Search

CN-122002103-A - Courseware making method and device based on artificial intelligence and storage medium

CN122002103ACN 122002103 ACN122002103 ACN 122002103ACN-122002103-A

Abstract

The invention relates to the technical field of artificial intelligence, in particular to a courseware manufacturing method, a courseware manufacturing device and a storage medium based on artificial intelligence, wherein the method comprises the steps of constructing a tone problem distribution table based on tone accuracy scores and machine bias error labels; the method comprises the steps of determining a target training group based on a tone problem distribution table, further determining courseware generation parameters, extracting a recent continuous training sequence based on the target training group, further generating training intensity adjustment parameters to adjust courseware generation parameters, constructing an individual cognition bias model based on a follow-up reading item in a user history period, generating a bias pair list, generating training content based on the courseware generation parameters, the target training group, the bias pair list and the tone problem distribution table, arranging a teaching flow based on the training content, and generating a courseware script. The invention can dynamically combine pronunciation data, personal error mode and linguistic knowledge base of learner to realize accurate closed loop from data diagnosis to targeted teaching.

Inventors

  • GENG YAOXING

Assignees

  • 北京启兴互动教育科技有限公司

Dates

Publication Date
20260508
Application Date
20260210

Claims (10)

  1. 1. The courseware manufacturing method based on artificial intelligence is characterized by comprising the following steps of: constructing a tone problem distribution table based on the tone accuracy score and the machine bias error label; determining a target training group based on a tone problem distribution table, further determining courseware generation parameters, extracting a recent continuous training sequence based on the target training group, further generating training intensity adjustment parameters, and adjusting the courseware generation parameters; constructing an individual cognition bias model based on the read-following items in the user history period, and generating a bias pair list; Generating training content based on courseware generation parameters, target training groups, a bias error pair list and a tone problem distribution table; and arranging a teaching flow based on the training content, and generating a courseware script.
  2. 2. The artificial intelligence based courseware making method of claim 1, wherein reading follow-up entries in a user history period, grouping according to two dimensions of a target tone and a tone position, wherein the target tone comprises one tone, two tones, three tones, four tones and light tone, and the syllable position comprises a single word, a double word first word and a double word last word; for each group, calculating the average value of the tone accuracy scores of all the follow-up entries under the group, and marking the average value as an average accuracy value Zp; Counting the proportion of the number of the follow-up entries with the tone accuracy score lower than a preset accuracy threshold value in the grouping to the total follow-up entries, and marking the proportion as an error occurrence rate Cf; Mapping machine error labels of all items with tone accuracy scores lower than a preset accuracy threshold under the group into specific error types, and counting the error type with the largest occurrence number as the main error type of the group; The average accuracy value, the error occurrence rate and the main error type are used as a tone problem distribution table.
  3. 3. The courseware making method based on artificial intelligence according to claim 2, wherein the error occurrence rate of all the groups in the tone problem distribution table is ranked from high to low, each group in the ranking is filtered, and after the filtering, the group with the highest error occurrence rate is selected and is determined as the target training group generated by the courseware; the filtering process is as follows: If the average accuracy value of the group is more than or equal to a preset accuracy filtering threshold value or the training state corresponding to the group in the teaching progress schedule is unopened, filtering the group, otherwise, not filtering the group; and calculating courseware generating parameters based on the average accuracy value A of the target training group, wherein the courseware generating parameters comprise a comparison word pair number N and a training form grade L.
  4. 4. A courseware making method based on artificial intelligence according to claim 3, wherein all the follow-up entries of the user for the target training group in the history period are extracted, the follow-up entries are arranged in ascending order according to the time stamps thereof to form a near-term continuous training sequence, and the length of the near-term continuous training sequence is denoted as Nt; When Nt is greater than a preset length, dividing a continuous exercise sequence meeting the conditions into three continuous subsequences in time sequence, wherein the subsequences comprise an early stage, a middle stage and a later stage, the early stage is a front floor (Nt/3) follow-up entries, the middle stage is a next floor (Nt/3) follow-up entries, the later stage is all the remaining entries, and floor () is a downward rounding function; respectively calculating the average value of the tone accuracy scores of all the entries in each subsequence, and recording the average value as Me, mm and Ml; Calculating an average difference delta m between the later stage and the earlier stage, wherein delta m=ml-Me; when delta m is smaller than a preset difference value, judging that a fatigue attenuation trend exists, otherwise, judging that no fatigue attenuation trend exists; When there is a fatigue decay trend, setting the training intensity adjustment parameter U to [1-min (1, max (0, (- [ delta ] m)/my)) ], my being a preset difference threshold; when no fatigue attenuation trend exists or Nt is smaller than or equal to a preset length, setting a training intensity adjusting parameter U to be 1; And adjusting the comparison word pair number N based on the training intensity parameter U, and setting the adjusted comparison word pair number N to be N1, wherein N1=max (2, ceil (N multiplied by U)), and the ceil () function is an upward rounding function.
  5. 5. The courseware making method based on artificial intelligence according to claim 4, wherein the following entries with the following entry tone accuracy score lower than 80 score in the user history period are extracted as sample entries, and for each sample entry, the target tone i of the sample entry is paired with the actual tone j according to the error type mapped by the machine bias label, all the paired bits are counted, and a 5×5 individual tone confusion matrix P is generated, wherein the calculation formula of each element Pij in the matrix is that pij=target tone is i and the number of entries which are missent as j/the total number of error entries whose target tone is i; Comparing the individual tone confusion matrix P with a tone confusion probability matrix M in a linguistic knowledge base element by element, and judging that the item (i, j) is the significance cognitive bias of the user if and only if Pij-Mij > Deltak, and Deltak is a preset significance threshold; All the significant cognitive errors (i, j) meeting the conditions are output as a list which is recorded as an error pair list E, and the list is arranged in descending order of the values of Pij-Mij.
  6. 6. The method for making courseware based on artificial intelligence according to claim 5, wherein according to the main error type and the bias error pair list E of the target training group, the word pairs matched with the items in the bias error pair list E are combined from the tone pairs of the hierarchical minimum tone pairs to form a high-priority candidate set; And for each group of word pairs in the core comparison word pair set, calling a corresponding template to generate a specific exercise sentence according to the form specified by the exercise form level L.
  7. 7. The courseware making method based on artificial intelligence according to claim 6, wherein the generating method of the training sentence comprises that if l=1, the training sentence is the word pair itself; If l=2, each word in the word pair is respectively embedded into a high-frequency double-word template to generate double words; if l=3, the words in the word pair are embedded into a preset simple sentence pattern template, and a simple sentence is generated.
  8. 8. The method for making courseware based on artificial intelligence according to claim 7, wherein training contents are sequentially arranged into independent basic teaching units according to the sequence of the training contents in the word pair set, a standardized interaction flow is designed for each basic teaching unit, and automatic scoring is performed after the training links follow up the training links, wherein the automatic scoring process is as follows: Acquiring the tone accuracy score of the learner for reading the current exercise sentence, recording as Sc, judging that the current exercise is mastered if Sc is greater than or equal to a first preset score, and automatically entering a next basic teaching unit in the sequence; if Sc is larger than or equal to the second preset score and smaller than the first preset score, judging that the current exercise is in a partial mastering state, feeding back a 'partial correct, please notice' prompt message, and automatically entering a next basic teaching unit in the process; and integrating the sequences of all the basic teaching units and the automatic scoring logic defined by the sequences to generate a courseware script.
  9. 9. An artificial intelligence based courseware making device, applied to an artificial intelligence based courseware making method according to any one of claims 1-8, comprising: The problem determining unit is used for constructing a tone problem distribution table based on the tone accuracy score and the machine bias error label; the parameter determining unit is used for determining a target training group based on the tone problem distribution table, further determining courseware generating parameters, extracting a recent continuous training sequence based on the target training group, further generating training intensity adjusting parameters, and adjusting the courseware generating parameters; the bias error pair determining unit is used for constructing an individual cognition bias error model based on the read-following items in the user history period and generating a bias error pair list; the training content generation unit is used for generating training content based on courseware generation parameters, target training groups, a bias error pair list and a tone problem distribution table; And the courseware generating unit is used for arranging the teaching flow based on the training content and generating a courseware script.
  10. 10. A computer readable storage medium, wherein the computer readable storage medium stores a computer program, and wherein the computer program is configured to control an electronic device in which the computer readable storage medium is located to execute the method for making an artificial intelligence-based courseware according to any one of claims 1 to 8 when the computer program is executed.

Description

Courseware making method and device based on artificial intelligence and storage medium Technical Field The invention relates to the technical field of artificial intelligence, in particular to a courseware manufacturing method and device based on artificial intelligence and a storage medium. Background In primary Chinese online teaching for English native language adult learners, tone mastering is a major core difficulty. The traditional online teaching platform has the following limitations in tone training that firstly, most of training materials are fixed and unified courseware, pertinence to individual pronunciation errors of learners is lacking, secondly, the training platform relies on teachers to manually listen and judge errors, is low in efficiency and difficult to scale, thirdly, feedback is lagged, and immediate and accurate correction cannot be provided during training. While some platforms introduce automatic speech scoring, most provide only simple scores, fail to diagnose error types deep and generate progressive training content therefrom. Therefore, a solution capable of automatically analyzing pronunciation errors of learners, dynamically generating personalized training courseware and realizing intelligent teaching flow management is needed to improve the effect and efficiency of tone teaching. Disclosure of Invention The invention aims to provide a courseware manufacturing method and device based on artificial intelligence and a storage medium, so as to solve at least one of the problems in the prior art. In order to achieve the above purpose, the invention adopts the following technical scheme: a courseware making method based on artificial intelligence comprises the following steps: constructing a tone problem distribution table based on the tone accuracy score and the machine bias error label; determining a target training group based on a tone problem distribution table, further determining courseware generation parameters, extracting a recent continuous training sequence based on the target training group, further generating training intensity adjustment parameters, and adjusting the courseware generation parameters; constructing an individual cognition bias model based on the read-following items in the user history period, and generating a bias pair list; Generating training content based on courseware generation parameters, target training groups, a bias error pair list and a tone problem distribution table; and arranging a teaching flow based on the training content, and generating a courseware script. Further, reading the follow-up entries in the history period of the user, and grouping according to two dimensions of a target tone and a syllable position, wherein the target tone comprises one tone, two tones, three tones, four tones and light tone, and the syllable position comprises a single word, a double-word first word and a double-word last word; for each group, calculating the average value of the tone accuracy scores of all the follow-up entries under the group, and marking the average value as an average accuracy value Zp; Counting the proportion of the number of the follow-up entries with the tone accuracy score lower than a preset accuracy threshold value in the grouping to the total follow-up entries, and marking the proportion as an error occurrence rate Cf; Mapping machine error labels of all items with tone accuracy scores lower than a preset accuracy threshold under the group into specific error types, and counting the error type with the largest occurrence number as the main error type of the group; The average accuracy value, the error occurrence rate and the main error type are used as a tone problem distribution table. Further, the error occurrence rate of all the groups in the tone problem distribution table is sequenced from high to low, each group in the sequencing is filtered, and after the filtering, the group with the highest error occurrence rate is selected and is determined to be the target training group generated by the courseware; the filtering process is as follows: If the average accuracy value of the group is more than or equal to a preset accuracy filtering threshold value or the training state corresponding to the group in the teaching progress schedule is unopened, filtering the group, otherwise, not filtering the group; and calculating courseware generating parameters based on the average accuracy value A of the target training group, wherein the courseware generating parameters comprise a comparison word pair number N and a training form grade L. Further, extracting all read-following items of a target training packet in a history period of a user, arranging the read-following items in ascending order according to the time stamps of the read-following items to form a near-term continuous training sequence, and marking the length of the near-term continuous training sequence as Nt; When Nt is greater than a preset length, dividing a contin