International trade teaching case dynamic generation method and device

Abstract

The invention is suitable for the technical field of international trade teaching, and particularly relates to an international trade teaching case dynamic generation method and device, and the method comprises the steps: obtaining the static data, behavior data and characteristic data of a student; based on the static data, the behavior data and the characteristic data, constructing an ability portrait of the student; generating an international trade teaching case according to the ability portrait, the real-time trade data and the trade knowledge graph; and obtaining operation data of the student, and optimizing an international trade teaching case generation process based on the operation data. According to the method, the problem of disjunction of fixed cases and contents in a traditional teaching method can be solved, and dynamic, personalized and real-time optimized international trade teaching case generation is realized.

Description

Technical Field

[0001] The present application belongs to the field of international trade teaching technology, and in particular to a method and device for dynamically generating international trade teaching cases. Background Art

[0002] International trade teaching cases refer to real or simulated international trade situation analysis for teaching purposes, usually involving goods, services, capital flows between countries, as well as related policies, regulations and market behaviors. Cases may cover tariff barriers, free trade agreements, multinational corporate strategies, trade disputes, exchange rate impacts, etc., aiming to help students understand the operation, challenges and opportunities of global trade, and cultivate their ability to analyze and solve international trade problems.

[0003] In the existing technology, international trade teaching often uses fixed cases, which cannot be personalized according to the abilities, risk preferences and learning progress of different students, resulting in some students being over-challenged or under-challenged. Traditional cases are usually based on past trade rules and market environments, which are difficult to adapt to the frequent changes in international trade policies and cannot allow students to learn and respond to the latest market changes in a timely manner.

[0004] To sum up, in the process of international trade teaching, there is a problem of being unable to meet differentiated teaching needs due to the use of fixed teaching cases and the disconnection between the case content and the real-time trade environment. Summary of the invention

[0005] The embodiment of the present application provides a method and device for dynamically generating international trade teaching cases, which can solve the problem in the related art that in the process of international trade teaching, the use of fixed teaching cases and the disconnection of case content from the real-time trade environment lead to the inability to meet differentiated teaching needs.

[0006] In a first aspect, an embodiment of the present application provides a method for dynamically generating international trade teaching cases, including: Obtaining the static data, behavioral data and trait data of the students; wherein the static data includes the course grades and error rates of international trade courses, the behavioral data includes the deviation of the transaction price from the baseline and the transaction rate, and the trait data includes the stress resistance index and risk preference; Constructing the ability profile of the student based on the static data, the behavioral data and the trait data; Generate international trade teaching cases based on the capability profile, real-time trade data and trade knowledge graph; wherein the real-time trade data includes international trade policies, government announcements and media news, and the trade knowledge graph is constructed based on international trade rules; Obtain the operation data of the student, and optimize the generation process of international trade teaching cases based on the operation data; wherein, the operation data is the operation data generated by the student in processing the international trade teaching cases, and the operation data includes the current error rate, the current risk preference, and the current error mode.

[0007] In the embodiments of the present application, the above technical solutions have at least the following technical effects: The method for dynamically generating international trade teaching cases provided by the present application first obtains the static data (course grades and error rates of international trade courses), behavioral data (the amplitude of the transaction price deviating from the baseline and the transaction rate), and trait data (stress resistance index and risk preference) of students, then constructs an ability portrait of the students based on the static data, behavioral data, and trait data, and then generates international trade teaching cases according to the ability portrait, real-time trade data (international trade policies, government announcements, and media news), and trade knowledge graph (constructed based on international trade rules), and finally obtains the operation data of the students (the operation data generated by the students in processing international trade teaching cases, including the current error rate, the current risk preference, and the current error mode), and optimizes the generation process of international trade teaching cases based on the operation data. This method generates teaching cases by combining the ability portraits of students, real-time trade data, and trade knowledge graphs, avoiding the limitations of fixed cases, enabling the learning process of each student to match their actual situation, thereby improving the learning effect. The teaching cases can reflect the latest dynamics in the field of international trade in real time, facilitating the timeliness and relevance of teaching content, enabling students to learn the most cutting-edge knowledge and skills. This method can be dynamically adjusted according to the real-time operation feedback of students, improving the real-time adaptability of teaching cases. The continuous monitoring of operation data allows teachers and systems to adjust teaching strategies and case designs in a timely manner, improving the efficiency and accuracy of teaching. This method can overcome the problems of fixed cases and content disconnection in traditional teaching methods, realizing the generation of dynamic, personalized, and real-time optimized international trade teaching cases, greatly improving the accuracy of teaching and the learning effect, meeting the needs of differentiated teaching, and thus enhancing the learning experience and ability development of students.

[0008] In a second aspect, an embodiment of the present application provides a device for dynamically generating international trade teaching cases, including: An acquisition unit, configured to acquire the static data, behavioral data, and trait data of students; wherein, the static data includes the course grades and error rates of international trade courses, the behavioral data includes the amplitude of the transaction price deviating from the baseline and the transaction rate, and the trait data includes the stress resistance index and risk preference; An ability portrait construction unit, configured to construct the ability portrait of the student based on the static data, the behavioral data, and the trait data; An international trade teaching case generation unit, configured to generate international trade teaching cases according to the ability portrait, real-time trade data, and trade knowledge graph; wherein, the real-time trade data includes international trade policies, government announcements, and media news, and the trade knowledge graph is constructed based on international trade rules; An optimization unit, configured to obtain the operation data of the student, and optimize the international trade teaching case generation process based on the operation data; wherein, the operation data is the operation data generated by the student in processing the international trade teaching case, and the operation data includes the current error rate, the current risk preference, and the current error pattern.

[0009] In a third aspect, an embodiment of the present application provides an international trade teaching case dynamic generation device, including a memory, a processor, and a computer program stored in the memory and executable on the processor. When the processor executes the computer program, the method described in any one of the embodiments in the first aspect is implemented.

[0010] It can be understood that the beneficial effects of the above second aspect to the third aspect can refer to the relevant descriptions in the first aspect above, and will not be elaborated here. Description of the Drawings

[0011] In order to more clearly illustrate the technical solutions in the embodiments of the present application, the following will briefly introduce the drawings required for use in the embodiments or the description of the prior art. Obviously, the drawings in the following description are only some embodiments of the present application. For those of ordinary skill in the art, other drawings can be obtained based on these drawings without creative efforts.

[0012] Figure 1 It is a schematic flowchart of an international trade teaching case dynamic generation method provided by an embodiment of the present application; Figure 2 It is an example diagram of a Prompt in step S330 of the international trade teaching case dynamic generation method provided by an embodiment of the present application; Figure 3 It is an example diagram of an international trade teaching case in step S330 of the international trade teaching case dynamic generation method provided by an embodiment of the present application; Figure 4 It is an example diagram of a case structure template in step S301 of the international trade teaching case dynamic generation method provided by an embodiment of the present application; Figure 5 It is an example diagram of initial generation parameters in the international trade teaching case dynamic generation method provided by an embodiment of the present application. Detailed Description of the Embodiments

[0013] In the following description, specific details such as specific system architectures and technologies are presented for the purpose of illustration rather than limitation, so as to provide a thorough understanding of the embodiments of the present application. However, those skilled in the art should clearly understand that the present application can also be implemented in other embodiments without these specific details. In other cases, detailed descriptions of well-known systems, devices, circuits, and methods are omitted to avoid unnecessary details from obstructing the description of the present application.

[0014] It should be understood that when used in the specification and claims of the present application, the term "comprising" indicates the presence of the described features, wholes, steps, operations, elements, and / or components, but does not exclude the presence or addition of one or more other features, wholes, steps, operations, elements, components, and / or their combinations.

[0015] It should also be understood that the term "and / or" as used in the specification and claims of the present application refers to any combination and all possible combinations of one or more of the associated listed items, and includes these combinations.

[0016] As used in the specification and claims of the present application, the term "if" can be interpreted as "when" or "once" or "in response to determining" or "in response to detecting" depending on the context. Similarly, the phrase "if determined" or "if detected [the described condition or event]" can be interpreted as meaning "once determined" or "in response to determining" or "once detected [the described condition or event]" or "in response to detecting [the described condition or event]" depending on the context.

[0017] In addition, in the description of the specification and claims of the present application, the terms "first", "second", "third", etc. are only used for distinguishing descriptions and should not be construed as indicating or implying relative importance.

[0018] The reference to "one embodiment" or "some embodiments" or the like described in the specification of the present application means that a specific feature, structure, or characteristic described in connection with the embodiment is included in one or more embodiments of the present application. Thus, the statements "in one embodiment", "in some embodiments", "in other some embodiments", "in still other embodiments", etc. that appear in different places in this specification do not necessarily all refer to the same embodiment, but mean "one or more but not all embodiments", unless otherwise specifically emphasized in other ways. The terms "comprising", "including", "having", and their variants all mean "including but not limited to", unless otherwise specifically emphasized in other ways.

[0019] In related technologies, international trade teaching often uses fixed cases, which cannot be personalized according to the abilities, risk preferences and learning progress of different students, resulting in some students being over-challenged or under-challenged. Traditional cases are usually based on past trade rules and market environments, which are difficult to adapt to the frequent changes in international trade policies and cannot allow students to learn and respond to the latest market changes in a timely manner.

[0020] To solve the above problems, the embodiment of the present application provides a method and device for dynamically generating international trade teaching cases. In the method, the static data (course grades and error rates of international trade courses), behavioral data (the deviation of transaction prices from the baseline and the transaction rate) and trait data (stress resistance index and risk preference) of students are first obtained, and then the ability portrait of students is constructed based on the static data, behavioral data and trait data. Then, according to the ability portrait, real-time trade data (international trade policies, government announcements and media news) and trade knowledge graph (based on international trade rules), international trade teaching cases are generated, and finally, the operation data of students (operation data generated by students processing international trade teaching cases, including current error rate, current risk preference and current error mode) are obtained, and the generation process of international trade teaching cases is optimized based on the operation data. The method generates teaching cases by combining students' ability portraits, real-time trade data and trade knowledge graphs, avoiding the limitations of fixed cases, so that each student's learning process can match their actual situation, thereby improving the learning effect. Teaching cases can reflect the latest developments in the field of international trade in real time, which is conducive to the timeliness and relevance of teaching content, so that students can learn the most cutting-edge knowledge and skills. This method can make dynamic adjustments based on students' real-time operational feedback, improve the real-time adaptability of teaching cases, and continuous monitoring of operational data allows teachers and systems to adjust teaching strategies and case designs in a timely manner, improving the efficiency and accuracy of teaching. This method can overcome the problem of fixed cases and content being out of touch in traditional teaching methods, and achieve dynamic, personalized, and real-time optimized generation of international trade teaching cases, which can greatly improve the accuracy of teaching and learning effects, meet the needs of differentiated teaching, and thus enhance students' learning experience and ability development.

[0021] The method for dynamically generating international trade teaching cases provided in the embodiment of the present application can be applied to a device for dynamically generating international trade teaching cases. At this time, the device for dynamically generating international trade teaching cases is the executing entity of the method for dynamically generating international trade teaching cases provided in the embodiment of the present application. The embodiment of the present application does not impose any restrictions on the specific type of the device for dynamically generating international trade teaching cases.

[0022] For example, the dynamic generation device of international trade teaching cases can be a mobile phone, a tablet computer, an augmented reality (AR) / virtual reality (VR) device, a laptop computer, an ultra-mobile personal computer (UMPC), a netbook, a personal digital assistant (PDA), a desktop computer, a computer, a laptop computer, customer premises equipment (CPE) and / or other devices used to communicate on wireless systems and next-generation communication systems, such as mobile terminals in 5G networks or mobile terminals in future evolved public land mobile networks (PLMN), etc.

[0023] In order to better understand the method for dynamically generating international trade teaching cases provided in the embodiment of the present application, the specific implementation process of the method for dynamically generating international trade teaching cases provided in the embodiment of the present application is exemplarily introduced below.

[0024] Figure 1 A schematic flow chart of a method for dynamically generating international trade teaching cases provided in an embodiment of the present application is shown. The method for dynamically generating international trade teaching cases includes: S100, obtaining the static data, behavioral data and trait data of the students. The static data includes the course grades and error rates of the international trade course, the behavioral data includes the deviation of the transaction price from the baseline and the transaction rate, and the trait data includes the stress resistance index and risk preference.

[0025] Understandably, static data is students’ course grades and error rates, which do not change much during the student’s learning process. Behavioral data reflects how students interact in learning and activities. This type of data is dynamic and changes over time. Trait data reflects students’ behavioral habits, stress tolerance, etc. This data can be obtained through questionnaires, psychological tests, or behavioral pattern analysis.

[0026] For example, relevant grade records can be extracted from the education management system or online teaching platform, and the grades can include courseware grades, test grades, homework grades, etc. The error rate can be calculated by the student's performance in answering questions in homework, tests or simulated transactions in the international trade course. For example, in an online test, the error rate can be calculated by the proportion of questions answered incorrectly by students to the total number of questions. The student's error rate data can be automatically counted and recorded through an online learning platform or an examination system (such as Quizlet or Google Forms).

[0027] In a scenario simulating international trade, the deviation between the student's transaction price and the benchmark price can be used as an indicator to measure the student's trading decision-making ability. For example, when students participate in virtual trade negotiations or simulated markets, their transaction prices can be recorded and compared with the benchmark price to calculate the deviation. The transaction price data may come from a simulated trading platform or actual transaction data recorded through artificial intelligence or systems during corporate internships. For example, use simulation platforms such as MarketPlace Simulations or SimTrade to record students' trading behaviors.

[0028] The transaction success rate reflects the frequency of students successfully completing transactions in simulated trade or actual operations and is calculated by statistically analyzing the ratio between the number of transactions initiated by students and the number of transactions ultimately concluded. It can be obtained through a trading system, an online simulation platform, or any actual transaction recording system that interacts with students.

[0029] The stress resistance index of students can be evaluated through simulated high-pressure scenarios or psychological tests. For example, use psychological scales (such as stress assessment scales) to evaluate students' ability to withstand stress during the learning process. It can be measured through regular questionnaires or stress test tools (such as stress assessment tools in psychology), or by analyzing students' performance when facing stress in simulated decision-making.

[0030] Risk preference can be obtained through decision analysis in simulated trading. For example, whether students' decisions in simulated international trade tend to be high-risk, high-return, or more inclined to low-risk, stable transactions. It can be inferred through questionnaires (such as risk preference scales) or through students' actual operating behaviors in the simulated environment. For example, the risk preference of students when making choices in a virtual market can be analyzed, or data can be collected through simulated financial trading tools (such as stock market simulation platforms).

[0031] Through this step, static data, behavioral data, and trait data of students can be comprehensively and systematically collected and analyzed, providing support for formulating personalized teaching plans and improving students' learning effects.

[0032] S200. Based on static data, behavioral data, and trait data, construct a student ability profile.

[0033] Exemplarily, due to the large differences in the dimensions and distributions of static data, behavioral data, and trait data, standardization (z-score standardization) or normalization can be used to standardize these data when constructing the ability profile.

[0034] The importance of course grades can be weighted using the weighted average method, and the error rate can be used as a reverse indicator. The higher the error rate, the lower the academic ability index, thereby calculating the student's academic ability index.

[0035] The deviation of transaction price from the baseline and the transaction rate can be summarized into a learning behavior index in a weighted manner to reflect students' ability in actual operations.

[0036] The stress resistance index and risk preference can be quantified between 0 and 1, and the two can be combined into a psychological trait index.

[0037] Weights can be assigned to the academic ability index, learning behavior index, and psychological trait index, and the total ability score can be calculated by weighted average. Regression models, cluster analysis, or principal component analysis (PCA) can be used to merge various data to form a multi-dimensional student ability portrait.

[0038] After constructing the student's ability portrait, you can use visualization tools (such as radar charts, bar charts, etc.) to display the student's ability distribution, so that teachers can quickly identify the student's strengths and weaknesses and provide personalized guidance. The three dimensions of academic ability, learning behavior, and psychological characteristics can be displayed in a radar chart to help teachers see the student's strengths and weaknesses at a glance; through distribution charts or scatter plots, you can analyze the distribution of students' various abilities and find commonalities or differences in the group.

[0039] This step can not only help teachers better understand students’ learning status, but also provide data support for personalized teaching, thereby improving teaching effectiveness and students’ learning experience.

[0040] In one possible implementation, S200 builds a student's ability profile based on static data, behavioral data, and trait data, including: S210, removing abnormal data from the static data, behavioral data and characteristic data to obtain cleaned static data, behavioral data and characteristic data.

[0041] Exemplarily, for static data: you can use statistical charts (such as histograms and box plots) to view the distribution of the data and identify values ​​that are significantly deviated from the normal range. You can set a reasonable threshold based on the actual range of the data. For example, course grades generally range from 0 to 100, and the error rate is generally not close to 0% or 100%. If a student's score is -10 points or above 100%, it is obviously abnormal data. You can calculate the mean and standard deviation of the data, and use the Z-Score to identify outliers. Z-Score values ​​greater than 3 or less than -3 are generally considered outliers. If the amount of data is large and some outliers are obviously input errors or unreasonable, you can choose to delete these outliers; if you do not want to lose data, you can replace the outliers with reasonable values ​​(such as mean, median, etc.).

[0042] For behavioral data: The deviation degree of the data can be identified by calculating the standard deviation of the behavioral data. Data points exceeding 2 or 3 times the standard deviation may be outliers. For cases where the transaction price deviates too much from the baseline (e.g., price fluctuations exceed a certain threshold) or the transaction rate is extremely high (close to 100%), rules can be set for elimination, and the specific threshold can be set based on historical data and the actual business scenario. For extremely unreasonable behavioral data, it can be directly deleted. If the data fluctuates greatly, methods such as the moving average method can be used for smoothing to reduce noise.

[0043] For trait data: The rationality of the data can be judged by statistical methods. For example, if the stress resistance index varies within a certain range (e.g., from 0 to 10), if extremely high or low values appear, it can be further processed. If some trait data is obviously unreasonable, the mean or median of the trait data can be used for substitution. Clustering algorithms (such as K-Means) can be used to identify and eliminate outliers in trait data. Outliers will be assigned to smaller groups or have different behavioral patterns from most samples. A reasonable range can be set according to the actual business scenario of the trait data to filter out data beyond this range.

[0044] The elimination of abnormal data is a key step in ensuring data quality. The accuracy of the data directly affects the quality of the generated teaching cases. Through this step, outliers in static data, behavioral data, and trait data can be effectively eliminated, which is beneficial to the accuracy and usability of the data, thus providing high-quality data support for subsequent analysis and model training.

[0045] S220, extract negotiation ability features from the cleaned behavioral data, and extract compliance ability features from the cleaned static data and the cleaned behavioral data.

[0046] It can be understood that negotiation ability features reflect the decision-making methods, risk management capabilities, strategy selections, etc. of students in simulated transactions or negotiations.

[0047] Exemplarily, for extracting negotiation ability features: The mean and standard deviation of the deviation degree of the transaction price from the baseline can be calculated, which reflects the overall decision-making behavior of students in multiple transactions (such as the degree of risk-taking). The maximum deviation degree of the transaction price from the baseline among all transactions of a student can be used as a feature of the student's extreme decision-making ability when facing negotiations. The positive and negative directions of the deviation of the transaction price from the baseline can be analyzed to know whether the student tends to close deals at high prices or low prices. The average transaction rate of a student in multiple transactions can be calculated to measure the overall performance of the student in transactions. Negotiation ability features can include the mean, standard deviation, and maximum deviation degree of the deviation degree of the transaction price from the baseline, the positive and negative directions of the deviation of the transaction price from the baseline, and the average transaction rate.

[0048] It can be understood that the compliance ability characteristics reflect whether students follow rules, laws, and compliance requirements when dealing with international trade teaching cases, and involve the degree of understanding and implementation of regulations.

[0049] Exemplarily, for extracting compliance ability characteristics: The compliance ability characteristics may include knowledge mastery, compliance error rate, baseline compliance degree, and compliance decision-making. The knowledge mastery can be evaluated by calculating the total score or unit score of students in the course to assess the degree of students' mastery of compliance-related knowledge.

[0050] The compliance error rate can be calculated by calculating the error rate of students in compliance-related tasks to infer whether students can effectively comply with the rules, and the error types involved in rule application in the tasks can be analyzed to identify compliance weaknesses (such as incorrect understanding of tariff rules, import and export restrictions, etc.).

[0051] The baseline compliance degree can analyze the deviation of the transaction price from the baseline. A smaller deviation may indicate that students are more inclined to comply with rules and standards.

[0052] The compliance decision-making can be inferred by analyzing the transaction rate and the deviation of the transaction price from the baseline whether students follow market rules and make compliant decisions in actual transactions.

[0053] Through this step, strong data support can be provided for subsequent teaching case design and teaching effect evaluation.

[0054] S230, extract the associated knowledge points between the negotiation ability characteristics, compliance ability characteristics, trait data and the knowledge points of the international trade course, and calculate the association degree of each associated knowledge point to construct a knowledge association graph.

[0055] It can be understood that the knowledge points of the international trade course may include international trade theories (such as the theory of comparative advantage, the theory of absolute advantage, the Ricardo model, the Heckscher-Ohlin model, etc.), international trade policies (tariffs, quotas, anti-dumping, import and export restrictions, etc.), international trade organizations and regulations (WTO rules, free trade agreements, regional economic integration, etc.), trade negotiation skills (negotiation strategies, application of game theory in trade negotiations, etc.), international market risk management (exchange rate risk, political risk, credit risk, etc.). The association degree is an indicator to measure the relationship between the negotiation ability characteristics, compliance ability characteristics, trait data and the knowledge points of the international trade course.

[0056] Exemplarily, the degree of association can be measured by calculating the correlation coefficients (Pearson correlation coefficient or Spearman rank correlation coefficient) between different ability characteristics (negotiation ability characteristics, compliance ability characteristics, and trait data) and the knowledge points of the course. Text-based matching analysis can be used, such as analyzing the semantic relationship between the text descriptions of students' ability characteristics and the knowledge points of the course through natural language processing (NLP) technology to calculate the text similarity. Machine learning methods (such as multiple regression, support vector machines, etc.) can be used to train a model to predict the degree of association between features and knowledge points.

[0057] Use the degree of association to construct a knowledge association graph. The nodes of the graph represent the knowledge points, negotiation ability characteristics, compliance ability characteristics, and trait data of the international trade course, and the edges represent the relationships between the knowledge points and ability characteristics. The weights of the edges can be set according to the degree of association. For example, if the degree of association between risk preference and international market risk management is 0.8, then the weight of the edge between these two nodes in the graph is 0.8.

[0058] Tools such as NetworkX (a Python library) or Gephi can be used to generate and visualize the knowledge association graph, showing the degree of association between each knowledge point and the characteristics. The strength of the association can be represented by colors, the thickness of the edges, etc.

[0059] Through this step, the relationships between certain knowledge points and negotiation ability, compliance ability, and trait data of students in the international trade course can be revealed, thereby helping to optimize teaching strategies and provide more targeted personalized learning paths for students.

[0060] S240, based on the knowledge association graph, negotiation ability characteristics, compliance ability characteristics, and stress resistance index, use a clustering algorithm to perform group division to obtain the portrait categories of students.

[0061] Exemplarily, an appropriate clustering algorithm can be selected according to the characteristic types and goals of the students. K-means clustering is suitable for cases where the dataset is large and the dimensions are small. The data is divided by setting the number of clusters (k), and each data point is assigned to the nearest cluster center. Hierarchical clustering is suitable for understanding the hierarchical relationships between student groups. The similarity between students is shown through a dendrogram, and the data points are grouped according to the distance. DBSCAN is suitable for processing unstructured data with noise and can automatically detect the number of clusters. Considering that there may be certain differences between student groups in this application, the K-means clustering method can be selected to perform group division on the characteristics of students.

[0062] Clustering model training process: students' characteristic data can be collected and used as clustering input. Each student has a set of characteristic vectors including negotiation ability, compliance ability, stress resistance index and knowledge point relevance. SSE (sum of squared errors) under different cluster numbers k can be calculated, and the k value corresponding to the inflection point where the SSE decreases less can be selected. The clustering effect can be evaluated by calculating the distance between the data points in each cluster and the data points outside the cluster. The closer the silhouette coefficient is to 1, the better the clustering effect. The Euclidean distance (or other distance metrics) can be used to calculate the distance between students and each cluster center. According to each student's characteristic vector, they are assigned to the nearest cluster center, thus forming different student groups and labeling each student group, such as high-risk, high-transaction rate group, low-risk, high-compliance group, etc.; the typical characteristics of each group can be summarized by viewing the characteristic mean of each group (such as the mean of negotiation ability characteristics, the mean of compliance ability, etc.).

[0063] Project students' characteristics into a two-dimensional plane or three-dimensional space, identify different student groups by dots of different colors, display the center point of each group, and mark the distance relationship between groups to help identify similarities and differences between groups.

[0064] The correlation degree, negotiation ability characteristics, compliance ability characteristics and stress resistance index in the student's knowledge association map can be input into the K-means algorithm, and the student can be matched with the corresponding group to obtain the student's portrait category.

[0065] This step can not only help understand students’ characteristics in international trade courses, but also adjust the content of teaching cases according to students’ individual characteristics, thereby improving teaching effectiveness and students’ learning experience.

[0066] S250, constructs students’ ability portraits based on knowledge association maps and portrait categories.

[0067] For example, the knowledge association map and the student's profile category can be integrated to obtain the student's ability profile. The student's ability profile can be presented in an intuitive form through data visualization tools (such as radar charts, bar charts, etc.), showing the student's scores in various ability dimensions, helping teachers to better understand the student's ability strengths and weaknesses, thereby providing more personalized and effective teaching support.

[0068] Optionally, at S240, based on the knowledge association graph and the negotiation ability characteristics, compliance ability characteristics, and stress resistance index, a clustering algorithm is used to divide the groups to obtain student portrait categories, including: S241, based on negotiation ability characteristics, compliance ability characteristics and stress resistance index, uses clustering algorithm to divide groups and obtain preliminary portrait categories.

[0069] Exemplarily, the clustering model trained in step S240 can be used, and the students' negotiation ability characteristics, compliance ability characteristics and stress resistance index can be used as input data of the clustering model. The model can divide students into different groups according to their characteristics, thereby outputting preliminary portrait categories of the students.

[0070] S242, based on the correlation degree of each associated knowledge point in the knowledge association graph, the preliminary portrait category is optimized to obtain the student's portrait category.

[0071] For example, when students are initially classified into a certain group, the knowledge association graph can be used to further optimize the student's portrait category. If a student shows a weaker specific ability in a certain group (such as low compliance ability), and this ability is strongly related to certain core knowledge points, the student's ability can be strengthened by increasing the teaching content and exercises of the core knowledge points; if the student's ability portrait category is inconsistent with the correlation between the knowledge points, the group division can be readjusted.

[0072] The optimized student portrait categories not only reflect students' performance in negotiation, compliance, stress resistance and other dimensions, but can also further refine student portraits based on the depth of knowledge mastery. The optimized student portrait categories will be more refined and able to reflect students' comprehensive abilities.

[0073] S300 generates international trade teaching cases based on capability profiles, real-time trade data and trade knowledge graphs. Real-time trade data includes international trade policies, government announcements and media news, and trade knowledge graphs are built based on international trade rules.

[0074] It is understandable that capability profiles can adjust the difficulty and complexity of cases. Real-time trade data can include changes in international trade policies, announcements from the General Administration of Customs, social media public opinion (such as hot topics on Reddit / r / SupplyChain), news (BBC, Reuters), etc., reflecting the actual situation of the current international trade environment. The trade knowledge graph is based on international trade rules and theories, which can help teaching cases have realism and depth.

[0075] For example, the case content can be customized by analyzing the student's ability profile. If the student's academic ability index is high, more challenging cases involving more complex international trade policies and advanced rule applications can be generated; if the academic ability is weak, basic and specific cases can be generated. For students with low transaction rates, more simulated decision-making tasks can be designed to improve their decision-making ability. If the student's stress tolerance is low, fewer stress situations can be designed, and more time and support can be provided; if the student's risk preference is high, more risky trade cases can be designed to allow students to experience and deal with high-risk situations in a simulated environment.

[0076] Real-time international trade policies, government announcements, and media news can be utilized to design cases that are closely related to the actual international trade environment. Cases can be generated based on the latest trade policies, tariff changes, free trade agreements, etc. of various countries. For example, if a country issues a new tariff policy, a case can be generated based on this policy for students to analyze the impact of the new policy on the international market and require students to propose coping strategies; if a country issues import and export restrictions, trade protection measures, etc., cases can be generated through these announcements and students are required to simulate how enterprises respond to these policy changes. Dynamic cases can be generated based on real-time international news, such as major events in the international market, economic sanctions, geopolitical conflicts, etc., for students to analyze how these news affect international trade.

[0077] The international trade rules in the trade knowledge graph can be used to ensure that the generated teaching cases meet the requirements of laws and regulations. Best practices or strategies in the trade knowledge graph can be used, and students can optimize their decisions by applying these strategies. The multi-dimensional relationships in the trade knowledge graph can help students understand the complexity of international trade from multiple perspectives.

[0078] By combining the student's ability profile, real-time trade data, and the trade knowledge graph, a specific teaching case can be generated.

[0079] By combining the student's ability profile, real-time trade data, and the trade knowledge graph, dynamic, personalized teaching cases that are closely related to the actual international trade environment can be generated. This can not only improve students' learning interest but also help students master practical application skills.

[0080] In a possible implementation, in S300, according to the ability profile, real-time trade data, and the trade knowledge graph, an international trade teaching case is generated, including: In S310, according to the ability profile, real-time trade data, and the trade knowledge graph, initial generation parameters are determined.

[0081] It can be understood that the initial generation parameters may include the difficulty of the teaching case, the content of the case, and the coverage of knowledge points.

[0082] Exemplarily, the difficulty level suitable for the student can be determined according to the student's ability profile. For example, if the student has strong academic ability, more difficult cases can be generated; if the academic ability is weak, simpler cases can be generated. Combining real-time trade data (such as international trade policies, global economic situation, etc.) and a trade knowledge graph, cases related to the current situation are generated. For example, when a new tariff policy is introduced, a case related to this policy can be generated. According to the fields with weaknesses in the student ability profile, the knowledge points in this field are appropriately strengthened. For example, if the student is weak in negotiation strategies, more content related to negotiation strategies can be added when generating cases.

[0083] Optionally, in S310, according to the ability profile, real-time trade data, and trade knowledge graph, initial generation parameters are determined, including: In S311, according to the real-time trade data, the case theme and market background are determined.

[0084] Exemplarily, the real-time trade data can be preprocessed. Irrelevant information such as advertisements and irrelevant links is cleared; the data format is converted into structured data that can be used for analysis, such as CSV, JSON, etc.; useful key information (such as policy names, implementation dates, industries affected, countries mainly affected, etc.) is extracted from the text.

[0085] Keywords in the text, such as tariff increase, trade agreement, market volatility, etc., can be extracted through natural language processing technology (NLP) to identify the core topics. Algorithms (such as deep learning models like LDA or BERT) can be used to cluster the text to identify potential themes from it.

[0086] It can be understood that the market background is used to describe the overall situation, environment, and market trends of the trade market related to the case. The market background can be inferred based on the real-time trade data.

[0087] Exemplarily, the scale, demand, and trends of the target market can be analyzed. For example, through data analysis, the industries most affected by tariffs (such as electronics, automobiles, mechanical equipment, etc.) are determined. The regions or countries affected can be determined according to the real-time trade data, such as which markets or countries' exporters are most affected by the tariff policies of a specific country. After the policy is implemented, the main competitors in the market and their coping strategies can be analyzed. For tariff policies, which countries or enterprises may benefit or be damaged can be identified.

[0088] This step is beneficial to the timeliness and real-world relevance of teaching cases, enabling students to learn and make decisions in an environment close to the real world.

[0089] In S312, according to the ability profile and knowledge association graph, the key knowledge points are determined.

[0090] It can be understood that the core objective of this step is to extract the knowledge points most relevant to the student's current ability from the knowledge association graph based on the student's learning history and knowledge mastery, and provide a basis for personalized teaching.

[0091] Exemplarily, by analyzing the student's ability profile, the weaknesses of the student in different fields (such as international trade theory, politics, etc.) can be determined. For example, if the student has a relatively high error rate in tariff policies and does not have a deep understanding of anti-dumping policies, then these two fields can be marked as the student's weak points. Based on the relationships between nodes in the knowledge association graph, the key knowledge points associated with the student's weak points can be found; by evaluating the student's knowledge mastery in certain fields, if the mastery in this field is relatively low, the relevant knowledge points in this field can be marked as key knowledge points.

[0092] This step effectively identifies the key knowledge points related to the student's weak points through the student's specific performance in the field of international trade and the association relationships between nodes in the knowledge association graph. It can not only help students identify their own learning weak points, but also provide targeted learning paths, improving learning efficiency and quality.

[0093] S313. According to the knowledge association graph, identify the student's knowledge weak points, and based on the knowledge weak points and the trade knowledge graph, determine the decision branches.

[0094] Exemplarily, the mastery level of the student in knowledge points can be evaluated through the association degree in the knowledge association graph. The knowledge points with relatively low association degrees can be selected as the student's knowledge weak points and marked.

[0095] Based on the student's knowledge weak points, the decision nodes related to the knowledge weak points can be extracted from the trade knowledge graph. According to the knowledge weak points, multiple branches are constructed at the decision nodes to help the student choose appropriate action plans. Each decision branch provides different option selections based on the student's knowledge weak points. Each decision branch can give an expected result, including positive results (such as increased market share, profit growth) and negative results (such as market loss, legal risks). Combining with the knowledge association graph, the expected results are generated to help the student understand the potential impacts of each decision.

[0096] Through this step, it can help students improve their abilities in weak fields through targeted decision-making training when dealing with complex international trade situations.

[0097] S314. Determine the case difficulty according to the ability profile.

[0098] Exemplarily, the competency profile data can be standardized to facilitate the comparison of different types of competency scores on the same scale. The data for each dimension (academic ability, negotiation ability, compliance ability, and stress resistance ability) can be transformed into values between 0 and 1, where 0 represents a lower ability and 1 represents a higher ability.

[0099] A model can be used to weight each competency dimension based on specific weights, which are automatically determined by the model and may be based on historical data or derived from the training set. The model can be a regression model that learns the relationship between the competency dimension and the case difficulty through regression analysis; or a machine learning model that uses algorithms such as decision trees, support vector machines (SVMs), or neural networks to predict the case difficulty corresponding to the competency characteristics of different students. The case difficulty is determined by dividing the difficulty according to the ability level and comparing the weighted competency dimension with the preset difficulty levels.

[0100] This step can automatically adjust the complexity of the case based on the actual performance of the student to provide personalized teaching content and decision-making challenges.

[0101] S315, Integrate the case theme, market background, key knowledge points, decision branches, and case difficulty to obtain the initial generation parameters.

[0102] Exemplarily, the case theme, market background, and key knowledge points can be transformed into structured data (such as using natural language processing techniques to transform text data into keywords, entities, and categories). For the case difficulty, it can be transformed into a standardized score or level (such as 1 representing low difficulty, 2 representing medium difficulty, and 3 representing high difficulty). Different decision branches can be associated with the knowledge points to form a decision tree or graph, clarifying the knowledge points and expected results corresponding to each decision point.

[0103] The parameters after the above processing can be integrated into a unified generation parameter object, which can be a structure or data frame containing all elements, providing a basis for generating targeted international trade teaching cases. For examples, please refer to Figure 5 。

[0104] S320, Adjust the difficulty of the initial generation parameters based on the three-dimensional difficulty matrix to obtain the case generation parameters. Among them, the three-dimensional difficulty matrix is constructed based on the competency profile.

[0105] It can be understood that the three dimensions of the three-dimensional difficulty matrix can include students' knowledge mastery, practical ability, and stress resistance. Knowledge mastery is used to measure students' mastery of international trade course knowledge, students' practical ability is used to reflect students' ability in actual operations, and stress resistance is used to measure students' performance when facing high-pressure situations. Students with strong stress resistance can handle more uncertain and high-risk situations and are suitable for facing more challenging teaching cases.

[0106] For example, knowledge mastery can be evaluated through course grades and error rates in the ability profile. Students with higher knowledge mastery can handle more complex cases, while those with lower knowledge mastery are more suitable for generating basic cases. Practical ability can be evaluated through students' performance in simulated transactions or actual operations in the ability profile (such as transaction rate, transaction price deviation from the baseline, etc.).

[0107] The difficulty level of teaching cases can be determined using a three-dimensional difficulty matrix. When knowledge mastery is low, practical ability is weak, or stress tolerance is poor, the generated cases will focus on basic knowledge, simplify situations, and provide more prompts and assistance; when knowledge mastery is high, practical ability is strong, and stress tolerance is strong, the generated cases will be more complex, containing multiple knowledge points and complex situations, requiring students to make high-risk, high-pressure decisions.

[0108] You can choose different international trade topics and knowledge points according to the difficulty level. Basic content can be the theoretical basis of international trade, tariffs, quotas, basic contract terms, etc.; advanced content can be global trade risk management, multinational negotiation strategies, complex trade negotiations, and coping with global economic uncertainty.

[0109] Adjusted case generation parameters enable students to learn in contexts appropriate to their current ability level and provide appropriate challenges and support.

[0110] Optionally, S320, adjusting the difficulty of the initial generation parameters based on the three-dimensional difficulty matrix to obtain case generation parameters includes: The three-dimensional difficulty matrix can include complexity, uncertainty, and cultural conflict. An example table of the three-dimensional difficulty matrix is ​​as follows: S321, adjusting the complexity of the initial generation parameters based on the compliance capability characteristics to obtain the first generation parameters.

[0111] It can be understood that complexity refers to the complexity of each link involved in international trade, including the number of nodes in trade decision-making, the application of policies and rules, etc., which can be reflected as the coupling degree of trade links, that is, the number of decisions, policies and rules involved.

[0112] Exemplarily, compliance ability characteristics (knowledge mastery, compliance error rate, baseline compliance, compliance decision-making) can be standardized to convert characteristics with different dimensions into a unified scoring range (such as between 0 and 1), which is convenient for calculation and comparison.

[0113] Weights can be assigned to each characteristic. Based on the standardized compliance ability characteristics and combined with the weights, a complexity adjustment factor can be calculated. The complexity adjustment factor is used to adjust the complexity of the case, and its range can be between 0.1 (single) and 1.0 (full chain). Adjust the complexity of the case based on the calculated complexity adjustment factor.

[0114] S322. Adjust the uncertainty of the initial generation parameters based on the stress resistance index to obtain the second generation parameters.

[0115] It can be understood that uncertainty represents uncertain factors in the market or policy environment in international trade cases, such as random events (such as policy changes, market fluctuations, etc.), decision-making risks, and information asymmetry. The adjustment range of uncertainty can be from 0 (no event) to 1.0 (multiple high-impact events).

[0116] Exemplarily, the stress resistance index can be standardized and converted to a value between 0 and 1. If the high stress resistance index is high (close to 1.0), the uncertainty can be adjusted to a higher value (such as 0.7 to 1.0) to simulate a complex and dynamically changing trade environment, bringing more risks and unpredictable factors; if the low stress resistance index is low (close to 0.0), the uncertainty can be adjusted to a lower value (such as 0.1 to 0.3) to reduce the uncertain factors in the decision-making environment and maintain a relatively stable market background and decision-making situation. The uncertainty adjustment factor calculation formula can be used to calculate the uncertainty adjustment factor based on the stress resistance index. Uncertainty adjustment factor calculation formula , where represents the uncertainty adjustment factor, represents the stress resistance index, and the dynamic range of uncertainty is adjusted by multiplying (1.0 - 0.1) to ensure that the uncertainty of students with a low stress resistance index does not exceed a certain upper limit.

[0117] Through the standardization of the stress resistance index and the calculation of the uncertainty adjustment factor, it is beneficial for the generated teaching cases to match the stress resistance ability of students in terms of challenge, enabling students to learn in an environment suitable for their coping ability and gradually improving their decision-making ability in high-pressure situations.

[0118] S323. Adjust the cultural conflict of the initial generation parameters based on the negotiation ability characteristics to obtain the third generation parameters.

[0119] It can be understood that cultural conflicts are caused by differences in values and behavior patterns under different cultural backgrounds. Cultural conflicts in negotiations reflect how students in different cultural environments respond to and handle negotiation challenges in a multi-cultural context.

[0120] Exemplarily, the negotiation ability characteristics (the mean, standard deviation, and maximum deviation from the baseline of the transaction price, the positive and negative directions of the transaction price deviation from the baseline, and the average transaction rate) can also be standardized and converted to a value between 0 and 1. The adjustment range of the cultural conflict intensity can range from 0 (no cultural differences) to 1.0 (extremely large cultural differences). Adjust the intensity of cultural conflicts according to the negotiation ability characteristics, and calculate the cultural conflict adjustment factor by weighted averaging the negotiation ability characteristics. Assuming that the contributions of the five characteristics are equally important, the weights corresponding to the five weights are all 0.2. The larger the calculated cultural conflict adjustment factor (the closer to 1), the greater the intensity of cultural conflicts, and vice versa.

[0121] By using the method of combining negotiation ability characteristics and weights to calculate the cultural conflict adjustment factor, and adjusting the intensity of cultural conflicts in the case according to this factor, a third generation parameter is generated, which is conducive to teaching cases adapting to students' negotiation ability and cultural adaptability.

[0122] S324, Combine the first generation parameter, the second generation parameter, and the third generation parameter to obtain the case generation parameter. Among them, the case generation parameter includes case theme, market background, key knowledge points, decision branches, complexity, uncertainty, and cultural conflicts.

[0123] Exemplarily, the above three generation parameters (complexity, uncertainty, cultural conflicts) can be combined with the case theme, market background, key knowledge points, and decision branches to form a complete case generation parameter for subsequent teaching case generation.

[0124] Through these steps above, it is possible to analyze students' performance in different ability dimensions, dynamically adjust the difficulty and decision complexity of the case, and facilitate the generated cases to meet the personalized learning needs and ability levels of students.

[0125] S330, According to the case generation parameter, construct a Prompt, and according to the Prompt, call the large language model fine-tuned by LoRA to generate international trade teaching cases.

[0126] Exemplarily, according to the case generation parameter, a suitable Prompt can be constructed to guide the large language model fine-tuned by LoRA to generate international trade teaching cases. The Prompt is the input provided to the model, and the model generates corresponding teaching cases according to the input content.

[0127] The structure of the Prompt can include students' portrait information (such as academic ability, negotiation ability, compliance ability, stress resistance ability, etc.), the background and context of the case (clearly describe the trade background required for the teaching case and the specific situation faced by the students), goals and tasks (teaching objectives), difficulty and support (the difficulty of the teaching case and whether additional hints or support are needed).

[0128] Suppose the Prompt is generated based on the students' ability portrait, three-dimensional difficulty matrix, real-time trade data, and trade knowledge graph. Please refer to Figure 2 .

[0129] It can be understood that LoRA (Low-Rank Adaptation) is a method for fine-tuning large language models. By making a small number of adjustments based on a pre-trained model, it can adapt to specific tasks. In this application, a large language model fine-tuned with LoRA can be used to generate customized international trade teaching cases.

[0130] Exemplarily, by providing the above Prompt to the large language model fine-tuned with LoRA, the model can generate corresponding teaching cases according to the Prompt. The generated teaching cases can include background settings, describing the background of trade policy changes and the market environment; decision-making tasks, the decisions that students need to make and the possible impacts of these decisions; teaching objectives, helping students improve negotiation skills, compliance, and stress resistance; case hints, case hints matching the students' abilities.

[0131] Suppose the above Prompt example is input into the large language model fine-tuned with LoRA, and the model outputs an international trade teaching case. Please refer to Figure 3 .

[0132] Through this step, an actual international trade teaching case can be generated, which combines academic knowledge and actual situations, and can improve students' decision-making ability, risk management ability, and compliance awareness. It not only helps students better understand the complex factors in international trade but also effectively exercises their thinking and decision-making abilities in high-pressure situations.

[0133] Optionally, in step S330, according to the case generation parameters, construct the Prompt, including: S331, determine the variables in the role definition layer according to the case theme and market background of the case generation parameters. The variables in the role definition layer include country parameters and trade environment.

[0134] Exemplarily, natural language processing (NLP) technology can be used to extract the country name from the case theme. For example, if it is mentioned in the case theme that a certain country has imposed additional tariffs on imported electronic products, the country can be identified through entity recognition technology and judged as the target country.

[0135] Text analysis techniques can be used to extract descriptions about the market from the market background and identify relevant trade environment variables. For example, if the market background mentions having a significant impact on the global electronics market, text analysis techniques can be used to identify the electronics market as the target industry and extract key metrics of this market (such as demand volume, price fluctuations, etc.). If the international market fluctuations are mentioned in the market background, the system can automatically analyze and estimate the risk level of this environment.

[0136] Integrate country parameters and trade environment parameters into the role definition layer, and use the role definition layer as one of the inputs for case generation, which helps the generated teaching cases to accurately reflect the decision-making background of a specific country in a specific market environment and provides the necessary context for subsequent case generation.

[0137] S332. Determine the variables in the basic setting layer based on the market background, key knowledge points, and decision branches of the case generation parameters. Among them, the variables in the basic setting layer include trade terms, commodity categories, and participating roles.

[0138] Exemplarily, trade terms involve the conditions and rules of goods transactions, such as payment methods, delivery methods, transfer of ownership of goods, etc. According to the scenarios provided in the market background, NLP techniques can be used to analyze the transaction conditions in the market background and extract relevant trade term vocabulary. For example, if it is mentioned in the market background that electronic products are exported under FOB terms, FOB is identified as a trade term.

[0139] Commodity categories refer to the types of goods involved in the case. The commodity types can be identified from the market background through keyword extraction and entity recognition. For example, if it is mentioned in the market background that a certain country imposes additional tariffs on imported electronic products, electronic products are extracted as the commodity category.

[0140] Participating roles refer to the parties involved in the transaction, including buyers, sellers, governments, etc. Entity recognition techniques can be used to identify the participants in the market background and map them to participating roles (such as governments, buyers, sellers, agents, etc.). For example, if it is mentioned in the market background that the government of a certain country imposes new import tariffs on electronic products, the government of that country is identified as a participating role.

[0141] Trade terms can be further deduced based on key knowledge points (such as tariff policies, market access, etc.). For example, tariff policy is a key knowledge point, and the possible trade terms involved can be deduced based on specific tariff terms (such as CIF terms are usually related to tariffs and transportation conditions).

[0142] Different product categories and participating roles can be determined based on decision branches. For example, if the decision branch involves choosing whether to raise prices in response to tariff changes, then it can be deduced that the product category involves electronic products, as well as the possible roles of buyers and sellers.

[0143] This step facilitates the generation of cases that can generate suitable transaction backgrounds and role information according to the specific market environment and students' knowledge levels, providing personalized scenario settings for subsequent teaching.

[0144] S333. Determine the variables at the teaching requirement level based on the complexity of the case generation parameters, key knowledge points, decision branches, market background, cultural conflicts, and uncertainties. Among them, the variables at the teaching requirement level include case structure complexity, teaching objectives, decision paths, current market dynamics, negotiation scenarios, and random events.

[0145] Exemplarily, the case structure complexity is used to measure the complexity of decision points, rule applications in the case, and how multiple factors (such as policies, market changes, etc.) are combined to affect decisions. The case structure complexity can be determined by complexity. For example, if the complexity adjustment factor value is between 0.1 and 0.3, the case structure complexity at the teaching requirement level can be set to simple; if the complexity adjustment factor is medium, between 0.4 and 0.7, the case structure complexity at the teaching requirement level can be set to medium; if the complexity adjustment factor is between 0.8 and 1.0, the case structure complexity at the teaching requirement level can be set to complex.

[0146] The teaching objectives can be determined by analyzing the key knowledge points and mapping the key knowledge points to the corresponding teaching objectives. For example, the teaching objective corresponding to the tariff policy can be to enable students to understand the basic principles, applications of the tariff policy and its impact on the market; the teaching objective corresponding to market access can be to help students master how to analyze and evaluate the conditions and obstacles of market access.

[0147] The choices that students need to make in the case can be extracted from the decision branches. For example, whether to raise product prices, whether to look for alternative markets, etc. Each decision branch is mapped to the corresponding decision path, and the decision-making process that students need to follow can be designed according to different decision paths, and ensure that each path has clear expected results.

[0148] The current market situation can be extracted from the market background, and based on the description in the market background, the current market dynamics can be analyzed and inferred. For example, if the case mentions that global tariffs have increased and affected the global electronic product supply chain, the market background can be transformed into current market dynamics, including reduced market demand, supply chain disruptions, price fluctuations, etc.

[0149] The negotiation scenario can be adjusted according to the intensity of cultural conflicts. For example, there may be significant cultural differences in the negotiation, and students need to understand and adapt to different negotiation styles; there are no major cultural differences involved in the negotiation, and the decision-making is relatively simple.

[0150] The number of random events and the impact level can be determined according to the uncertainty adjustment factor in the uncertainty. For example, when the uncertainty adjustment factor is 0.8, 3 high-impact random events can be generated, such as changes in tariff policies, changes in global market demand, and changes in competitors' behaviors.

[0151] Integrate the case structure complexity, teaching objectives, decision-making paths, current market dynamics, negotiation scenarios, and random events to generate a complete teaching requirement layer. It can help students better understand the core content and decision-making scenarios in the case.

[0152] S334, generate the decision branches of the parameters according to the case, and quantify the decision consequences of the output specification layer.

[0153] Exemplarily, each decision option, the expected results corresponding to each decision option (subsequent quantification and calculations are based on these results), and the basis for each decision option (different situations of decision consequences are speculated based on these bases) can be extracted from the decision branches.

[0154] Define the relevant quantification indicators according to the key influencing factors in the expected results. The quantification indicators can include changes in market share (percentage), changes in profit (currency unit or percentage), changes in cost (currency unit or percentage), and risk level (value from 0 to 1).

[0155] Pre-defined quantification rules or models (such as market models, economic formulas, risk assessment models, etc.) can be used to calculate the quantification consequences of each option according to the specific information in the decision options. The change in market share can be calculated based on the change in market demand and price adjustment. For example, when the price increases by 10%, calculate the percentage of the decrease in market share based on the market demand elasticity.

[0156] The change in profit can be calculated based on the price adjustment and the change in market share. Based on the pre-defined profit model, combine the change in market share with the change in cost to obtain the change in profit.

[0157] If the decision involves increasing costs (such as finding alternative markets may bring higher entry costs), the change in cost can be calculated.

[0158] The risk level can be evaluated based on the number of random events, the scope of influence, and the students' stress resistance.

[0159] Generate the quantified consequences corresponding to each decision option through the above steps, organize these quantified results according to the decision options, and generate the final decision consequence output. The quantified consequences of each decision option can be organized into a unified format and output as the decision consequence part for subsequent analysis, feedback, or decision support.

[0160] The consequences of each decision can be quantified and accurately reflected in the decision consequences at the output specification level, providing clear feedback on the decision-making impact of students in real-world scenarios.

[0161] In a possible implementation, the method for dynamically generating international trade teaching cases further includes: S301, construct a case structure template. The case structure template includes background, conflict, decision point, and result.

[0162] Exemplarily, constructing a standardized case structure template is to systematically label and generate international trade teaching cases and provide a clear structure for training large language models with LoRA fine-tuning. There are four core parts in the case structure template: The background can provide background information of the case, providing students with a clear context to help them understand the background and general environment in which the whole case occurs. The background can include time, place, market, industry background, relevant global or regional economic changes (such as trade wars, economic crises, etc.), participants (such as companies, countries, trade unions, etc.) and their backgrounds.

[0163] The conflict is the main challenge or problem in the case, which is the core problem for students to solve and provides one or more situations for students to make decisions. It can include core problems (such as price fluctuations, market access issues, trade barriers, etc.), challenges faced (such as uncertainties brought by new trade policies, risks of exchange rate changes, etc.), and opposition of roles (in trade negotiations, goal conflicts between buyers and sellers, opposition in price negotiations, etc.).

[0164] The decision point is where students make choices in the case. Students make corresponding decisions based on the existing information and problems. Each decision point represents the action plan chosen by students and affects the final result. It can include optional solutions, expected results of each solution, and impacts of the decision.

[0165] The result is the possible result after students make decisions, including positive and negative impacts of the decision, which can reflect the interaction between students' decisions and background information and conflicts. It can include successful results, failed results, feedback, and suggestions.

[0166] For an example of the case structure template, please refer to Figure 4 。

[0167] S302. According to the case structure template, annotate the case structure of the verified international trade teaching cases to obtain the training dataset. Among them, the verified international trade teaching cases are international trade teaching cases that comply with international trade rules.

[0168] Exemplarily, the verified cases (verified international trade teaching cases) can be split into words and sentences for grammatical and semantic analysis of each part; identify entity information in the cases, such as countries, companies, policy names, etc.; analyze the dependency relationships between words in the sentences to determine which parts belong to the background, conflict, decision points, or results.

[0169] It is possible to identify information such as time, location, policy changes, etc. in the verified cases and map them to the background template, extract the specific economic situation and policies, and confirm the match with the background. It is possible to search for keywords such as challenges, problems, conflicts, and the interactions of stakeholders, extract the sentences describing the core problems and conflicts, and mark them as the conflict part. It is possible to search for sentences containing keywords such as choices, decisions, responses, etc., extract the consequences corresponding to each choice, and clearly label the different decision paths. It is possible to search for sentences containing keywords such as results, feedback, impacts, etc., and extract the relevant content, generate positive and negative results for each decision path, and label them as the result part. According to the above methods, the extracted content can be filled into the case structure template to generate the standardized case structure corresponding to the verified cases for use as the training dataset.

[0170] Automatically extracting the background, conflict, decision points, and results from the verified international trade teaching cases through natural language processing technology combined with a predefined case structure template and performing annotation is the core of automatic annotation, which helps to generate a structured case dataset. The case dataset can be used for subsequent machine learning model training to improve the quality and pertinence of the generated cases.

[0171] S303. Initialize the parameters of LoRA and select the fine-tuning layer in the large language model.

[0172] It can be understood that the key to the LoRA fine-tuning technology lies in how to initialize the low-rank matrix and ensure that it can effectively customize the training of the large language model (LLM).

[0173] Exemplarily, a pre-trained large language model (LLM), such as models in the GPT series, T5 series, BERT, etc., can be selected as the basis for fine-tuning. The LoRA module adjusts the parameters of the LLM by introducing low-rank matrices. LoRA does not directly update the weights of the entire model but adds a low-rank adaptation matrix to the weight matrix of each layer, which can significantly reduce the number of parameters and computational costs. Initialize a low-rank matrix, the size of which is much smaller than the weights of the original model (LLM), and insert the low-rank matrix into the weight matrix of the LLM. The learning rate of the LoRA module can be set smaller than other parameters because it is specifically used to adjust some parameters of the model. In this way, computational resources can be efficiently utilized and the focus can be on fine-tuning the key parts.

[0174] In the large language model (LLM), there are multiple layers available for fine-tuning, but not all layers need to be fine-tuned. Layers most relevant to the task can be selected for fine-tuning. The fine-tuning layers that can be selected include Transformer layers. The core layers of most models are Transformer layers, and these layers are crucial for the performance of the model; Attention layers. Since the self-attention mechanism plays an important role in language models, sometimes fine-tuning the self-attention layer in the model can bring good results; Feed-Forward layers. Sometimes fine-tuning these layers can help the model better handle specific tasks.

[0175] S304, based on the training dataset, train the large language model and adjust the parameters of LoRA according to the decrease in the loss function in the large language model to optimize the fine-tuning layers in the large language model and obtain the LoRA fine-tuned large language model.

[0176] Exemplarily, cross-entropy loss or maximum likelihood estimation can be selected as the loss function. The purpose of the loss function is to minimize the gap between the predicted output and the target output (true label). Optimizers such as AdamW can be used. AdamW performs well in large-scale training and supports large learning rate adjustments. The low-rank matrix of LoRA can adopt a smaller learning rate to avoid causing too much interference to the parameters of the large model.

[0177] The training dataset can be input, and forward propagation can be performed through the model to calculate the output result of the model. By calculating the difference between the model output and the actual target (i.e., the correct answer or decision), the loss value can be obtained. Use the backpropagation algorithm to pass the loss value back to the model and update the parameters of the LoRA low-rank matrix to reduce the loss.

[0178] The loss function value for each training epoch can be recorded. If the loss decreases steadily, it indicates that the model is learning effectively; if the loss stagnates or increases, it means there may be problems in the training process (such as too high a learning rate, etc.). During the training process, according to the decrease of the loss function, adjust the learning rate and other parameters of the low-rank matrix of LoRA. The parameters of the low-rank matrix of LoRA can be gradually optimized according to the change of the loss function to better adapt to specific tasks. To avoid overfitting, an early stopping strategy can be used. When the loss function does not decrease significantly within a certain period, stop the training.

[0179] When the loss function reaches a stable and low value on both the training set and the validation set, it means that the parameters of the LoRA fine-tuning have been optimized, and the fine-tuning layer of the model has converged to a good state capable of handling the generation of international trade teaching cases. At this time, the final LoRA fine-tuned large language model can be saved and used to generate new international trade teaching cases. The fine-tuned model can be used to generate new cases, and check whether it meets the expected teaching goals and content structure during the generation process. The effect of the model can be further evaluated by calculating indicators such as the accuracy, rationality, and relevance of the cases generated by the model.

[0180] Through the large language model fine-tuned by LoRA, personalized cases for international trade teaching can be efficiently generated. Initialize the low-rank matrix of LoRA and select the appropriate fine-tuning layer. By monitoring the decrease of the loss function and adjusting the parameters of LoRA, finally optimize the large language model to generate high-quality teaching cases. This process improves the training efficiency and generation ability of the model while saving computing resources through a small number of parameter adjustments.

[0181] S400, obtain the operation data of students, and optimize the international trade teaching case generation process based on the operation data. Among them, the operation data is the operation data generated by students when dealing with international trade teaching cases, and the operation data includes the current error rate, the current risk preference, and the current error pattern.

[0182] It can be understood that the current error rate is the error frequency of students in solving the generated international trade teaching cases. A high error rate may mean that students have difficulties in certain knowledge points or skills. The current risk preference is the decision-making style of students in the case, reflecting the students' risk tolerance. For example, whether students tend to choose high-risk and high-return decisions in simulated transactions or prefer conservative strategies. The current error pattern is the common error types shown by students when solving cases. For example, whether students frequently apply certain trade rules incorrectly or lack the correct framework when analyzing international trade policies.

[0183] Exemplarily, the system can record the number of errors made by students at each stage (such as each decision point, task, or subtask), and calculate the error rate. The preferences can be judged by analyzing the decision-making behaviors of students in the simulated transactions of international trade teaching cases. For example, record the risk level of each transaction of a student in the simulated market, and then analyze the overall risk preference. By tracking the decision-making process of students, it can be analyzed in which specific tasks students often make mistakes and classify the types of mistakes.

[0184] Through machine learning algorithms and data analysis tools (such as cluster analysis, regression analysis, etc.), the generation process of teaching cases can be further optimized. By analyzing the operation data of students (current error rate, current risk preference, current error pattern), the learning trends and difficulties of students can be identified, and more adaptable teaching cases can be automatically generated. That is, the case difficulty can be adjusted according to the current error rate, the case strategy can be adjusted according to the current risk preference, and the case content can be adjusted according to the current error pattern.

[0185] Through this step, a highly personalized teaching experience can be achieved, helping students continuously improve their decision-making ability, learning ability, and psychological quality when solving practical problems. This data-driven optimization process not only improves teaching efficiency but also enhances students' sense of participation and learning effectiveness.

[0186] In a possible implementation manner, in step S400, optimizing the generation process of international trade teaching cases based on operation data includes: S410, constructing the current ability profile of students based on operation data.

[0187] Exemplarily, the current error rate can be standardized to a value between 0 and 1, where 0 represents no errors and 1 represents all decision-making errors. Determine the academic ability score according to the standardized current error rate. Students with a lower error rate correspond to a higher academic ability score, that is, use 1 minus the standardized current error rate to obtain the academic ability score.

[0188] The current risk preference can be quantified to a value between 0 and 1, where 0 represents extremely low risk and 1 represents extremely high risk. Determine the decision-making ability score according to the quantified current risk preference. Students who prefer low-risk decisions get higher scores, while students who tend to take risks get lower scores, that is, the quantified current risk preference is the decision-making ability score.

[0189] According to the current error pattern, the decision-making quality of students in different decision-making situations can be evaluated, and a stress response ability score can be generated.

[0190] The scores of each ability dimension can be synthesized to generate the current ability profile of the student. The current ability profile provides the performance of the student in different ability dimensions (academic ability, decision-making ability, adaptability), which can help understand the support and teaching strategies required by the student in international trade teaching cases, and provide data support for subsequent teaching case optimization, decision-making path planning, etc.

[0191] 420. Set the reward mechanism according to the current error rate in the operation data.

[0192] It can be understood that the goal of the reward mechanism is to motivate students to reduce errors and gradually improve their decision-making ability.

[0193] Exemplarily, the reward can be dynamically adjusted according to the current error rate. For example, the reward calculation formula can be set as: , where represents the reward, represents the current error rate. When the current error rate is 0, the reward is 1; when the error rate is 1, the reward is 0.

[0194] By setting the relationship between the reward and the error rate, the reward value can be adjusted in real time, thereby motivating students to improve their decision-making ability and automatically adjusting the difficulty and support level of the teaching case according to the performance of the students, which is conducive to students' learning and growth while adapting to their own ability levels.

[0195] S430. Based on the current ability profile and the reward mechanism, use the proximal policy optimization algorithm to optimize the case generation parameters to obtain the optimized case generation parameters.

[0196] Exemplarily, the optimization goal can be defined. The optimization goal can include reducing the error rate, increasing the reward, and improving the learning efficiency. The policy network is defined as a function. The policy network will adjust the generation parameters (such as decision-making path, case complexity, etc.) according to the ability profile and the reward mechanism of the student. The goal of the policy network is to enable students to obtain more rewards and reduce errors during the learning process.

[0197] The exploration mechanism in the proximal policy optimization (PPO) algorithm can be used to adjust the case generation parameters according to the reward mechanism and the ability profile of the student. By randomly or slightly adjusting the generation parameters, simulate the students' reactions and record their reward values. For different case generation parameters, the simulated error rate and reward changes can be recorded. After each decision, calculate the reward value of the student based on the reward mechanism, and adjust the parameters of the policy network based on the students' feedback (error rate and reward), thereby optimizing the case generation parameters. Through the clipped objective function in the PPO algorithm, optimize the case generation parameters to maximize the reward value in multiple explorations and gradually reduce the error rate of the students.

[0198] Optimize through multiple iterations. Each time, adjust the policy network according to the student's performance (error rate and reward), and optimize the case generation parameters. After each iteration, the student's ability profile and reward mechanism are updated, and the optimized case generation parameters will be used in the next round of learning. After several rounds of optimization and exploration, the optimized case generation parameters are finally obtained. The case generation parameters are automatically adjusted according to the student's current ability level and reward mechanism, enabling the student to obtain higher rewards and fewer errors during the learning process, facilitating the student to learn in a situation suitable for the current ability, and gradually improving the decision-making ability and error correction ability.

[0199] It should be understood that the magnitudes of the sequence numbers of the steps in the above embodiments do not imply the order of execution. The order of execution of each process should be determined by its function and internal logic, and should not constitute any limitation to the implementation process of the embodiments of the present application.

[0200] Corresponding to the dynamic generation method of international trade teaching cases described in the above embodiments, an embodiment of the present application also provides a device for dynamically generating international trade teaching cases. Each unit of the device can implement each step of the dynamic generation method of international trade teaching cases.

[0201] The device includes: An acquisition unit, configured to acquire the static data, behavior data, and trait data of the student. Among them, the static data includes the course grades and error rates of international trade courses, the behavior data includes the deviation amplitude of the transaction price from the baseline and the transaction success rate, and the trait data includes the stress resistance index and risk preference.

[0202] An ability profile construction unit, configured to construct the student's ability profile based on the static data, behavior data, and trait data.

[0203] An international trade teaching case generation unit, configured to generate international trade teaching cases according to the ability profile, real-time trade data, and trade knowledge graph. Among them, the real-time trade data includes international trade policies, government announcements, and media news, and the trade knowledge graph is constructed based on international trade rules.

[0204] An optimization unit, configured to acquire the operation data of the student and optimize the international trade teaching case generation process based on the operation data. Among them, the operation data is the operation data generated by the student when processing international trade teaching cases, and the operation data includes the current error rate, the current risk preference, and the current error pattern.

[0205] It should be noted that for the information interaction, execution process, etc. between the above units, since they are based on the same concept as the method embodiment of the present application, their specific functions and the technical effects brought are specifically described in the method embodiment part, and will not be elaborated here.

[0206] Those skilled in the art can clearly understand that, for the convenience and brevity of description, only the above division of each functional unit and module is used as an example. In actual applications, the above functions can be allocated to different functional units and modules as needed, that is, the internal structure of the device is divided into different functional units or modules to complete all or part of the functions described above. Each functional unit and module in the embodiments can be integrated into a processing unit, or each unit can exist physically alone, or two or more units can be integrated into one unit. The above integrated unit can be implemented in the form of hardware or in the form of a software functional unit. In addition, the specific names of each functional unit and module are only for the convenience of mutual distinction and do not limit the protection scope of this application. The specific working processes of the units and modules in the above device can refer to the corresponding processes in the foregoing method embodiments and will not be elaborated here.

[0207] The embodiment of this application also provides an international trade teaching case dynamic generation device. The international trade teaching case dynamic generation device in this embodiment includes: at least one processor, at least one memory, and a computer program stored in the at least one memory and executable on the at least one processor. When the processor executes the computer program, the international trade teaching case dynamic generation device implements the steps in any of the above international trade teaching case dynamic generation method embodiments, or enables the international trade teaching case dynamic generation device to implement the functions of each unit in the above device embodiments.

[0208] Exemplarily, the computer program can be divided into one or more modules / units. The one or more modules / units are stored in the memory and executed by the processor to complete this application. The one or more modules / units can be a series of computer program instruction segments capable of completing specific functions, and these instruction segments are used to describe the execution process of the computer program in the international trade teaching case dynamic generation device.

[0209] The international trade teaching case dynamic generation device can be a computing device such as a desktop computer, a notebook, a palm computer, and a cloud server. The international trade teaching case dynamic generation device can include, but is not limited to, a processor and a memory. It can include more or fewer components, or combine certain components, or different components. For example, it can also include input and output devices, network access devices, a bus, etc.

[0210] The processor may be a Central Processing Unit (CPU), or it may also be other general-purpose processors, Digital Signal Processors (DSPs), Application Specific Integrated Circuits (ASICs), Field-Programmable Gate Arrays (FPGAs), or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. The general-purpose processor may be a microprocessor or any conventional processor, etc.

[0211] In some embodiments, the memory may be an internal storage unit of the international trade teaching case dynamic generation device, such as the hard disk or memory of the international trade teaching case dynamic generation device. In other embodiments, the memory may also be an external storage device of the international trade teaching case dynamic generation device, such as a plug-in hard disk, Smart Media Card (SMC), Secure Digital (SD) card, Flash Card, etc. equipped on the international trade teaching case dynamic generation device. Further, the memory may also include both the internal storage unit and the external storage device of the international trade teaching case dynamic generation device. The memory is used to store an operating system, application programs, a BootLoader, data, and other programs, such as the program code of the computer program. The memory may also be used to temporarily store data that has been output or is to be output.

[0212] An embodiment of the present application also provides a computer-readable storage medium storing a computer program, and when the computer program is executed by a processor, the steps in any of the above method embodiments are implemented.

[0213] An embodiment of the present application provides a computer program product, and when the computer program product runs on the international trade teaching case dynamic generation device, the international trade teaching case dynamic generation device implements the steps in any of the above method embodiments.

[0214] When the integrated unit is implemented in the form of a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, to implement all or part of the processes in the above-described embodiment methods of this application, a computer program can be used to instruct relevant hardware to complete. The computer program can be stored in a computer-readable storage medium. When the computer program is executed by a processor, the steps of the above-described method embodiments can be implemented. Among them, the computer program includes computer program code, and the computer program code can be in the form of source code, object code, executable file or some intermediate form, etc. The computer-readable medium can at least include: any entity or device capable of carrying the computer program code to the international trade teaching case dynamic generation device, recording medium, computer memory, read-only memory (ROM, Read-Only Memory), random access memory (RAM, Random Access Memory), electrical carrier signal, telecommunication signal, and software distribution medium. For example, a USB flash drive, a mobile hard disk, a magnetic disk, or an optical disc, etc. In some jurisdictions, according to legislation and patent practice, the computer-readable medium cannot be an electrical carrier signal and a telecommunication signal.

[0215] In the above embodiments, the descriptions of the various embodiments have their own emphases. For the parts not detailed or recorded in a certain embodiment, reference can be made to the relevant descriptions of other embodiments.

[0216] Those of ordinary skill in the art can realize that the units and algorithm steps of the examples described in combination with the embodiments disclosed in this article can be implemented by electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are executed in a hardware or software manner depends on the specific application and design constraints of the technical solution. Professional technicians can use different methods for each specific application to implement the described functions, but such implementation should not be considered to exceed the scope of this application.

[0217] In the embodiments provided in this application, it should be understood that the disclosed international trade teaching case dynamic generation device, equipment and method can be implemented in other ways. For example, the international trade teaching case dynamic generation device and equipment embodiments described above are only illustrative. For example, the division of the modules or units is only a logical function division. In actual implementation, there can be other division methods. For example, multiple units or components can be combined or integrated into another system, or some features can be ignored or not executed. Another point is that the displayed or discussed coupling, direct coupling or communication connection to each other can be through some interfaces. The indirect coupling or communication connection of the device or unit can be in an electrical, mechanical or other form.

[0218] The unit described as the separation component may or may not be physically separated. The component displayed as a unit may or may not be a physical unit, that is, it may be located in one place or distributed across multiple network units. Some or all of the units can be selected according to actual needs to achieve the purpose of the solution of this embodiment.

[0219] The above-described embodiments are only used to illustrate the technical solutions of the present application, rather than to limit them. Although the present application has been described in detail with reference to the foregoing embodiments, those of ordinary skill in the art should understand that they can still modify the technical solutions described in the foregoing embodiments, or perform equivalent replacements on some of the technical features; and these modifications or replacements do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of the present application, and should all be included within the protection scope of the present application.

Claims

1. A method for dynamically generating international trade teaching cases, characterized in that: include: Obtaining the static data, behavioral data and trait data of the students; wherein the static data includes the course grades and error rates of international trade courses, the behavioral data includes the deviation of the transaction price from the baseline and the transaction rate, and the trait data includes the stress resistance index and risk preference; Constructing the ability profile of the student based on the static data, the behavioral data and the trait data; Generate international trade teaching cases based on the capability profile, real-time trade data and trade knowledge graph; wherein the real-time trade data includes international trade policies, government announcements and media news, and the trade knowledge graph is constructed based on international trade rules; The operation data of the students is obtained, and the process of generating international trade teaching cases is optimized based on the operation data; wherein the operation data is the operation data generated by the students when processing the international trade teaching cases, and the operation data includes the current error rate, the current risk preference and the current error mode.

2. The method for dynamically generating international trade teaching cases according to claim 1, characterized in that: The constructing the student's ability profile based on the static data, the behavior data and the trait data includes: Eliminate abnormal data from the static data, the behavioral data, and the characteristic data to obtain cleaned static data, behavioral data, and the characteristic data; Extracting negotiation capability features from the cleansed behavioral data, and extracting compliance capability features from the cleansed static data and the cleansed behavioral data; Extracting the associated knowledge points between the negotiation capability feature, the compliance capability feature and the trait data and the knowledge points of the international trade course and calculating the degree of association of each associated knowledge point to construct a knowledge association graph; Based on the knowledge association graph, the negotiation ability characteristics, the compliance ability characteristics, and the stress resistance index, a clustering algorithm is used to divide the groups to obtain the student portrait category; Construct the student's ability portrait based on the knowledge association map and the portrait category.

3. The method for dynamically generating international trade teaching cases according to claim 2, characterized in that: The method of dividing the groups by using a clustering algorithm based on the knowledge association graph, the negotiation ability characteristics, the compliance ability characteristics and the stress resistance index to obtain the student portrait categories includes: Based on the negotiation ability characteristics, the compliance ability characteristics and the stress resistance index, the clustering algorithm is used to divide the groups and obtain preliminary portrait categories; Based on the association degree of each associated knowledge point in the knowledge association graph, the preliminary portrait category is optimized to obtain the portrait category of the student.

4. The method for dynamically generating international trade teaching cases according to claim 2, characterized in that: The generating of international trade teaching cases based on the capability portrait, real-time trade data and trade knowledge graph includes: Determining initial generation parameters according to the capability portrait, the real-time trade data, and the trade knowledge graph; Adjusting the difficulty of the initial generation parameters based on a three-dimensional difficulty matrix to obtain case generation parameters; wherein the three-dimensional difficulty matrix is ​​constructed based on the capability portrait; Generate parameters according to the case, construct a Prompt, and based on the Prompt, call the LoRA fine-tuned large language model to generate the international trade teaching case.

5. The method for dynamically generating international trade teaching cases according to claim 4, characterized in that: The method further comprises: Constructing a case structure template; wherein the case structure template includes background, conflict, decision point, and result; According to the case structure template, the verified international trade teaching cases are annotated with case structures to obtain a training data set; wherein the verified international trade teaching cases are international trade teaching cases that comply with the international trade rules; Initialize the parameters of LoRA and select the fine-tuning layer in the large language model; Based on the training data set, the large language model is trained, and according to the decrease of the loss function in the large language model, the parameters of the LoRA are adjusted to optimize the fine-tuning layer in the large language model to obtain the LoRA fine-tuned large language model.

6. The method for dynamically generating international trade teaching cases according to claim 4, characterized in that: The determining of initial generation parameters according to the capability portrait, the real-time trade data and the trade knowledge graph includes: Determine the case topic and market context based on the real-time trade data; Determine key knowledge points according to the capability portrait and the knowledge association map; According to the knowledge association graph, identifying the student's knowledge weaknesses, and determining a decision branch based on the knowledge weaknesses and the trade knowledge graph; Determine the difficulty of the case based on the capability profile; The case theme, the market background, the key knowledge points, the decision branches and the case difficulty are integrated to obtain the initial generation parameters.

7. The method for dynamically generating international trade teaching cases as claimed in claim 4, wherein the three-dimensional difficulty matrix includes complexity, uncertainty and cultural conflict, and is characterized in that: The difficulty adjustment of the initial generation parameters based on the three-dimensional difficulty matrix to obtain case generation parameters includes: Adjusting the complexity of the initial generation parameters based on the compliance capability characteristics to obtain first generation parameters; Adjusting the uncertainty of the initial generation parameter based on the compression resistance index to obtain a second generation parameter; adjusting the cultural conflict of the initial generation parameters based on the negotiation capability characteristics to obtain third generation parameters; The first generation parameter, the second generation parameter and the third generation parameter are combined to obtain the case generation parameters; wherein the case generation parameters include case theme, market background, key knowledge points, decision branches, complexity, uncertainty and cultural conflict.

8. The method for dynamically generating international trade teaching cases as claimed in claim 7, wherein Prompt comprises a role definition layer, a basic setting layer, a teaching requirement layer and an output specification layer, characterized in that: The generating parameters according to the case and constructing a prompt include: Determining the variables of the role definition layer according to the case theme and market background of the case generation parameters; wherein the variables of the role definition layer include country parameters and trade environment; Determine the variables of the basic setting layer according to the market background, key knowledge points and decision branches of the case generation parameters; wherein the variables of the basic setting layer include trade terms, commodity categories and participating roles; Determine the variables of the teaching requirement layer according to the complexity of the case generation parameters, key knowledge points, decision branches, market background, cultural conflicts, and uncertainties; wherein the variables of the teaching requirement layer include case structure complexity, teaching objectives, decision paths, current market dynamics, negotiation scenarios, and random events; A decision branch of parameters is generated according to the case, and the decision consequence of the output specification layer is quantified.

9. The method for dynamically generating international trade teaching cases according to claim 4, characterized in that: The optimizing the process of generating the international trade teaching case based on the operation data includes: Building a current ability profile of the student based on the operation data; Setting a reward mechanism based on the current error rate in the operational data; Based on the current capability portrait and the reward mechanism, the case generation parameters are optimized using a proximal strategy optimization algorithm to obtain optimized case generation parameters.

10. A device for dynamically generating international trade teaching cases, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that: When the processor executes the computer program, the method according to any one of claims 1 to 9 is implemented.

Images