Teaching interactive content generation method, device, equipment and computer program product

Abstract

The present invention provides a teaching interactive content generation method, device, equipment and computer program product, which relates to the field of computer technology. The method includes: obtaining a teaching prediction background according to previously acquired teaching target information; obtaining at least one teaching prediction content according to the teaching prediction background, the teaching prediction content includes a teaching prediction task and a problem prediction chain corresponding to the teaching prediction task, and the problem prediction chain is constructed based on at least one prediction problem obtained by the corresponding prediction of the teaching prediction task; according to the problem prediction chain, searching a previously constructed knowledge base to obtain a target question-answer chain matching the question prediction chain; using the question-answer chain matching the question prediction chain, replacing the question prediction chain to obtain corresponding teaching update content; according to the teaching update content, identifying the task association between each teaching prediction task and the corresponding question-answer chain in the teaching update content to obtain teaching interactive content. The present invention can ensure the rationality of teaching interactive content.

Description

Technical Field

[0001] The present invention relates to the field of computer technology, and in particular to a method, device, equipment and computer program product for generating interactive teaching content. Background Art

[0002] Teaching interaction refers to the information exchange, viewpoint sharing and thought collision carried out through various forms in the teaching process. It is a key link in the teaching process and provides a practical foothold for guiding and inspiring students to understand knowledge. It is of great significance in heuristic teaching design.

[0003] At present, teaching interaction mainly generates background information and question chains through designing prompts, which call large models to generate background information, learning tasks and question chains that meet teaching objectives. However, the question chains obtained in this way will deviate from the teaching focus and are highly repetitive, and the accuracy, language and richness of the answers to the questions are poor.

[0004] But at the same time, it also faces the test of quality control, logic and teaching relevance of content generation. Summary of the invention

[0005] The present invention provides a method, device, equipment and computer program product for generating interactive teaching content, which are used to solve the defects in the prior art that the question chain deviates from the teaching focus and the accuracy of the question answers is poor, and can generate logical, accurate and rich interactive teaching content on a large scale.

[0006] The present invention provides a method for generating interactive teaching content, comprising: obtaining teaching target information; obtaining a teaching prediction background according to the teaching target information; obtaining at least one teaching prediction content according to the teaching prediction background, the teaching prediction content comprising a teaching prediction task and a problem prediction chain corresponding to the teaching prediction task, the problem prediction chain being constructed based on at least one predicted problem obtained by a corresponding prediction of the teaching prediction task; searching a knowledge base according to the problem prediction chain to obtain a target question-answer chain matching the problem prediction chain, the knowledge base being constructed based on problems determined in advance based on historical teaching content, analysis and interpretation content corresponding to the problems, and answers to corresponding problems determined based on the analysis and interpretation content; replacing the problem prediction chain with the question-answer chain to obtain corresponding teaching update content; identifying the task association between each teaching prediction task in the teaching update content and the corresponding question-answer chain according to the teaching update content, to obtain interactive teaching content.

[0007] According to a teaching interactive content generation method provided by the present invention, the knowledge base includes at least one configuration file corresponding to the historical teaching content and stored in a preset format, the configuration file is obtained based on the question field, the analysis and interpretation field and the answer field respectively storing the question corresponding to the teaching content, the analysis and interpretation content and the answer; according to the question prediction chain, the knowledge base is searched to obtain a target question and answer chain matching the question prediction chain, including: keyword extraction is performed on each predicted question in the question prediction chain to obtain corresponding question keywords; according to the question keywords, the knowledge base is searched in combination with the target index to obtain the target question and answer chain; wherein the target index is first obtained by creating a mapping relationship based on the target teaching content corresponding to the teaching target information and the configuration file corresponding to the historical teaching content.

[0008] According to a teaching interactive content generation method provided by the present invention, according to the question keywords, combined with the target index, the knowledge base is searched to obtain the target question and answer chain, including: according to the question keywords, combined with the target index, the question field of the corresponding configuration file in the knowledge base is searched, and the question keywords are matched with the questions stored in the corresponding question field obtained by the search to obtain the target question; according to the target question, the target analysis and interpretation content corresponding to the target question in the analysis and interpretation field of the corresponding configuration file is obtained; according to the target question, the target answer corresponding to the target question in the answer field of the corresponding configuration file is obtained; according to the target question, the target analysis and interpretation content and the target answer are obtained to obtain the target question and answer chain.

[0009] According to a teaching interactive content generation method provided by the present invention, after obtaining a target question-answer chain, the method includes: deleting a target index.

[0010] According to a teaching interactive content generation method provided by the present invention, before searching the knowledge base according to the question prediction chain, the method includes: obtaining historical teaching content; performing text segmentation on the historical teaching content to obtain multiple text paragraphs corresponding to the historical teaching content; inputting the multiple text paragraphs corresponding to the historical teaching content into the knowledge prediction model to obtain the questions corresponding to each text paragraph output by the knowledge prediction model, the analysis and interpretation content corresponding to the questions, and the answers to the corresponding questions determined based on the analysis and interpretation content; wherein the knowledge prediction model is trained based on teaching content training data and the question labels, interpretation content labels and answer labels corresponding to the teaching content training data; the knowledge prediction model is used to predict the questions, the analysis and interpretation content corresponding to the questions, and the answers to the questions for each text paragraph corresponding to the input historical teaching content; the questions corresponding to each text paragraph, the analysis and interpretation content corresponding to the questions, and the answers to the corresponding questions determined based on the analysis and interpretation content are stored in the database as a corresponding question-answer chain.

[0011] According to a teaching interactive content generation method provided by the present invention, the knowledge prediction model includes a feature extraction layer, a question prediction layer, an interpretation analysis layer and an answer prediction layer; a plurality of text paragraphs corresponding to the history teaching content are input into the knowledge prediction model to obtain the questions corresponding to each text paragraph output by the knowledge prediction model, the analysis and interpretation content corresponding to the questions and the answers to the corresponding questions determined based on the analysis and interpretation content, including: inputting a plurality of text paragraphs corresponding to the history teaching content into the feature extraction layer for feature extraction to obtain the text features corresponding to each text paragraph; inputting the text features corresponding to each text paragraph into the question prediction layer for question prediction to obtain the questions corresponding to each text paragraph; inputting the questions corresponding to each text paragraph into the interpretation analysis layer to predict the corresponding text learning results and teaching key point analysis results for each text paragraph, and predicting the corresponding teaching text analysis results for each text paragraph and the questions of each text paragraph, and obtaining the analysis and interpretation content of the corresponding text paragraph according to the predicted text learning results, teaching key point analysis results and teaching text analysis results; inputting the analysis and interpretation content of each text paragraph and the questions corresponding to each text paragraph into the answer prediction layer for answer prediction to obtain the answers corresponding to each text paragraph output by the answer prediction layer.

[0012] According to a teaching interactive content generation method provided by the present invention, before the questions corresponding to each text paragraph, the analysis and interpretation content corresponding to the questions, and the answers to the corresponding questions determined based on the analysis and interpretation content are stored in a database as a corresponding question-answer chain, the method includes: inputting the questions corresponding to each text paragraph into a question detection model to obtain a question check result output by the question detection model, and the question check result is used to characterize whether the corresponding question inquiry object is the target object; wherein the question detection model is trained based on historical questions asked about the target object, historical questions asked about other objects different from the target object, and historical question labels corresponding to the historical questions; if the corresponding question inquiry object is the target object based on the question check result, it is retained; otherwise, multiple text paragraphs corresponding to the historical teaching content are re-input into the knowledge prediction model for prediction.

[0013] The present invention also provides a teaching interactive content generation device, comprising: a target acquisition module, which acquires teaching target information; a background prediction module, which obtains a teaching prediction background according to the teaching target information; a content prediction module, which obtains at least one teaching prediction content according to the teaching prediction background, wherein the teaching prediction content includes a teaching prediction task and a problem prediction chain corresponding to the teaching prediction task, wherein the problem prediction chain is constructed based on at least one predicted problem corresponding to the prediction of the teaching prediction task; a search module, which searches a knowledge base according to the problem prediction chain to obtain a target question-answer chain matching the problem prediction chain, wherein the knowledge base is constructed based on a problem determined based on historical teaching content, an analysis and interpretation content corresponding to the problem, and an answer corresponding to the problem determined based on the analysis and interpretation content; a replacement module, which replaces the question prediction chain with a question-answer chain matching the question prediction chain to obtain corresponding teaching update content; and a content generation module, which identifies the task association between each teaching prediction task and the corresponding question-answer chain in the teaching update content according to the teaching update content to obtain teaching interactive content.

[0014] The present invention also provides an electronic device, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein when the processor executes the computer program, the method for generating interactive teaching content as described above is implemented.

[0015] The present invention also provides a non-transitory computer-readable storage medium on which a computer program is stored. When the computer program is executed by a processor, the teaching interactive content generation method as described in any one of the above is implemented.

[0016] The present invention also provides a computer program product, including a computer program, which, when executed by a processor, implements any of the above-mentioned teaching interactive content generation methods.

[0017] The teaching interactive content generation method, device, equipment and computer program product provided by the present invention obtain teaching target information so that each subsequent step of operation is derived and promoted based on this, ensuring that all teaching predictions, content updates and interactive designs are closely aligned with the teaching targets set initially, avoiding teaching activities from deviating from the expected direction; obtaining the teaching prediction background through the teaching target information to preliminarily predict the teaching prediction content, and using the previously constructed knowledge base to quickly find the target question-answer chain related to the current prediction question chain to replace the question prediction chain of the teaching prediction content, thereby continuously optimizing and improving on the basis of previous teaching, ensuring that the teaching content always remains fresh and targeted, better meeting the needs that may arise in different learning stages, avoiding the obsolescence and rigidity of the teaching content, and then identifying the task association between each teaching prediction task and the corresponding question-answer chain according to the teaching update content, determining the teaching interactive content, ensuring that the interaction is more personalized, and ensuring the rationality of the teaching interactive content, thereby improving the teaching quality. BRIEF DESCRIPTION OF THE DRAWINGS

[0018] In order to more clearly illustrate the technical solutions in the present invention or the prior art, the following briefly introduces the drawings required for use in the embodiments or the description of the prior art. Obviously, the drawings described below are some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.

[0019] Figure 1 This is one of the flow charts of the teaching interactive content generation method provided by the present invention;

[0020] Figure 2 This is the second flow chart of the teaching interactive content generation method provided by the present invention;

[0021] Figure 3 It is a structural schematic diagram of the teaching interactive content generating device provided by the present invention;

[0022] Figure 4 It is a structural schematic diagram of the electronic device provided by the present invention. DETAILED DESCRIPTION

[0023] In order to make the purpose, technical solution and advantages of the present invention clearer, the technical solution of the present invention will be clearly and completely described below in conjunction with the drawings of the present invention. Obviously, the described embodiments are part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0024] Figure 1It is a flow chart of the teaching interactive content generation method provided by the present invention, such as Figure 1 As shown, the method includes:

[0025] S11, obtaining teaching target information;

[0026] S12, obtaining teaching prediction background according to teaching target information;

[0027] S13, obtaining at least one teaching prediction content according to the teaching prediction background, the teaching prediction content including a teaching prediction task and a problem prediction chain corresponding to the teaching prediction task, the problem prediction chain being constructed based on at least one prediction problem obtained by the prediction corresponding to the teaching prediction task;

[0028] S14, according to the question prediction chain, searching the knowledge base to obtain the target question-answer chain matching the question prediction chain, the knowledge base is constructed based on the questions determined based on the history teaching content, the analysis and interpretation content corresponding to the questions, and the answers to the corresponding questions determined based on the analysis and interpretation content;

[0029] S15, using the question-answer chain that matches the question prediction chain, replacing the question prediction chain to obtain corresponding teaching update content;

[0030] S16, according to the teaching update content, identifying the task association between each teaching prediction task in the teaching update content and the corresponding question-answer chain, and obtaining the teaching interaction content.

[0031] It should be noted that the step numbers "S1N" in this manual do not represent the order of the method for generating teaching interactive content. Figure 2 The teaching interactive content generating method of the present invention is described.

[0032] Step S11, obtaining teaching target information.

[0033] It should be added that the teaching objective information includes the name of the target teaching content and the content generation prompt for generating teaching interactive content. The name of the target teaching content is determined based on the target teaching content. For example, if the target teaching content is an article, the corresponding name of the target teaching content is the article name.

[0034] In addition, the teaching objective information also includes a background generation prompt, a task content generation prompt, and a task association prompt. The background generation prompt, the task content generation prompt, and the task association prompt can be set according to actual teaching needs, and no further limitation is made here.

[0035] Step S12, obtaining the teaching prediction background according to the teaching target information.

[0036] In this embodiment, a teaching prediction background is obtained according to the teaching objective information, including: inputting the name of the target teaching content and the background generation prompt into the background prediction model to obtain the teaching prediction background output by the background prediction model; wherein the background prediction model is trained according to the teaching content training data and the background labels corresponding to the teaching content training data.

[0037] It should be added that background information includes the environment in which people interact with objects or with each other in the target teaching content and the basis for solving problems. The basis can include pictures, charts, and texts, etc. The basis can come from the target teaching content or from other sources of knowledge.

[0038] For example, the background generation prompt can be: You are a language expert, and you are teaching Yu Dafu's "Autumn in the Old Capital". Please generate a context corresponding to this text according to the connotation of the context. The context connotation is as follows: {qingjingneihan}, and the example context is as follows: {few_shot_context}.

[0039] Accordingly, few_shot_background information can be:

[0040] "Since ancient times, many men of letters have been sad about spring and autumn. Autumn is one of the most common and typical images in Chinese literature. Autumn contains rich emotions. When we mention autumn, we will think of the bleakness of "boundless fallen leaves rustling down, the endless Yangtze River rolling in" in Du Fu's "Climbing the High Tower", and the grandeur of "since ancient times, autumn has been sad and lonely, I say autumn is better than spring" in Liu Yuxi's "Autumn Lyrics". It can be seen that the scenery moves with the emotions, and the scenery described by the writer is actually the sustenance of the writer's personal emotions. Reading literary works is actually reading the author's life. Today, we will walk into Yu Dafu's "Autumn in the Old Capital" together, feel the unique emotions that Yu Dafu placed in "Autumn in the Old Capital", and walk through Yu Dafu's life together.

[0041] As autumn approaches, the school has launched a reading sharing activity with the theme of "autumn". Students need to read works related to "autumn" and write their reflections. Excellent works will be compiled into a booklet. We will take Yu Dafu's essay "Autumn in the Old Capital" as an example to learn how to transform our reading and thinking into a focused, layered and in-depth reflection. "

[0042] In addition, before inputting the name of the target teaching content and the background generation prompt into the background prediction model, it includes: obtaining the teaching content training data and the background labels corresponding to the teaching content training data; using the teaching content training data as the input data for training, and using the background labels corresponding to the teaching content training data as the labels for training, and training the background prediction model to be trained until the model converges or reaches a preset number of iterations, thereby obtaining a trained background prediction model.

[0043] Step S13, obtaining at least one teaching prediction content according to the teaching prediction background, wherein the teaching prediction content includes a teaching prediction task and a problem prediction chain corresponding to the teaching prediction task, wherein the problem prediction chain is constructed based on at least one prediction problem obtained by the corresponding prediction of the teaching prediction task.

[0044] In this embodiment, at least one teaching prediction content is obtained based on the teaching prediction background, including: inputting the teaching prediction background and the task content generation prompt into the teaching content prediction model, and obtaining at least one teaching prediction content output by the teaching content prediction model; wherein the teaching content prediction model is trained based on the background training data and the task labels and question labels corresponding to the background training data.

[0045] It should be added that the content generation prompts can be set according to the actual required teaching prediction content, for example, three, and each teaching prediction content corresponds to one teaching prediction task, and each teaching prediction task corresponds to three prediction questions.

[0046] For example, the content generation prompt can be: Now you are a language expert, teaching Yu Dafu's "Autumn in the Old Capital". Please generate three learning tasks based on the context, and each learning task corresponds to three questions. The context is as follows: {qingjing}, and the example learning task and question chain are as follows: {few_shot_learning task and question chain}.

[0047] Accordingly, the teaching forecast content is:

[0048] "Teaching prediction task 1: Write a reading note on "Appreciation of Scenery Description"

[0049] Prediction question 1. In “Autumn in the Old Capital”, which scene description impressed you the most and why?

[0050] Prediction Question 2. How does the author use language to express the beauty of these scenes?

[0051] Prediction Question 3. What role do these scene descriptions play in the entire work, and what impact do they have on the construction of the story context?

[0052] Teaching prediction task 2: Write a reading note on "the method of integrating situation and context"

[0053] Prediction Question 1. In “Autumn in the Old Capital”, how are emotions and scenery intertwined?

[0054] Prediction Question 2. Provide an example of contextualization and explain how it enhances the emotional expression of the text.

[0055] Prediction Question 3. What impression does this fusion of situation and emotion leave on you?

[0056] Teaching prediction task three: Write a reading note on "emotional understanding of the work"

[0057] Prediction Question 1. Analyze which emotional expressions in "Autumn in the Old Capital" may be related to the author Yu Dafu's life background.

[0058] Prediction Question 2. How important are these emotional expressions to understanding the entire work?

[0059] Prediction Question 3. How did you feel the emotional dynamics in the work? How did these emotions interact with the author’s writing style?

[0060] In addition, before the teaching prediction background and task content generation prompts are input into the teaching content prediction model, it includes: obtaining background training data and task labels and question labels corresponding to the background training data; using the background training data as input data for training, using the task labels and question labels corresponding to the background training data as labels for training, training the teaching content prediction model to be trained until the model converges or reaches a preset number of iterations, and obtaining a trained teaching content prediction model.

[0061] In an optional embodiment, before searching the knowledge base according to the question prediction chain, the process includes: obtaining historical teaching content; performing text segmentation on the historical teaching content to obtain multiple text paragraphs corresponding to the historical teaching content; inputting the multiple text paragraphs corresponding to the historical teaching content into the knowledge prediction model to obtain the questions corresponding to each text paragraph output by the knowledge prediction model, the analysis and interpretation content corresponding to the questions, and the answers to the corresponding questions determined based on the analysis and interpretation content; wherein the knowledge prediction model is trained based on the teaching content training data and the question labels, interpretation content labels and answer labels corresponding to the teaching content training data; the knowledge prediction model is used to predict the questions, the analysis and interpretation content corresponding to the questions, and the answers to the questions for each text paragraph corresponding to the input historical teaching content; the questions corresponding to each text paragraph, the analysis and interpretation content corresponding to the questions, and the answers to the corresponding questions determined based on the analysis and interpretation content are stored in the database as a corresponding question and answer (QA) chain.

[0062] Specifically, the text segmentation of the history teaching content is performed to obtain multiple text paragraphs corresponding to the history teaching content, including: identifying the history teaching content, and segmenting the history teaching content according to the preset number of words to obtain multiple text paragraphs; calling the semantic completion model, combining the history teaching content corresponding to the corresponding text paragraph, and semantically completing each text paragraph, so that the content of each text paragraph has a more complete semantic structure. It should be noted that the semantic completion model can use a general large model, which is not further explained here; the preset number of words can be set according to the actual number of words required for the text paragraph, for example, if each paragraph of text requires about 300 words, the preset number of words can be 300 words, which is not further limited here.

[0063] In addition, the knowledge prediction model includes a feature extraction layer, a question prediction layer, an interpretation analysis layer and an answer prediction layer; multiple text paragraphs corresponding to the history teaching content are input into the knowledge prediction model to obtain the questions corresponding to each text paragraph output by the knowledge prediction model, the analysis and interpretation content corresponding to the questions, and the answers to the corresponding questions determined based on the analysis and interpretation content, including: inputting multiple text paragraphs corresponding to the history teaching content into the feature extraction layer for feature extraction to obtain the text features corresponding to each text paragraph; inputting the text features corresponding to each text paragraph into the question prediction layer for question prediction to obtain the questions corresponding to each text paragraph; inputting the questions corresponding to each text paragraph into the interpretation analysis layer to predict the corresponding text learning results and teaching key points analysis results for each text paragraph, and predicting the corresponding teaching text analysis results for each text paragraph and the questions of each text paragraph, and obtaining the analysis and interpretation content of the corresponding text paragraph based on the predicted text learning results, teaching key points analysis results and teaching text analysis results; inputting the analysis and interpretation content of each text paragraph and the questions corresponding to each text paragraph into the answer prediction layer for answer prediction to obtain the answers corresponding to each text paragraph output by the answer prediction layer.

[0064] It should be noted that the text learning results represent the text knowledge in the teaching content as well as discourse reading skills and summarization skills; the teaching points analysis results are obtained by understanding and analyzing the meaning of the problem based on the combination of the text’s internal and external content and the context. For example, the question is: In what order does Yu Dafu describe the autumn scenery? Correspondingly, the teaching points analysis results are: First, you need to understand what the order means. By combining the text’s internal and external content and the context, you can understand and analyze the writing order of the article as a whole, and conduct a comprehensive and thorough analysis to truly understand the writing order of the article. Students not only need to answer other related questions: What scenery is described in "Autumn in the Ancient Capital", but also need to analyze the characteristics of the scenery and the writing order from a macro perspective. The text content corresponding to the problem of the teaching text analysis results is very rich, including both features extracted directly from the text and writing logic deeply excavated from the text.

[0065] In addition, if the knowledge prediction model adopts a general large language model, accordingly, before inputting multiple text paragraphs corresponding to the history teaching content into the knowledge prediction model, a question prompt, a text learning prompt, a key point analysis prompt, a text analysis prompt and an answer prompt are also configured. The question prompt can refer to the above description, and the text learning prompt, the key point analysis prompt and the text analysis prompt can be configured according to the actual required text learning results, the teaching key point analysis results and the teaching text analysis results. The corresponding question, result or answer obtained according to the prompt can be determined based on the knowledge prediction model obtained by actual training, and no further limitation is made here.

[0066] For example, if the history teaching content is "Autumn in the Ancient Capital", the question prompt can be: Now you are a language expert, teaching Yu Dafu's "Autumn in the Ancient Capital". Now you need to generate questions based on the appreciation fragment. Note that the questions should be closely related to "Autumn in the Ancient Capital". The questions need to be in-depth and meaningful, and can guide students to think about the content of the appreciation fragment. Questions are for students, and students need to answer them. Do not generate questions about teaching. Appreciation fragment: {shangxipianduan}, question example: {few_shot_question example}.

[0067] Correspondingly, the questions can be: "1. Read the whole text, sort out the autumn scenes selected in "Autumn in the Old Capital", and try to summarize them in the form of "×× picture". 2. What kind of feelings does Yu Dafu have about the autumn in Beijing? Judging from the overall tone of the article, what is the author's mood? 3. Combined with the text, analyze what descriptive techniques are used in "Autumn in the Old Capital" to deal with the relationship between scenery and emotions, and what kind of effect do they have? 4. Appreciate the key sentence "But, the autumn in the north is particularly clear, quiet, and sad", summarize what characteristics of the autumn in the old capital? 5. Why did the author travel thousands of miles to the embrace of the autumn in the north to taste the flavor of autumn? Try to analyze from the perspective of the contrast between the autumn in the south and the autumn in the north, and the perspective of the writing background."

[0068] For another example, the text learning prompt can be: Now you are a Chinese language expert, and you are teaching Yu Dafu's "Autumn in the Old Capital". Based on the expected Chinese language learning results of "Autumn in the Old Capital", please analyze what abilities students can improve through problems and analysis and explanations, and what kind of learning effects can be achieved. The problems are as follows: {wenti}, the analysis and explanations are as follows: {fenxichanshi}, and the expected learning effects of "Autumn in the Old Capital" are as follows: {few_shot_Chinese language learning results}. Pay attention to the correlation between Chinese language learning results and problems and analysis and explanations. Do not output problems and analysis and explanations. Summarize them in paragraphs, no more than 3 paragraphs, and about 50 words.

[0069] Accordingly, for question 1 above, the text learning results can be: 1. Comparative reading ability: understand and analyze the role of contrast in expressing emotions, and improve sensitivity to text structure. 2. Emotional grasp and interpretation: learn to understand the author's complex emotional changes between the lines, and cultivate the ability to deeply interpret the text. 3. Improve writing skills: learn from Yu Dafu's contrast and detailed description, and improve the skills of expressing emotions and themes in writing.

[0070] For question 2 above, the text learning results can be: 1. Emotional understanding and resonance: students can appreciate Yu Dafu's emotional changes and learn to interpret the emotional depth in the text from different perspectives. 2. Language appreciation: through delicate description, students improve their sensitivity and appreciation of literary language and understand the beauty of artistic conception. 3. Writing skills: imitate the contrast and emotional transition techniques in the text to improve the ability to express their own emotions and opinions.

[0071] For question 3 above, the text learning results can be: 1. Improve text appreciation: students can identify and understand the emotional color of quantifiers and comprehend their role in creating artistic conception. 2. Emotional expression and interpretation: learn to capture the author's emotions from quantifiers, reduplication and details, enhance emotional resonance and text interpretation ability. 3. Use of rhetorical devices: understand the role of rhetorical devices such as contrast and metaphor in presenting themes and conveying emotions, and improve writing skills.

[0072] For question 4 above, the text learning results can be: 1. Perceive aesthetic integration: understand how the author integrates subjective emotions into natural scenery and cultivate aesthetic appreciation. 2. Experience emotional expression: learn to interpret the author's emotional world through literary works and improve emotional resonance. 3. Construct literary space: learn how to construct personalized literary artistic conception with the help of text and enhance the depth of text expression.

[0073] For question 5 above, the text learning results can be: 1. Sensory interpretation and aesthetic appreciation: students will learn to understand the deep meaning of the text through detailed description, and improve their aesthetic perception and text sensitivity. 2. Cultural and emotional resonance: students can appreciate the cultural characteristics of Beijing and Yu Dafu's emotional sustenance, and enhance their understanding of the humanistic connotation of the work. 3. Writing skills and emotional expression: by imitating the writing method of observation and emotional integration in the text, improve the ability of scene integration and emotional transmission when writing.

[0074] For another example, the prompt for the key point analysis can be: Now you are a Chinese language expert, and you are teaching Yu Dafu's "Autumn in the Old Capital". Please obtain the teaching key point analysis corresponding to the question based on the analysis and explanation and the analysis of the textbook "Autumn in the Old Capital". The question is as follows: {wenti}, the analysis and explanation are as follows: {fenxichanshi}, and the example is: {few_shot_teaching key point analysis}. Pay attention to the correlation between the question, analysis and explanation, and the teaching key point analysis. The teaching key point analysis is written for teachers. Do not output questions and analysis and explanation. The output content should be a continuous and fluent paragraph, and do not summarize it in points.

[0075] Accordingly, for question 1 above, the analysis results of the key points of teaching can be: when teaching Yu Dafu's "Autumn in the Old Capital", students should be guided to pay attention to how the author uses contrast to show emotional changes. First, by interpreting the word "good" at the beginning and the desire for autumn at the end, students can understand the author's transformation from sadness to appreciation of autumn, and feel the level of "love because of sadness". Secondly, guide students to analyze five autumn scenes, appreciate the author's delicate description of autumn in the north, and experience the process of "love because of love". At the same time, it is necessary to discuss the contrast between autumn in the south and the north, highlight the distinctive characteristics of autumn in the north, and the author's deep thinking about the living conditions and natural environment contained therein. In addition, by discussing the dialogue between urban idlers, analyze the author's reflection on the city and the countryside, the hustle and bustle and the tranquility, and strengthen the love for autumn in the north. On the whole, teaching should focus on analyzing the richness of the author's emotions and the depth of the connotation of the work, and help students understand Yu Dafu's yearning for a fresh and simple life and his insight into the philosophy of life.

[0076] For question 2 above, the key points of teaching analysis can be obtained as follows: Yu Dafu's pursuit of autumn in the north is actually a projection and search of his inner emotions. This behavior reflects his need for a deep experience of life and his sensitivity to nature and humanistic feelings. In teaching, students should be guided to understand the change in Yu Dafu's emotional attitude: starting from the sad feeling of autumn, digging out his thinking about the philosophy of life and the sustenance of sincere emotions behind it. By analyzing his description of the autumn scenery in the north, emphasizing his yearning for a simple life and his true love for nature, this is a complex emotion of "love because of sadness, and joy because of love". At the same time, let students experience the author's joy and pursuit of an ideal life in the sad autumn scenery through comparative analysis, and cultivate their ability to interpret the works at multiple levels. In this way, students can not only deeply understand Yu Dafu's emotional world, but also enhance their appreciation of the emotional expression of literary works.

[0077] For question 3 above, the key points of teaching analysis can be: When teaching Yu Dafu's "Autumn in the Old Capital", attention should be paid to how the author projects emotions through quantifiers. First, guide students to analyze the choice of words such as "a rafter" and "a bowl" to appreciate the author's appreciation of simple life and the unique emotions of life in the old capital. Secondly, explain the use of reduplicated words, such as "every household" and "raised at home", so that students can understand their role in creating atmosphere and emotional rendering. Through comparison and metaphor, analyze the author's treatment of the relationship between autumn cicadas and house insects, and reveal his different orientations to real life and traditional poetic expression. Emphasize the subjective feelings in the details of the text, such as "delicate", "leisurely" and "lonely", so that students can understand the integration of the author's emotions and natural scenes. At the same time, discuss the author's selection and detailed treatment of images, such as the description of the autumn morning in the broken courtyard, to show his pursuit of secular beauty and ideal life. In terms of structure, explain the characteristics of the beginning and the end echoing and the combination of narration and discussion to help students understand the author's thinking on the philosophy of life. Finally, combined with Yu Dafu’s life background, students can be helped to interpret the life sentimentality and feelings for the old capital in the text more deeply, and enhance their comprehensive understanding of the connotation of the work.

[0078] For question 4 above, the key points of teaching analysis can be: When teaching Yu Dafu's "Autumn in the Old Capital", students should be guided to deeply understand the relationship between the author's subjective feelings and natural scenery through "clear" and "quiet". First, explain the natural characteristics represented by these two words in the text, so that students can appreciate the purity and tranquility of autumn in the north. Then, analyze the author's emotional projection, so that students can understand how Yu Dafu integrates his sad mood into the autumn scenery, forming a resonant aesthetic experience. In the classroom discussion, emphasize the creative concept of "scenery created by the heart", so that students understand that the text is not only a description of the scenery, but also an interweaving of emotions and memories. Through careful reading of the text, students should learn to capture the author's personal emotional clues and understand that "Yu Dafu's Autumn" is not a simple visual experience, but contains rich life philosophy and deep feelings for the hometown. In this way, students can not only appreciate the beauty of literature, but also understand the deep connection between imagery and emotion in literary works, and how the author uses words to construct a unique spiritual space.

[0079] For question 5 above, the key points of teaching analysis can be: When teaching "Autumn in the Old Capital", students should be guided to deeply understand Yu Dafu's multi-level description of "autumn flavor". First, emphasize that this is the product of the author's personal emotions and Beijing culture. Through the description of sensory details, let students experience Beijingers' artistic appreciation of life. Discussions can be organized in class to analyze the autumn meaning in ordinary scenes in the text, such as the tranquility of the broken house in the morning and the touch of the falling pistils of the locust tree, so as to enhance students' aesthetic perception. Secondly, explain how Yu Dafu shows the harmony between people and the environment and emotional communication through character dialogues, so that students understand that "autumn flavor" is not only about the scenery, but also about the emotional connotation behind the words. Finally, interpret the image of "people in the old capital" at the end of the article, reveal that "autumn flavor" is both an objective scene and an interweaving of subjective emotions, and cultivate their reading ability of blending emotions and reason. Through such teaching, it not only helps students understand the text, but also allows them to feel the restrained and implicit beauty of oriental culture.

[0080] For another example, the text analysis prompt can be: You are a language expert, teaching Yu Dafu's "Autumn in the Old Capital". Now you need to analyze the problem based on the appreciation passage and generate an analysis and interpretation. Pay attention to combining the appreciation passage and consider the relevance between the problem and the analysis and interpretation. Question: {wenti}, appreciation passage: {shangxipianduan}.

[0081] Correspondingly, for question 1 above, the result of the teaching text analysis can be: Yu Dafu expressed his complex and rich emotional changes towards the autumn in the north through contrast in "Autumn in the Old Capital". First, he started with "sorrow", borrowed the ancients' common sadness about autumn, combined with his personal life experience, and formed resonance. However, he did not indulge in sadness, but repeatedly emphasized his love for the autumn in the north, such as the word "good" at the beginning and the end of the exchange of life for autumn, showing the emotional transformation of "love because of sadness". In the article, he showed his deep love and heartfelt joy for autumn through the detailed description of the autumn scenery in the north, especially through the portrayal of five autumn scenes, which embodies the process of "joy because of love". Secondly, Yu Dafu skillfully used contrast to compare the autumn in the north with the autumn in the south of the Yangtze River, highlighting the characteristics of the autumn in the north, such as clear, quiet, and sad, which triggered his emotional resonance, and also let readers feel his yearning for a simple life and his love for real nature. Through the description of the conversations between urban idlers, he expressed his thoughts on the city and the countryside, the hustle and bustle and the quiet life in a deeper level, and further strengthened his love for the autumn in the north. In general, "Autumn in the Old Capital" is not a single sad story, but contains the author's multi-level emotional experience, with sadness, love, and joy. This emotional change makes the work more rich in connotation. Through contrast, detailed portrayal and deep perception, Yu Dafu showed his unique understanding and affectionate praise of the autumn in the north, thus conveying his yearning for a fresh and simple life and his understanding of the philosophy of life.

[0082] For question 2 above, the result of the teaching text analysis can be: Yu Dafu traveled thousands of miles to pursue the autumn in the north because he deeply resonated with the sad characteristics of autumn. This sadness was not a simple negative emotion, but triggered his thinking and emotional resonance, thus forming the emotional basis of love. In the article, he clearly expressed his love for the autumn in the north, such as the praise of autumn in the first paragraph and the reason for the long journey, as well as the strong desire to express his feelings at the end, showing "love because of sadness". At the same time, his delicate description and love of the autumn scenery in the north further developed into "joy because of love". Through the depiction of five autumn scenes, he conveyed his yearning for a light and simple life and his joy for real nature. The article also highlights the unique charm of autumn in the north through contrast, allowing readers to feel the author's joy and love for life in addition to sadness. Therefore, Yu Dafu's emotional attitude is not a single sadness, but contains deep love and joy, as well as the pursuit of an ideal life, showing the complexity and layering of emotions.

[0083] For question 3 above, the result of the teaching text analysis can be: Yu Dafu skillfully uses quantifiers to show his emotional projection in "Autumn in the Old Capital". For example, "a rafter", "a bowl", "a seat", "a thread by a thread", these words seem simple, but they actually contain deep meaning. Choosing "a rafter broken house" instead of "a room" expresses the author's satisfaction with life in the old capital, his love for simple life, and his deep attachment to the old capital. The selection and matching of quantifiers, such as "a seat" and "a thread by a thread", create a leisurely and quiet artistic conception, while also implying a lonely mood, allowing readers to deeply experience the author's emotional world. The use of reduplication, such as "every household" and "raised at home", not only increases the rhythmic beauty of the language, but also strengthens the relationship between autumn cicadas and residents' lives, implying the sad emotions brought by autumn cicadas. Through contrast and metaphor, such as the connection between autumn cicadas and house insects, the author conveys the realistic feelings of autumn, rather than traditional poetic expressions. In the careful reading of the text, the author's subjective feelings are closely combined with objective images, such as the description of the falling petals of the autumn locust tree, which is not only a symbol of life, but also a carrier of the author's emotions. Through the careful depiction of details and the touch of emotions, such as "feeling delicate", "feeling leisurely", and "feeling lonely", the author's emotions blend with the natural scene to form a strong emotional tension. In addition, the author considered the selection and details of images when creating, such as focusing on the description of the autumn morning in the broken courtyard, reflecting his pursuit of the beauty of secular life, which is closely related to his personal life experience and aesthetic ideals. At the same time, the echo of the beginning and the end of the text structure and the combination of the narrative and discussion in the middle part reflect the author's deep thinking about autumn and people.

[0084] For question 4 above, the result of the teaching text analysis is as follows: The relationship between Yu Dafu's subjective feelings and natural scenery through "clear" and "quiet" in "Autumn in the Old Capital" is actually the author's projection of his emotions onto the objective scenery, forming a unique aesthetic experience. The two words are used separately. On the one hand, they emphasize the pure natural characteristics of autumn in the north, such as the high sky, the flight of trained pigeons, the leaking sunlight and morning glory. These "clear" and "quiet" depict a quiet and profound autumn. On the other hand, they also carry Yu Dafu's inner sadness. This fusion of subjective feelings and the environment allows readers to experience the author's unique understanding and deep nostalgia for the autumn in the old capital when reading. Yu Dafu's writing technique cleverly uses the concept of "scenery created by the heart". He not only depicts the autumn scenery in the text, but also injects his own emotional world and personal experience into it. His "very autumn feeling" comes from a deep reflection on life and a deep attachment to his hometown, which makes the autumn in the old capital not only a visual scene, but also an interweaving of emotions and memories. Through delicate descriptions and strong subjective emotions, Yu Dafu's "Autumn" has become a carrier of his personal emotions, that is, "Yu Dafu's Autumn". Therefore, when we talk about "Autumn in the Old Capital", we should not only see the quiet and sad natural scenery in the text, but also understand Yu Dafu's personal subjective world and his deep affection for the old capital. This relationship reflects the fusion of imagery and emotion in literary works, and how the author constructs a unique literary space through words, allowing readers to not only feel the external scenery, but also deeply understand the author's inner world.

[0085] For question 5 above, the result of the teaching text analysis can be: Yu Dafu's multi-level description of the "autumn flavor" in "Autumn in the Old Capital" is not only his personal unique feeling, but also deeply integrated into the cultural connotation of Beijing. He repeatedly emphasized the "autumn flavor", which is actually to show the artistic appreciation attitude of Beijingers towards life through delicate description of the senses, which reflects the taste and aesthetic taste of Beijing culture. As a cultural person who once lived in the old capital, Yu Dafu guided readers to experience the charm of autumn in ordinary life scenes (such as the morning of a broken house and the details of autumn) from the perspective of a "Beijinger". This aesthetic experience that goes deep into life makes the "autumn flavor" have both the leisurely and leisurely beauty and the deep feeling of loneliness, forming a unique "Beijing culture" charm. Yu Dafu is not limited to the surface of natural landscapes, but by guiding readers to pay attention to details, such as the touch of the fallen pistils of the locust tree, the meaning of the residual sound of the autumn cicada, and the leisurely posture after the rain, etc., to show the delicate levels of the "autumn flavor of the old capital". His description of the art of speaking of Beijingers further reveals the harmonious coexistence of people and the environment, as well as the emotional exchange contained in speech, and strengthens the rich connotation of the flavor of autumn. At the end of the article, Yu Dafu integrates the "flavor" of nature, emotion and language through the image of "people in the old capital", showing that the "flavor of autumn" is both an objective landscape and a manifestation of subjective emotions, forming an aesthetic effect of blending emotion and reason. This is not only a deep interpretation of the autumn colors of the old capital, but also a reflection of the restrained and implicit characteristics of oriental culture. In short, the multi-level description of the "flavor of autumn" in "Autumn in the Old Capital" is not only a sustenance of Yu Dafu's personal emotions, but also a carrier of the cultural spirit of Beijing. Through meticulous observation and emotional expression, readers can feel the charm of culture and the art of life in their taste.

[0086] For another example, the answer prompt can be configured according to the actual answer required, such as: You are now a language expert, teaching Yu Dafu's "Autumn in the Old Capital", and now you need to generate a reference answer corresponding to the question based on the question and analysis and explanation. Question: {wenti}, analysis and explanation: {fenxichanshi}.

[0087] Accordingly, the answer to question 1 above can be: Yu Dafu combines his own tragic experience with his love for autumn in the north through contrast in "Autumn in the Old Capital". He uses the ancients' sadness for autumn as an introduction to express the resonance in his own life, but by repeatedly emphasizing his love for autumn, such as the word "good" and his persistent pursuit of autumn, he shows the process of turning from sadness to love. The article depicts five autumn scenes to show his deep affection and joy for autumn in the north, reflecting the deepening of the emotion of "joy because of love". At the same time, through the contrast between Jiangnan and the north, the clear, quiet and sad autumn in the north is highlighted, expressing his true love for simple life and nature. The dialogue between urban idlers reveals his reflection on urban and rural life and strengthens his identification with autumn in the north. Overall, "Autumn in the Old Capital" contains rich emotional levels, including sadness, love and joy, showing the author's unique perception of fresh life and life philosophy.

[0088] For question 2 above, the answer is: Yu Dafu's pursuit of autumn in the north reflects his deep feelings for autumn, including his resonance with the sad and bleak nature of autumn and his love for this sadness, forming an emotional attitude of "love because of sadness". His description and yearning for the autumn scenery in the north reflects the joy of "joy because of love" and his desire for a simple and unsophisticated life. At the same time, his emotions also include the pursuit of an ideal life and his love for real nature, showing the richness and depth of his emotions.

[0089] For question 3 above, the answer can be: Yu Dafu accurately depicts the daily life scenes of the ancient capital through quantitative words such as "a rafter" and "a bowl" in "Autumn in the Ancient Capital". These words not only show the simplicity and leisure of life, but also contain the author's recognition of the simple life and his attachment to the ancient capital. For example, the "a rafter" in "a broken house" emphasizes the simplicity of life, but reveals the author's unique aesthetics. The use of reduplicated words, such as "every household" and "raised at home", strengthens the rendering of emotions, such as the connection between the wailing of autumn cicadas and the lives of residents, conveying a faint sadness. The author uses contrast and metaphor, such as the comparison between autumn cicadas and house insects, to show his personal perception of the real autumn scenery, rather than traditional beautification. In the text, the choice of quantitative words is intertwined with emotional vocabulary (such as "delicate", "leisurely" and "lonely"), forming a fusion of emotions and natural scenes, showing a strong emotional tension. At the same time, the author deliberately highlights certain images, such as the autumn morning in the broken courtyard, reflecting his attention to the beauty of ordinary life and the inner world. On the whole, the use of these quantifiers reflects the author's deep feelings about life and his deep affection for his hometown, and needs to be fully interpreted in conjunction with his personal experience and the background of the times.

[0090] For question 4 above, the answer is: In "Autumn in the Old Capital", Yu Dafu closely connects subjective emotions with natural scenery through the two words "clear" and "quiet". He shows the purity and tranquility of autumn in the north by describing the high autumn sky, the flying pigeons, the leaking sunlight and the quiet beauty of morning glory, which is his intuitive reflection of natural scenery. However, "clear" and "quiet" also contain Yu Dafu's inner world, especially his sadness and nostalgia for his hometown, which makes these scenes have a strong emotional color. This writing style of "scenery created by the heart" makes the autumn in the old capital a projection of Yu Dafu's personal emotions, and becomes his unique life experience and literary creation. Therefore, while appreciating these scenes, readers can also feel the author's deep emotional resonance and special emotional sustenance for the autumn in the old capital. This fusion of imagery and emotion reflects the depth of emotional expression and artistic charm in literary works.

[0091] For question 5 above, the answer is: Yu Dafu's multi-level description of the "autumn flavor" in "Autumn in the Old Capital" shows the artistic perception of Beijingers on life through the description of sensory details, reflecting their taste and aesthetic taste. He closely combines personal emotions with the culture of the old capital, and conveys the characteristics of "Beijing culture" intertwined with leisure and loneliness through ordinary life scenes such as broken houses in the early morning and autumn scenes. From the touch of the pistil of the locust tree, the residual sound of the autumn cicada to the art of speech, it reveals the harmony and emotional exchange between people and the environment, forming an aesthetic effect of blending emotion and reason. This is not only a poetic expression of the autumn colors of the old capital, but also a vivid display of the introverted characteristics of oriental culture, and a carrier of Yu Dafu's emotional sustenance and Beijing spirit.

[0092] In an optional embodiment, before the questions corresponding to each text paragraph, the analysis and interpretation content corresponding to the questions, and the answers to the corresponding questions determined based on the analysis and interpretation content are stored in a database as a corresponding question-answer (QA) chain, the following steps are included: the questions corresponding to each text paragraph are input into a question detection model to obtain a question check result output by the question detection model, the question check result being used to characterize whether the object of the corresponding question inquiry is the target object; wherein the question detection model is trained based on historical questions asked about the target object, historical questions asked about other objects different from the target object, and historical question labels corresponding to the historical questions; if the object of the corresponding question inquiry is the target object based on the question check result, it is retained; otherwise, the multiple text paragraphs corresponding to the history teaching content are re-input into the knowledge prediction model for prediction.

[0093] It should be added that when inputting the questions corresponding to each text paragraph into the question detection model, if the question checking model selects the general large model, correspondingly, the question checking prompt is also set and input into the question checking model. The question checking prompt can be configured according to the actual inspection needs, for example, it can be: Now you are a language expert, teaching Yu Dafu's "Autumn in the Old Capital", now you need to determine whether the question is for students, if so, please directly output the original question, if not, please refer to the appreciation paragraph to regenerate. Appreciation paragraph: {shangxipianduan}, question: {wenti}.

[0094] In an optional embodiment, before inputting multiple text paragraphs corresponding to the historical teaching content into the knowledge prediction model, it includes: obtaining teaching content training data and question labels, explanation content labels and answer labels corresponding to the teaching content training data; using the teaching content training data as input data for training, using the question labels, explanation content labels and answer labels corresponding to the teaching content training data as labels for training, training the knowledge prediction model to be trained until the model converges or reaches a preset number of iterations, and obtaining a trained knowledge prediction model.

[0095] In an optional embodiment, the knowledge base includes at least one configuration file related to the historical teaching content and stored in a preset format. The configuration file is obtained based on the question field, the analysis and interpretation field, and the answer field, which respectively store the questions, analysis and interpretation content, and answers corresponding to the corresponding teaching content. Correspondingly, the questions corresponding to each text paragraph, the analysis and interpretation content corresponding to the questions, and the answers to the corresponding questions determined based on the analysis and interpretation content are stored in the database as a corresponding question-answer (QA) chain, including: the questions corresponding to each text paragraph, the analysis and interpretation content corresponding to the questions, and the answers to the corresponding questions determined based on the analysis and interpretation content are stored as a corresponding question-answer (QA) chain, and stored in the knowledge base in a preset format to obtain a configuration file for the corresponding text paragraph. It should be supplemented that the preset format can be in json format, and accordingly, each json file includes three field elements: question, analysis and interpretation, and reference answer.

[0096] Step S14, according to the question prediction chain, search the knowledge base to obtain the target question-answer chain that matches the question prediction chain. The knowledge base is constructed based on the questions determined based on the history teaching content, the analysis and interpretation content corresponding to the questions, and the answers to the corresponding questions determined based on the analysis and interpretation content.

[0097] In this embodiment, first, before searching the knowledge base according to the question keywords and combining the target index to obtain the target question-answer chain, it includes: creating a target index corresponding to the configuration file in the knowledge base based on the teaching target information.

[0098] Furthermore, based on the teaching objective information, a target index corresponding to the configuration file in the knowledge base is created, including: configuring the index name based on the name of the target teaching content based on the teaching objective information; searching the knowledge base based on the name of the target teaching content, determining all configuration files of the historical teaching content corresponding to the target teaching content, and obtaining index data; determining the mapping field and the text configuration type corresponding to the mapping field according to the index data; creating the target index according to the preset number of shards, the preset number of copies, the mapping field and the text configuration type corresponding to the mapping field, combined with the previously configured index creation statement, and saving it according to the index name; wherein, the preset number of shards is used to limit the index configuration file to be processed as a whole as several shards, and the preset number of copies is used to limit the number of backups of the index configuration file.

[0099] Furthermore, the index creation statement includes a setting statement and a mapping statement. Before creating the target index according to the preset number of shards, the preset number of copies, the mapping fields and the text configuration type corresponding to the mapping fields, combined with the previously configured index creation statement, it includes: configuring the setting statement based on the preset number of shards and the preset number of copies, and configuring the mapping statement based on the mapping fields and the text configuration type corresponding to the mapping fields to obtain the index creation statement; executing the index creation statement to obtain the target index.

[0100] For example, an index creation statement can be expressed as:

[0101] index_name = "Autumn in the Ancient Capital"

[0102] index_config = {

[0103] "settings": {

[0104] "number_of_shards": 1,

[0105] "number_of_replicas": 0

[0106] },

[0107] "mappings": {

[0108] "properties": {

[0109] "wenti": {"type": "text"},

[0110] "fenxichanshi": {"type": "text"},

[0111] "cankaodaan": {"type": "text"}

[0112] }

[0113] }

[0114] }

[0115] Secondly, according to the question prediction chain, the knowledge base is searched to obtain a target question and answer chain that matches the question prediction chain, including: extracting keywords for each predicted question in the question prediction chain to obtain corresponding question keywords; according to the question keywords, combined with the target index, the knowledge base is searched to obtain the target question and answer chain; wherein, the target index is first obtained by creating a mapping relationship based on the configuration file corresponding to the target teaching content corresponding to the teaching objective information and the historical teaching content.

[0116] Specifically, according to the question keywords, combined with the target index, the knowledge base is searched to obtain the target question and answer chain, including: according to the question keywords, combined with the target index, the question field of the corresponding configuration file in the knowledge base is searched, and the question keywords are matched with the questions stored in the corresponding question field obtained by the search to obtain the target question; according to the target question, the target analysis and explanation content corresponding to the target question in the analysis and explanation field of the corresponding configuration file is obtained; according to the target question, the target answer corresponding to the target question in the answer field of the corresponding configuration file is obtained; according to the target question, the target analysis and explanation content and the target answer, the target question and answer chain is obtained.

[0117] Furthermore, according to the question keywords, in combination with the target index, the question field of the corresponding configuration file in the knowledge base is searched, and the question keywords are matched with the questions stored in the corresponding question fields obtained by the search to obtain the target question, including: determining the source of the query data according to the index data of the target index, and determining the query field according to the question keywords; determining the query conditions according to the query data source and the query field; searching according to the query conditions to obtain query results that meet the query conditions, the query results including at least one record, and each record including the target question field that meets the query conditions, the analysis and explanation content and the answer corresponding to the question field; traversing each record in the query result, performing similarity matching between the corresponding question field and the question keywords to obtain the target question.

[0118] It should be noted that the query condition statement can be:

[0119] index_name = "Autumn in the Ancient Capital"

[0120] query = {

[0121] 'query': {

[0122] 'multi_match': {

[0123] 'query': characters,

[0124] 'fields': 'wenti'

[0125] }

[0126] },

[0127] 'size': 1 # Return only one result

[0128] }

[0129] It should be noted that characters can be defined according to the index data to be matched and searched, and the query conditions can be obtained by configuring the query object query and the field field to search for the corresponding problem field in characters, thereby realizing the query; in addition, size can be configured according to the actual number of results to be returned, which is not further limited here.

[0130] The statement to get the query result can be:

[0131] res = es.search(index=index_name, body=query)

[0132] The statement for obtaining the target question and obtaining the corresponding target analysis and interpretation content and the target answer according to the target question can be:

[0133] for hit in res['hits']['hits']:

[0134] print(hit['_source'].get('wenti'))

[0135] print(hit['_source'].get('fenxichanshi'))

[0136] print(hit['_source'].get('cankaodaan'))

[0137] Finally, after obtaining the target question-answer chain, the target index is deleted to ensure that the questions in each teaching interaction are not repeated. It should be added that the statement for deleting the target index can be:

[0138] # Delete the searched documents

[0139] es.delete(index=index_name, id=hit['_id'])

[0140] Step S15, using the question-answer chain that matches the question prediction chain, replace the question prediction chain to obtain the corresponding teaching update content.

[0141] Step S16, according to the teaching update content, identifying the task association between each teaching prediction task in the teaching update content and the corresponding question-answer chain, and obtaining the teaching interaction content.

[0142] In this embodiment, according to the teaching update content, the task association between each teaching prediction task and the corresponding question and answer chain in the teaching update content is identified to obtain the teaching interaction content, including: inputting the teaching update content and the task association prompt into the interactive content prediction model to obtain the teaching interaction content output by the interactive content prediction model; wherein the interactive content prediction model is trained according to the teaching update content training data and the task association labels and content labels corresponding to the teaching update content training data, and the interactive content prediction model is used to predict the task association between the teaching prediction task and the corresponding question and answer chain based on the input teaching update content, and combine the task association analysis results and the target question and answer chain to predict the corresponding answer to obtain the teaching interaction content.

[0143] Furthermore, the interactive content prediction model includes a feature extraction layer, an association prediction layer, an answer prediction layer and a content generation layer, and the teaching update content and the task association prompt are input into the interactive content prediction model to obtain the teaching interactive content output by the interactive content prediction model, including: inputting the teaching update content, the task association prompt and the content generation prompt into the feature extraction layer to extract features of the teaching update content to obtain the task features of each teaching prediction task and the problem features of each teaching prediction task; inputting the task features of each teaching prediction task and the problem features of each teaching prediction task into the association prediction layer, and predicting the association between the teaching prediction task and the target question corresponding to the teaching prediction task according to the task association prompt to obtain the task association analysis result corresponding to each teaching prediction task; inputting each teaching prediction task, the target question and answer chain corresponding to each teaching prediction task and the task association analysis result corresponding to each teaching prediction task into the answer prediction layer to predict the answer and obtain the predicted answer; inputting each teaching prediction task, the target question corresponding to each teaching prediction task, the task association analysis result corresponding to each teaching prediction task and the corresponding predicted answer into the content generation layer, and generating the prompt according to the content to obtain the teaching interactive content.

[0144] It should be added that the teaching interaction content includes each teaching prediction task, the target question corresponding to each teaching prediction task, and the supplemented target analysis and interpretation content and predicted answers corresponding to each target question. In addition, the supplemented target analysis and interpretation content is obtained by supplementing the target analysis and interpretation content in the target question-answering (QA) chain based on the task association analysis results.

[0145] It should be noted that task-related analysis includes the understanding and analysis of the order of prose writing, which is of great significance for improving the aesthetic ability of prose. After a preliminary understanding of the scenery and characteristics of the ancient capital's autumn, we can deeply analyze the logical order of writing afterwards, that is, understand how Yu Dafu used his senses to write about the autumn scenery. For example, the task-related prompt can be: Now you are a language expert and are teaching Yu Dafu's "Autumn in the Ancient Capital". Please analyze the role of the questions and the analysis and interpretation of "Autumn in the Ancient Capital" in promoting the completion of the learning task, and get the task-related analysis. The questions and analysis and interpretation are as follows: {QA}, and the tasks are as follows: {xvexirenwu}.

[0146] In an optional embodiment, before inputting the teaching update content and the task-associated prompt into the interactive content prediction model, it includes: obtaining the teaching update content training data and the task-associated labels and content labels corresponding to the teaching update content training data; using the teaching update content training data as the input data for training, using the task-associated labels and content labels corresponding to the teaching update content training data as the labels for training, training the interactive content prediction model to be trained until the model converges or reaches a preset number of iterations, and obtaining a trained interactive content prediction model.

[0147] In summary, the embodiment of the present invention obtains teaching objective information so that each subsequent step of operation is derived and promoted based on this, ensuring that all teaching predictions, content updates, interactive designs and other links are closely aligned with the initially set teaching objectives, avoiding teaching activities from deviating from the expected direction; obtaining the teaching prediction background through the teaching objective information to preliminarily predict the teaching prediction content, and using the previously constructed knowledge base to quickly find the target question and answer chain related to the current prediction question chain to replace the question prediction chain of the teaching prediction content, thereby continuously optimizing and improving on the basis of previous teaching, ensuring that the teaching content always remains fresh and targeted, better meeting the needs that may arise in different learning stages, avoiding the obsolescence and rigidity of the teaching content, and then identifying the task association between each teaching prediction task and the corresponding question and answer chain according to the teaching update content, determining the teaching interaction content, while ensuring that the interaction is more personalized, ensuring the rationality of the teaching interaction content, and achieving improved teaching quality.

[0148] The teaching interactive content generating device provided by the present invention is described below. The teaching interactive content generating device described below and the teaching interactive content generating method described above can refer to each other.

[0149] Figure 3 A schematic diagram of the structure of a teaching interactive content generation device is shown, the device comprising:

[0150] A target acquisition module 31, which acquires teaching target information;

[0151] A background prediction module 32, which obtains a teaching prediction background according to the teaching target information;

[0152] A content prediction module 33 obtains at least one teaching prediction content according to the teaching prediction background, wherein the teaching prediction content includes a teaching prediction task and a problem prediction chain corresponding to the teaching prediction task, wherein the problem prediction chain is constructed based on at least one prediction problem obtained by predicting the teaching prediction task;

[0153] A search module 34 searches a knowledge base according to the problem prediction chain to obtain a target question-answer chain matching the problem prediction chain, wherein the knowledge base is constructed based on the problems determined based on the history teaching content, the analysis and interpretation content corresponding to the problems, and the answers corresponding to the problems determined based on the analysis and interpretation content;

[0154] A replacement module 35 replaces the question prediction chain with a question-answer chain matching the question prediction chain to obtain corresponding teaching update content;

[0155] The content generation module 36 identifies the task association between each teaching prediction task and the corresponding question-answer chain in the teaching update content according to the teaching update content, and obtains the teaching interaction content.

[0156] It should be added that the teaching goal information includes the name of the target teaching content and the content generation prompt for generating teaching interactive content. The name of the target teaching content is determined according to the target teaching content. For example, if the target teaching content is an article, the name of the corresponding target teaching content is the article name. In addition, the teaching goal information also includes background generation prompt, task content generation prompt and task association prompt.

[0157] In this embodiment, the background prediction module 32 is used to: input the name of the target teaching content and the background generation prompt into the background prediction model to obtain the teaching prediction background output by the background prediction model; wherein the background prediction model is trained based on the teaching content training data and the background labels corresponding to the teaching content training data.

[0158] In addition, the device also includes: a first data acquisition module, which acquires teaching content training data and background labels corresponding to the teaching content training data before inputting the name of the target teaching content and the background generation prompt into the background prediction model; a first training module, which uses the teaching content training data as input data for training, and uses the background labels corresponding to the teaching content training data as labels for training, and trains the background prediction model to be trained until the model converges or reaches a preset number of iteration steps, thereby obtaining a trained background prediction model.

[0159] The content prediction module 33 is used to: input the teaching prediction background and task content generation prompts into the teaching content prediction model to obtain at least one teaching prediction content output by the teaching content prediction model; wherein the teaching content prediction model is trained based on the background training data and the task labels and question labels corresponding to the background training data.

[0160] In addition, the device also includes: a second data acquisition module, which acquires background training data and task labels and question labels corresponding to the background training data before inputting the teaching prediction background and task content generation prompts into the teaching content prediction model; a second training module, which uses the background training data as input data for training, and uses the task labels and question labels corresponding to the background training data as labels for training, to train the teaching content prediction model to be trained until the model converges or reaches a preset number of iterations, thereby obtaining a trained teaching content prediction model.

[0161] In an optional embodiment, the device also includes: a historical content acquisition module, which acquires historical teaching content before searching the knowledge base according to the question prediction chain; a text segmentation module, which performs text segmentation on the historical teaching content to obtain multiple text paragraphs corresponding to the historical teaching content; a question and answer prediction module, which inputs the multiple text paragraphs corresponding to the historical teaching content into the knowledge prediction model to obtain the questions corresponding to each text paragraph output by the knowledge prediction model, the analysis and interpretation content corresponding to the questions, and the answers to the corresponding questions determined based on the analysis and interpretation content; wherein the knowledge prediction model is trained based on the teaching content training data and the question labels, interpretation content labels and answer labels corresponding to the teaching content training data; the knowledge prediction model is used to predict the questions, the analysis and interpretation content corresponding to the questions, and the answers to the questions for each text paragraph corresponding to the input historical teaching content; a storage module, which stores the questions corresponding to each text paragraph, the analysis and interpretation content corresponding to the questions, and the answers to the corresponding questions determined based on the analysis and interpretation content as a corresponding question and answer (QA) chain in the database.

[0162] Specifically, the text segmentation module includes: a text segmentation unit, which identifies historical teaching content and segments the historical teaching content according to a preset number of words to obtain multiple text paragraphs; a semantic completion unit, which calls the semantic completion model, combines the historical teaching content corresponding to the corresponding text paragraph, and performs semantic completion on each text paragraph, so that the content of each text paragraph has a more complete semantic structure.

[0163] In addition, the knowledge prediction model includes a feature extraction layer, a question prediction layer, an interpretation analysis layer and an answer prediction layer; the question and answer prediction module is used to: input multiple text paragraphs corresponding to the history teaching content into the feature extraction layer for feature extraction, and obtain the text features corresponding to each text paragraph; input the text features corresponding to each text paragraph into the question prediction layer for question prediction, and obtain the questions corresponding to each text paragraph; input the questions corresponding to each text paragraph into the interpretation analysis layer, so as to predict the corresponding text learning results and teaching points analysis results for each text paragraph, and predict the corresponding teaching text analysis results for each text paragraph and the questions of each text paragraph, and obtain the analysis and interpretation content of the corresponding text paragraph based on the predicted text learning results, teaching points analysis results and teaching text analysis results; input the analysis and interpretation content of each text paragraph and the questions corresponding to each text paragraph into the answer prediction layer for answer prediction, and obtain the answers corresponding to each text paragraph output by the answer prediction layer.

[0164] In an optional embodiment, the device also includes: a checking module, which inputs the questions corresponding to each text paragraph into a question detection model before storing the questions corresponding to each text paragraph, the analysis and interpretation content corresponding to the questions, and the answers to the corresponding questions determined based on the analysis and interpretation content as a corresponding question and answer (QA) chain in a database, and obtains the question checking results output by the question detection model, wherein the question checking results are used to characterize whether the corresponding question inquiry object is the target object; wherein the question detection model is trained based on historical questions asked about the target object, historical questions asked about other objects different from the target object, and historical question labels corresponding to the historical questions; and a result processing module, which retains the corresponding question inquiry object if the question checking result indicates that the corresponding question inquiry object is the target object; otherwise, re-inputs the multiple text paragraphs corresponding to the history teaching content into the knowledge prediction model for prediction.

[0165] In an optional embodiment, the device also includes: a third data acquisition module, which obtains teaching content training data and question labels, explanation content labels and answer labels corresponding to the teaching content training data before inputting multiple text paragraphs corresponding to the historical teaching content into the knowledge prediction model; a third training module, which uses the teaching content training data as input data for training, and uses the question labels, explanation content labels and answer labels corresponding to the teaching content training data as labels for training, and trains the knowledge prediction model to be trained until the model converges or reaches a preset number of iteration steps, thereby obtaining a trained knowledge prediction model.

[0166] In an optional embodiment, the knowledge base includes at least one configuration file related to the historical teaching content and stored in a preset format. The configuration file is obtained based on the question field, the analysis and interpretation field, and the answer field, which respectively store the questions, analysis and interpretation content, and answers corresponding to the teaching content. Correspondingly, the preservation module is used to: take the questions corresponding to each text paragraph, the analysis and interpretation content corresponding to the questions, and the answers to the corresponding questions determined based on the analysis and interpretation content as a corresponding question-answer (QA) chain, and store them in the knowledge base in a preset format to obtain a configuration file for the corresponding text paragraph. It should be supplemented that the preset format can be in json format, and accordingly, each json file includes three field elements: question, analysis and interpretation, and reference answer.

[0167] In an optional embodiment, the device further includes: an index creation module, which creates a target index corresponding to the configuration file in the knowledge base based on the teaching target information before searching the knowledge base according to the question keywords in combination with the target index to obtain the target question and answer chain.

[0168] Furthermore, the index creation module includes: a name configuration unit, which configures the index name based on the name of the target teaching content of the teaching objective information; a data acquisition unit, which searches the knowledge base for the name of the target teaching content, determines all configuration files of the historical teaching content corresponding to the target teaching content, and obtains index data; a text configuration unit, which determines the mapping field and the text configuration type corresponding to the mapping field according to the index data; an index creation unit, which creates the target index according to the preset number of shards, the preset number of copies, the mapping field and the text configuration type corresponding to the mapping field, combined with the previously configured index creation statement, and saves it according to the index name; wherein the preset number of shards is used to limit the index configuration file to be processed as a whole as several shards, and the preset number of copies is used to limit the number of backups of the index configuration file.

[0169] Furthermore, the index creation statement includes a setting statement and a mapping statement. The index creation module also includes a base statement configuration unit, which is used to configure the setting statement based on the preset number of shards, the preset number of copies, the mapping fields and the text configuration type corresponding to the mapping fields, and configure the mapping statement based on the mapping fields and the text configuration type corresponding to the mapping fields, before creating the target index in combination with the previously configured index creation statement, to obtain the index creation statement; execute the index creation statement to obtain the target index.

[0170] The search module 34 includes: a keyword extraction unit, which extracts keywords for each predicted question in the question prediction chain to obtain corresponding question keywords; a search unit, which searches the knowledge base according to the question keywords and the target index to obtain the target question and answer chain; wherein the target index is obtained by first creating a mapping relationship based on the target teaching content corresponding to the teaching target information and the configuration file corresponding to the historical teaching content.

[0171] Specifically, the search unit includes: a question search sub-unit, which searches for the question field of the corresponding configuration file in the knowledge base according to the question keywords and the target index, and matches the question keywords with the questions stored in the searched question field to obtain the target question; a content search sub-unit, which obtains the target analysis and interpretation content of the corresponding configuration file analysis and interpretation field according to the target question; a reply search sub-unit, which obtains the target answer of the corresponding configuration file answer field according to the target question; and a question and answer chain acquisition sub-unit, which obtains the target question and answer chain according to the target question, the target analysis and interpretation content and the target answer.

[0172] Furthermore, the question search sub-unit includes: a query condition determination grandchild unit, which determines the query data source based on the index data of the target index, and determines the query field based on the question keywords, and determines the query condition based on the query data source and the query field; a query search grandchild unit, which searches according to the query conditions to obtain query results that meet the query conditions, and the query results include at least one record, and each record includes the target question field that meets the query conditions, and the analysis and explanation content and answer corresponding to the question field; a matching grandchild unit, which traverses each record in the query result, and performs similarity matching between the corresponding question field and the question keyword to obtain the target question.

[0173] In an optional embodiment, the device further includes: a deletion module, which deletes the target index after obtaining the target question-answer chain to ensure that questions in each teaching interaction are not repeated.

[0174] The content generation module 36 is used to: input the teaching update content and the task association prompt into the interactive content prediction model to obtain the teaching interactive content output by the interactive content prediction model; wherein the interactive content prediction model is trained based on the teaching update content training data and the task association labels and content labels corresponding to the teaching update content training data, and the interactive content prediction model is used to predict the task association between the teaching prediction task and the corresponding question and answer chain based on the input teaching update content, and will combine the task association analysis results and the target question and answer chain to predict the corresponding answer to obtain the teaching interactive content.

[0175] Furthermore, the interactive content prediction model includes a feature extraction layer, an association prediction layer, an answer prediction layer and a content generation layer, and the content generation module 36 is used to: input the teaching update content, the task association prompt and the content generation prompt into the feature extraction layer to extract features of the teaching update content and obtain the task features of each teaching prediction task and the problem features of each teaching prediction task; input the task features of each teaching prediction task and the problem features of each teaching prediction task into the association prediction layer, and predict the association between the teaching prediction task and the target question corresponding to the teaching prediction task according to the task association prompt to obtain the task association analysis result corresponding to each teaching prediction task; input each teaching prediction task, the target question and answer chain corresponding to each teaching prediction task and the task association analysis result corresponding to each teaching prediction task into the answer prediction layer to predict the answer and obtain the predicted answer; input each teaching prediction task, the target question corresponding to each teaching prediction task, the task association analysis result corresponding to each teaching prediction task and the corresponding predicted answer into the content generation layer, and generate the prompt according to the content to obtain the teaching interactive content.

[0176] In an optional embodiment, the device also includes: a fourth data acquisition module, which obtains the teaching update content training data and the task association labels and content labels corresponding to the teaching update content training data before inputting the teaching update content and the task association prompt into the interactive content prediction model; a fourth training module, which uses the teaching update content training data as input data for training, and uses the task association labels and content labels corresponding to the lesson update content training data as labels for training, and trains the interactive content prediction model to be trained until the model converges or reaches a preset number of iteration steps, thereby obtaining a trained interactive content prediction model.

[0177] In summary, the embodiment of the present invention obtains teaching target information through the target acquisition module, so that each subsequent step of operation is derived and promoted based on this, ensuring that all teaching predictions, content updates, and interactive designs are closely aligned with the teaching targets set initially, and avoiding teaching activities from deviating from the expected direction; the teaching prediction background is obtained according to the teaching target information through the background prediction module, so as to preliminarily use the content prediction module to predict the teaching prediction content, and use the knowledge base previously constructed by the search module to quickly find the target question-answer chain related to the current prediction question chain, so as to replace the question prediction chain of the teaching prediction content through the replacement module, thereby continuously optimizing and improving on the basis of previous teaching, ensuring that the teaching content always remains fresh and targeted, better meeting the needs that may arise in different learning stages, avoiding the obsolescence and rigidity of the teaching content, and then identifying the task association between each teaching prediction task and the corresponding question-answer chain according to the teaching update content through the content generation module, determining the teaching interaction content, and ensuring that the interaction is more personalized while ensuring the rationality of the teaching interaction content, thereby improving the teaching quality.

[0178] Figure 4 An example of a physical structure diagram of an electronic device is shown in FIG. Figure 4 As shown, the electronic device may include: a processor (processor) 410 , a communication interface (Communications Interface) 420 , a memory (memory) 430 and a communication bus 440 , wherein the processor 410 , the communication interface 420 , and the memory 430 communicate with each other through the communication bus 440 . The processor 410 can call the logic instructions in the memory 430 to execute the teaching interactive content generation method, which includes: obtaining teaching target information; obtaining a teaching prediction background based on the teaching target information; obtaining at least one teaching prediction content based on the teaching prediction background, the teaching prediction content includes a teaching prediction task and a problem prediction chain corresponding to the teaching prediction task, and the problem prediction chain is constructed based on at least one predicted problem obtained by the corresponding prediction of the teaching prediction task; according to the problem prediction chain, searching the knowledge base to obtain a target question-answer chain matching the problem prediction chain, the knowledge base is constructed based on the problems determined based on the historical teaching content, the analysis and interpretation content corresponding to the problems, and the answers to the corresponding problems determined based on the analysis and interpretation content; using the question-answer chain matching the problem prediction chain, replacing the question prediction chain to obtain the corresponding teaching update content; according to the teaching update content, identifying the task association between each teaching prediction task and the corresponding question-answer chain in the teaching update content to obtain the teaching interactive content.

[0179] In addition, the logic instructions in the above-mentioned memory 430 can be implemented in the form of a software functional unit and can be stored in a computer-readable storage medium when it is sold or used as an independent product. Based on this understanding, the technical solution of the present invention, in essence, or the part that contributes to the prior art or the part of the technical solution, can be embodied in the form of a software product, and the computer software product is stored in a storage medium, including a number of instructions for a computer device (which can be a personal computer, a server, or a network device, etc.) to perform all or part of the steps of the method described in each embodiment of the present invention. The aforementioned storage medium includes: U disk, mobile hard disk, read-only memory (ROM, Read-Only Memory), random access memory (RAM, Random Access Memory), disk or optical disk, etc. Various media that can store program codes.

[0180] On the other hand, the present invention also provides a computer program product, which includes a computer program. The computer program can be stored on a non-transitory computer-readable storage medium. When the computer program is executed by a processor, the computer can execute the teaching interactive content generation method provided by the above methods, which method includes: obtaining teaching objective information; obtaining a teaching prediction background based on the teaching objective information; obtaining at least one teaching prediction content based on the teaching prediction background, the teaching prediction content including a teaching prediction task and a problem prediction chain corresponding to the teaching prediction task, the problem prediction chain is constructed based on at least one predicted problem obtained by the corresponding prediction of the teaching prediction task; according to the problem prediction chain, searching a knowledge base to obtain a target question-answer chain matching the problem prediction chain, the knowledge base is constructed based on the questions determined based on the historical teaching content, the analysis and interpretation content corresponding to the questions, and the answers to the corresponding questions determined based on the analysis and interpretation content; using the question-answer chain matching the question prediction chain, replacing the question prediction chain to obtain the corresponding teaching update content; according to the teaching update content, identifying the task association between each teaching prediction task and the corresponding question-answer chain in the teaching update content to obtain the teaching interactive content.

[0181] On the other hand, the present invention also provides a non-transitory computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, is implemented to execute the teaching interactive content generation method provided by the above-mentioned methods, the method comprising: obtaining teaching objective information; obtaining a teaching prediction background based on the teaching objective information; obtaining at least one teaching prediction content based on the teaching prediction background, the teaching prediction content comprising a teaching prediction task and a problem prediction chain corresponding to the teaching prediction task, the problem prediction chain being constructed based on at least one predicted problem obtained by the corresponding prediction of the teaching prediction task; searching a knowledge base based on the problem prediction chain to obtain a target question-answer chain matching the problem prediction chain, the knowledge base being constructed based on problems determined based on historical teaching content, analysis and interpretation content corresponding to the problems, and answers to corresponding problems determined based on the analysis and interpretation content; replacing the problem prediction chain with a question-answer chain matching the problem prediction chain to obtain corresponding teaching update content; identifying the task association between each teaching prediction task and the corresponding question-answer chain in the teaching update content based on the teaching update content to obtain teaching interactive content.

[0182] The device embodiments described above are merely illustrative, wherein the units described as separate components may or may not be physically separated, and the components displayed as units may or may not be physical units, that is, they may be located in one place, or they may be distributed on multiple network units. Some or all of the modules may be selected according to actual needs to achieve the purpose of the scheme of this embodiment. Ordinary technicians in this field can understand and implement it without paying creative labor.

[0183] Through the description of the above implementation methods, those skilled in the art can clearly understand that each implementation method can be implemented by means of software plus a necessary general hardware platform, and of course, can also be implemented by hardware. Based on this understanding, the above technical solution is essentially or the part that contributes to the prior art can be embodied in the form of a software product, and the computer software product can be stored in a computer-readable storage medium, such as ROM / RAM, a disk, an optical disk, etc., including a number of instructions for a computer device (which can be a personal computer, a server, or a network device, etc.) to execute the methods described in each embodiment or some parts of the embodiments.

[0184] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit it. Although the present invention has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that they can still modify the technical solutions described in the aforementioned embodiments, or make equivalent replacements for some of the technical features therein. However, these modifications or replacements do not deviate the essence of the corresponding technical solutions from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. A method for generating interactive teaching content, characterized in that: include: Obtain teaching objective information; According to the teaching target information, a teaching prediction background is obtained; According to the teaching prediction background, at least one teaching prediction content is obtained, wherein the teaching prediction content includes a teaching prediction task and a problem prediction chain corresponding to the teaching prediction task, wherein the problem prediction chain is constructed based on at least one prediction problem obtained by the prediction corresponding to the teaching prediction task; According to the problem prediction chain, a knowledge base is searched to obtain a target question-answer chain matching the problem prediction chain, wherein the knowledge base is constructed based on the problems determined based on the history teaching content, the analysis and interpretation content corresponding to the problems, and the answers corresponding to the problems determined based on the analysis and interpretation content; Using the question-answer chain that matches the question prediction chain, the question prediction chain is replaced to obtain corresponding teaching update content; According to the teaching update content, identifying the task association between each teaching prediction task in the teaching update content and the corresponding question-answer chain to obtain teaching interaction content; The knowledge base includes at least one configuration file corresponding to the history teaching content and stored in a preset format, wherein the configuration file is obtained based on the question field, the analysis and interpretation field, and the answer field, which respectively store the question corresponding to the teaching content, the analysis and interpretation content, and the answer; According to the question prediction chain, a knowledge base is searched to obtain a target question-answer chain matching the question prediction chain, including: Extract keywords for each predicted question in the question prediction chain to obtain corresponding question keywords; According to the question keywords, combined with the target index, the knowledge base is searched to obtain the target question-answer chain; wherein the target index is obtained by creating a mapping relationship based on the target teaching content corresponding to the teaching target information and the configuration file corresponding to the history teaching content; According to the question keywords, combined with the target index, the knowledge base is searched to obtain the target question-answer chain, including: According to the question keyword, combined with the target index, the question field of the corresponding configuration file in the knowledge base is searched, and the question keyword is matched with the question stored in the searched question field to obtain the target question; According to the target question, obtaining the target analysis and interpretation content corresponding to the target question in the analysis and interpretation field of the corresponding configuration file; According to the target question, obtaining a target answer corresponding to the target question in the answer field of the corresponding configuration file; Obtaining a target question-answer chain according to the target question, the target analysis and interpretation content, and the target answer; Before searching the knowledge base according to the problem prediction chain, it includes: Obtain historical teaching content; Performing text segmentation on the history teaching content to obtain a plurality of text paragraphs corresponding to the history teaching content; Input multiple text paragraphs corresponding to the history teaching content into the knowledge prediction model, and obtain the questions corresponding to each of the text paragraphs, the analysis and interpretation content corresponding to the questions, and the answers corresponding to the questions determined based on the analysis and interpretation content output by the knowledge prediction model; wherein the knowledge prediction model is trained based on the teaching content training data and the question labels, interpretation content labels, and answer labels corresponding to the teaching content training data; the knowledge prediction model is used to predict the questions, the analysis and interpretation content corresponding to the questions, and the answers corresponding to the questions for each text paragraph corresponding to the input history teaching content; The question corresponding to each of the text paragraphs, the analysis and interpretation content corresponding to the question, and the answer to the question determined based on the analysis and interpretation content are stored in the database as a corresponding question-answer chain.

2. The teaching interactive content generation method according to claim 1, characterized in that: After obtaining the target question-answer chain, the following steps are included: Delete the target index.

3. The teaching interactive content generation method according to claim 1, characterized in that: The knowledge prediction model includes a feature extraction layer, a question prediction layer, an interpretation analysis layer and an answer prediction layer; a plurality of text paragraphs corresponding to the history teaching content are input into the knowledge prediction model, and the questions corresponding to the text paragraphs, the analysis and interpretation content corresponding to the questions and the answers corresponding to the questions determined based on the analysis and interpretation content output by the knowledge prediction model are obtained, including: Inputting a plurality of text paragraphs corresponding to the history teaching content into the feature extraction layer for feature extraction to obtain text features corresponding to each of the text paragraphs; Inputting the text features corresponding to each of the text paragraphs into the question prediction layer to perform question prediction, and obtaining the questions corresponding to each of the text paragraphs; Inputting the questions corresponding to each of the text paragraphs into the interpretation analysis layer, so as to predict the corresponding text learning results and teaching key point analysis results for each text paragraph, and predict the corresponding teaching text analysis results for each text paragraph and the questions of each text paragraph, and obtaining the analysis and interpretation content of the corresponding text paragraph based on the predicted text learning results, teaching key point analysis results and teaching text analysis results; The analysis and interpretation content of each of the text paragraphs and the questions corresponding to each of the text paragraphs are input into the answer prediction layer for answer prediction, and the answers corresponding to each of the text paragraphs are output by the answer prediction layer.

4. The teaching interactive content generation method according to claim 3, characterized in that: Before storing the question corresponding to each text paragraph, the analysis and interpretation content corresponding to the question, and the answer corresponding to the question determined based on the analysis and interpretation content as a corresponding question-answer chain in a database, the method includes: Inputting the questions corresponding to each of the text paragraphs into a question detection model, obtaining a question check result output by the question detection model, wherein the question check result is used to characterize whether the object of the corresponding question inquiry is the target object; wherein the question detection model is trained based on historical questions inquired about the target object, historical questions inquired about other objects different from the target object, and historical question labels corresponding to the historical questions; If the question check result indicates that the object of the question inquiry is the target object, then retain it; Otherwise, the multiple text paragraphs corresponding to the history teaching content are re-input into the knowledge prediction model for prediction.

5. A teaching interactive content generation device, characterized in that: include: Target acquisition module, to obtain teaching target information; A background prediction module, which obtains a teaching prediction background according to the teaching target information; A content prediction module, which obtains at least one teaching prediction content according to the teaching prediction background, wherein the teaching prediction content includes a teaching prediction task and a problem prediction chain corresponding to the teaching prediction task, wherein the problem prediction chain is constructed based on at least one predicted problem corresponding to the teaching prediction task; A search module searches a knowledge base according to the question prediction chain to obtain a target question-answer chain matching the question prediction chain, wherein the knowledge base is constructed based on questions determined based on history teaching content, analysis and interpretation content corresponding to the questions, and answers corresponding to the questions determined based on the analysis and interpretation content; A replacement module, using a question-answer chain matching the question prediction chain to replace the question prediction chain to obtain corresponding teaching update content; A content generation module, based on the teaching update content, identifies the task association between each teaching prediction task and the corresponding question-answer chain in the teaching update content to obtain teaching interaction content; The knowledge base includes at least one configuration file corresponding to the history teaching content and stored in a preset format, wherein the configuration file is obtained based on the question field, the analysis and interpretation field, and the answer field, which respectively store the question corresponding to the teaching content, the analysis and interpretation content, and the answer; The search module comprises: A keyword extraction unit extracts keywords from each predicted question in the question prediction chain to obtain corresponding question keywords; A search unit searches the knowledge base according to the question keywords and the target index to obtain a target question-answer chain; wherein the target index is obtained by creating a mapping relationship based on the target teaching content corresponding to the teaching target information and the configuration file corresponding to the history teaching content; The searching unit comprises: The question search subunit searches for a question field corresponding to a configuration file in a knowledge base according to the question keyword and in combination with a target index, and matches the question keyword with the question stored in the searched question field to obtain a target question; The content search subunit obtains the target analysis and interpretation content corresponding to the target question in the analysis and interpretation field of the corresponding configuration file according to the target question; The answer search subunit obtains a target answer corresponding to the target question in the answer field of the corresponding configuration file according to the target question; A question-answer chain acquisition subunit obtains a target question-answer chain according to the target question, the target analysis and interpretation content, and the target answer; The device further comprises: A historical content acquisition module, which acquires historical teaching content before searching the knowledge base according to the problem prediction chain; A text segmentation module performs text segmentation on the history teaching content to obtain a plurality of text paragraphs corresponding to the history teaching content; The question-answer prediction module inputs a plurality of text paragraphs corresponding to the history teaching content into the knowledge prediction model, and obtains the questions corresponding to each of the text paragraphs, the analysis and interpretation content corresponding to the questions, and the answers corresponding to the questions determined based on the analysis and interpretation content output by the knowledge prediction model; wherein the knowledge prediction model is trained based on the teaching content training data and the question labels, interpretation content labels, and answer labels corresponding to the teaching content training data; the knowledge prediction model is used to predict the questions, the analysis and interpretation content corresponding to the questions, and the answers corresponding to the questions for each text paragraph corresponding to the input history teaching content; The storage module stores the question corresponding to each of the text paragraphs, the analysis and interpretation content corresponding to the question, and the answer corresponding to the question determined based on the analysis and interpretation content as a corresponding question-answer chain in a database.

6. An electronic device comprising a memory, a processor, and a computer program stored in the memory and running on the processor, characterized in that: When the processor executes the computer program, the teaching interactive content generating method as described in any one of claims 1 to 5 is implemented.

7. A computer program product, comprising a computer program, characterized in that When the computer program is executed by a processor, the teaching interactive content generating method as described in any one of claims 1 to 5 is implemented.

Images