized composition document. Otherwise, identifying the current English composition document as a non-plagiarized composition document includes: directly setting the total score of the current English composition document identified as a plagiarized composition document to zero.

5. The language learning ability assessment system based on artificial intelligence assistance according to claim 3, characterized in that: The system further comprises: The total score evaluation mechanism is connected to the synchronous evaluation mechanism and is used to calculate the total score value of the current English composition document based on the score values ​​of each sub-item of the current English composition document using a sub-item weighted scoring mode; The method of calculating the total score of the current English composition document based on the score values ​​of each sub-item of the current English composition document using the sub-item weighted scoring mode includes: multiplying the score values ​​of each sub-item of the current English composition document by their respective score weights to obtain respective products, and adding the respective products to obtain the total score of the current English composition document; Among them, the score values ​​of each sub-item of the current English composition document are multiplied by their respective score weights to obtain each product, and each product is added together to obtain the total score value of the current English composition document, including: each score weight corresponding to each sub-item score value is a value less than 1, and the cumulative value of each score weight corresponding to each sub-item score value is equal to 1.

6. The artificial intelligence-assisted language learning ability assessment system according to claim 3, characterized in that: The system further comprises: The wireless uploading mechanism is connected to the synchronous evaluation mechanism, and is used to receive the score values ​​of each sub-item and the percentage of plagiarized text of the current English composition document, and enter the score values ​​of each sub-item and the percentage of plagiarized text of the current English composition document into a network data packet for wireless transmission to a remote evaluation management server.

7. The language learning ability assessment system based on artificial intelligence assistance according to claim 3, characterized in that: The system further comprises: The instant display mechanism is connected to the synchronous evaluation mechanism, and is used to receive the score values ​​of each sub-item and the percentage of plagiarized text of the current English composition document, and synchronously display the score values ​​of each sub-item and the percentage of plagiarized text of the current English composition document.

8. The artificial intelligence-assisted language learning ability assessment system according to any one of claims 3 to 7, characterized in that: Based on the AI ​​English proficiency assessment model, the sub-item score values ​​of the current English composition document and the proportion of plagiarized content are intelligently assessed according to the limited number of characters in the English composition, the latest and previous sub-item score values ​​of the author of the current English composition document, multiple related information of the current English composition document, and the digital conversion content of the current English composition document. The method also includes: inputting the limited number of characters in the English composition, the latest and previous sub-item score values ​​of the author of the current English composition document, multiple related information of the current English composition document, and the digital conversion content of the current English composition document into the AI ​​English proficiency assessment model in parallel; Among them, based on the AI ​​English proficiency assessment model, intelligently assessing the sub-item score values ​​and plagiarized length ratio of the current English composition document according to the limited number of characters in the English composition, the sub-item score values ​​of the latest and past English compositions of the author of the current English composition document, multiple related information of the current English composition document, and the digital conversion content of the current English composition document also includes: executing the AI ​​English proficiency assessment model to obtain the sub-item score values ​​and plagiarized length ratio values ​​of the current English composition document output by the AI ​​English proficiency assessment model; Among them, the character limit for English composition, the sub-item score values ​​of the latest and past English compositions of the author to whom the current English composition document belongs, multiple related information of the current English composition document, and the digital conversion content of the current English composition document are input into the AI ​​English proficiency assessment model in parallel, including: a programmable logic device designed with VHDL language inputs the character limit for English composition, the sub-item score values ​​of the latest and past English compositions of the author to whom the current English composition document belongs, multiple related information of the current English composition document, and the digital conversion content of the current English composition document into the AI ​​English proficiency assessment model in parallel.

9. The artificial intelligence-assisted language learning ability assessment system according to any one of claims 3 to 7, characterized in that: The latest and most recent multiple English compositions of the author of the current English composition document are multiple English compositions completed by the author of the current English composition document immediately before the current English composition document, and the number of the multiple English compositions is proportional to the character limit of the English composition, including: using a numerical conversion formula to express the numerical conversion relationship that the number of the multiple English compositions is proportional to the character limit of the English composition; Among them, the ASCII codes of each English character appearing in each English text in the current English composition document are connected end to end to obtain a binary code stream with a fixed number of bits, and the binary code streams corresponding to each English text in the current English composition document are output as the digital conversion content of the current English composition document. It also includes: when the number of bits of the binary code stream obtained by connecting the ASCII codes of each English character appearing in a certain English text end to end is equal to the value of the fixed number of bits, the binary code stream obtained by directly connecting the ASCII codes of each English character appearing in the English text end to end is used as the binary code stream with a fixed number of bits corresponding to the English text.

10. A method for assessing language learning ability based on artificial intelligence, characterized in that: The method comprises: Connecting the ASCII codes of each English character in each English text in the current English composition document end to end to obtain a binary code stream with a fixed number of bits, and outputting each binary code stream corresponding to each English text in the current English composition document as the digital conversion content of the current English composition document; Outputting the total number of English characters, the percentage of repeated characters, the standard deviation of the number of English characters corresponding to each English sentence, the number of English paragraphs, and the number of English characters corresponding to the longest English paragraph as multiple pieces of related information of the current English composition document; Performing multiple learning operations on the feedforward neural network to obtain a feedforward neural network after performing the multiple learning operations and outputting the feedforward neural network as the AI ​​English proficiency assessment model, wherein the number of learning operations performed by the feedforward neural network is positively correlated with the number of characters limited in the English composition; Based on the AI ​​English proficiency assessment model, the sub-item score values ​​and plagiarism ratio of the current English composition document are intelligently assessed according to the character limit of the English composition, the latest and past English compositions of the author of the current English composition document, multiple related information of the current English composition document, and the digital conversion content of the current English composition document.

Images