Learning management system, information processing method and program for a learning management system

Abstract

We can freely build systems that operate with proprietary software that generates practical, work-level deliverables by reverse engineering using unique data processing structures and inference algorithms. [Solution] The learning management system 10 includes an inference unit 1A-8 that has a knowledge database for storing structured data, analyzes input information obtained from learners based on the structured data, infers missing variables in light of dependencies, calculates the priority of missing parameters to be obtained, and dynamically branches and generates question items to be presented to learners based on that priority, and an output generation unit 1A-9 that combines the learner's individual information obtained by the inference unit 1A-8 with the structured data, works backward to determine the components of the output based on dependencies, and automatically generates a practical-level output tailored to the learner.

Description

Technical Field

[0001] The present invention relates to a learning management system, an information processing method of the learning management system, and a program.

Background Art

[0002] In recent years, e-learning is known in which learners connect to a school server via the Internet to learn. In such e-learning, many learning management systems for managing learners' learning have been proposed (see Patent Document 1).

[0003] In Patent Document 2 below, "In order to provide an individual optimal education support system that can analyze a target person based on objective elements and recommend a learning method according to the characteristics of the target person, the individual optimal education support system of the present disclosure includes an action information acquisition means for acquiring action information regarding the actions of the target person, an analysis means for analyzing the action characteristics of the target person based on the action information acquired by the action information acquisition means, and a learning method recommendation means for recommending a learning method according to the action characteristics of the target person analyzed by the analysis means." is disclosed.

Prior Art Documents

Patent Documents

[0004]

Patent Document 1

Patent Document 2

Summary of the Invention

Problems to be Solved by the Invention

[0005] In conventional skill learning, a "learning-first" flow in which trainees create a product through trial and error after learning basic knowledge is common. However, since beginners do not understand "the correct form at the practical level" or "what elements are required", the acquisition efficiency is extremely low, and it takes a great deal of time to reach a quality that can be used in practice.

[0006] For example, even if you learn English and understand and memorize vocabulary and grammar, you may not reach a level where you can converse fluently with native speakers. You may learn programming and be able to develop according to the textbook, but you may be completely lost in a real development environment. Even if you learn design skills and acquire specific knowledge of color schemes, placement, layout, and design essences for different target audiences and purposes, you may not be able to create designs that are well-received in the market when working on actual projects. In the traditional learning-first flow, the problem of "the extremely high hurdle to reaching the (so-called) applied level required in the field after acquiring basic skills" has not been solved, and even with sincere learning, it has been difficult for all students to achieve the results they expected before learning.

[0007] Furthermore, in conventional learning systems, simply introducing a general-purpose generative AI and adding functions such as "asking the AI ​​questions" or "having the AI ​​create examples" is merely a design modification that can be appropriately made by those skilled in the field. However, in specific business domains (such as web strategies), it is fundamentally impossible for a novice to obtain practical-level results by simply giving instructions to a general-purpose AI. The reasons for this are as follows.

[0008] For practical-level output, expert-level information—specifically, tacit knowledge—is essential, such as what assumptions and variables should be input. Beginners often lack the knowledge to even know what to input into the AI, making it fundamentally impossible to solve this problem by simply connecting a general-purpose AI to an existing system. Furthermore, while general-purpose AI learns from a wide range of open-source resources on the web, obtaining practical-level results requires highly confidential expert knowledge and know-how (the so-called "secret recipe"), but current generative AI cannot select the know-how specific to each business domain. Therefore, in principle, it has been difficult for beginners to use generative AI to generate practical-level output in a specific business domain.

[0009] Cognitive burden of prompt construction: When beginners try to create deliverables that are suitable for practical use using general-purpose AI, they are required to engage in complex prompt engineering through trial and error. As a result, they end up in a counterproductive situation where their learning effort is diverted to "AI operation methods" rather than "acquiring actual business skills."

[0010] Furthermore, in today's world, the correct answers required in the field are constantly changing due to shifts in trends and algorithms. As a result, general-purpose AI based on past data cannot fully reflect the latest market demands, and there is always a risk that "the generated output will be outdated methods that are no longer applicable in practice." To address this, it is common practice for those in the industry to provide content that explains the latest trends and market demands. However, due to the aforementioned limitations of learning-first approaches and the "unknown unknowns barrier," it remains difficult for students to generate output that accurately captures market demands.

[0011] Furthermore, since the very purpose of education is for students to become able to do things on their own, when introducing AI into existing educational systems, it is often thought that "having the AI ​​create the complete output (answer) in advance deprives students of the opportunity for trial and error and reduces the effectiveness of the education." For this reason, there have been strong obstacles to companies in the field adopting a configuration in which "the system uses AI to generate a practical, end-to-end output from the beginning and provide it to the students."

[0012] Furthermore, in the traditional "learning-first" flow, even if students create deliverables through trial and error, beginners face the challenge of being extremely unable to accurately self-evaluate whether their work has reached a level that is acceptable in the market or in practical work (the correct answer).

[0013] For example, even if you learn web design and writing and create a product according to the course materials, beginners who lack their own evaluation criteria based on practical experience cannot judge whether it is of a quality that can produce results (such as conversions) in a real business setting. As a result, they may continue learning with incorrect interpretations or low-quality work, and as a result, they may not reach a practical level.

[0014] In other words, even if one attempts to obtain feedback using existing systems or general-purpose generative AI, general-purpose AI can only perform superficial corrections based on general grammar and theory, but it cannot provide evaluations based on a "professional perspective" (market trends and highly confidential know-how) specific to a particular business domain. Furthermore, it is often physically and financially impossible for expert instructors to provide individual feedback on a regular basis.

[0015] Therefore, in conventional learning methods where students proceed without having their own absolute standard of correct answers beforehand, it has been fundamentally difficult to overcome the limitations of self-assessment.

[0016] Furthermore, in conventional human resource development systems and learning management systems, the curriculum presented to students typically consists of tasks aimed at inputting abstract knowledge, such as "watch the video for Chapter 1." However, such abstract tasks present a significant challenge: students cannot concretely visualize how today's learning directly relates to the final product they want to create, which easily leads to decreased motivation and frustration (failure of self-management).

[0017] Those skilled in the art have attempted to solve this problem (maintaining motivation and managing behavior) by adding management functions to existing learning management systems, such as "automatic reminder function when not logged in" and "progress visualization graphs," and simultaneously introducing methods to encourage partial output, such as "submission of end-of-chapter assignments" and "creation of simple deliverables (mini-work)." However, these are merely design changes that those skilled in the art can make as appropriate, and no matter how these conventional methods are combined, it is fundamentally impossible to bring students to a practical level and prevent a fundamental decline in motivation.

[0018] This is because the "tasks and mini-work" in existing systems are merely independent and partial tasks aimed at "confirming the retention of the basic knowledge learned immediately beforehand." The practical deliverables required in actual business settings are not simply a collection of single pieces of knowledge, but rather a complex interplay of market trends and multiple preconditions (the tacit knowledge of experts). Therefore, even if beginners complete conventional partial tasks and receive automatic reminders from the system or have their progress managed with progress graphs, they cannot dispel the fundamental doubt of "will doing this really result in skills that can be used in the field?" and as a result, they do not reach the practical quality that is acceptable in the market.

[0019] The real reason for the decline in motivation and poor skill retention lies in the fact that abstract learning and partial output are being accumulated without a defined or presented "ultimate practical-level goal (the correct answer required in the market)." Therefore, in the conventional system structure, where the "final practical-level deliverables for each student" are not defined in advance by the system and daily tasks are not designed by working backward from that, it is impossible to translate daily tasks into "concrete actions directly related to practical work," making it difficult to fundamentally solve the limitations of this behavioral management and skill acquisition.

[0020] This invention has been made in view of the circumstances described above, and its purpose is to provide a learning management system, an information processing method for a learning management system, and a program that can operate with proprietary software that generates practical-level deliverables in reverse using the following unique data processing structure and inference algorithm, rather than using general-purpose AI as a mere "passive text generation tool," in order to overcome all of the aforementioned fundamental limitations. [Means for solving the problem]

[0021] To achieve the aforementioned objectives, the learning management system according to the present invention is a learning management system in which a server device that operates the learning management system and a data terminal operated by the learner communicate with each other to provide learning support services to the learner, wherein the server device provides knowledge information including teaching know-how, problem-solving methods, optimal deliverable configurations tailored to the target group, and market trend analysis results possessed by experts active in a specific business domain, to multiple final deliverables at the expert's practical level. Includes target audience, objectives, budget, and assumptions. Variable names defined by variable names, dependency information defining the type of dependency that identifies the dependency between the multiple variables, and constraint conditions identified by rule type, Subtracting the variable names entered by the learner from the variable names including the multiple target groups, objectives, budget, and assumptions mentioned above. Missing variables given name The system includes a knowledge database that stores structured data structured to associate with inference priorities that define the priority of each variable, and further analyzes input information, including the specifications, objectives, conditions, and learner attributes of the desired deliverables, which the learner inputs via the data terminal, with the final deliverables defined in the multiple variable name tables, infers missing parameters for variables not included in the input information based on the dependencies, and further determines the acquisition priority of the missing parameters by referring to the inference priority table. identification The system includes an inference unit that dynamically generates question items to be presented to the learner based on their priority, and further includes a unit that combines the learner's individual information obtained through the presentation of the question items by the inference unit with the structured data, and determines the values ​​of the remaining variables based on the dependencies so as to satisfy the constraints defined in the structured data, and the determined rest The value of the variable In order to produce deliverables of a practical standard, including Exclusive for learners sentence It is characterized by comprising a product generation unit that automatically generates book data.

Effect of the Invention

[0022] According to the learning management system according to the present invention, a learning management system that can operate with unique software that inversely generates work products at the practical level can be freely constructed by a unique data processing structure and inference algorithm.

[0023] As described above, the present invention has been briefly explained. Furthermore, the details of the present invention will be further clarified by reading through the embodiments (hereinafter referred to as "embodiments") for carrying out the invention described below with reference to the attached drawings.

Brief Description of the Drawings

[0024] The drawings show specific embodiments of the present invention and include not only essential components of the invention but also optional and preferred embodiments. [Figure 1] FIG. 1(A) and FIG. 1(B) are overall conceptual diagrams of a learning management system according to an embodiment of the present invention. [Figure 2] It is a block diagram for explaining the functional configuration of the server device shown in FIG. 1. [Figure 3] It is an overall block diagram of the learning management system. [Figure 4] It is a block diagram of the target learning form information storage unit. [Figure 5] It is a block diagram of the target learner curriculum selection unit. [Figure 6] It is a flowchart showing the operation until a target learner curriculum is presented to a target learner in the learning management system. [Figure 7] It is a flowchart showing the operation of managing the daily schedule and ToDo list. [Figure 8] It is a block diagram of the general learning template storage unit. [Figure 9]This flowchart illustrates the process of presenting an approximate learning template and workflow to the target learner. [Modes for carrying out the invention]

[0025] The following embodiments relate to the learning management system shown in Figures 1 to 9, and include not only essential components of the invention but also optional and preferred components.

[0026] [Embodiment] Figure 1(A) shows a conceptual diagram of the learning management system 10 according to the present invention. The learning management system is exemplified by a learning management system in which a server device 1A that operates the learning management system communicates with a data terminal operated by a learner. Note that 10A and 10B are processors that execute a functional processing program described later and perform information processing according to the flowchart described later.

[0027] As shown in Figure 1(A), the learning management system 10 is connected to the electronic devices (e.g., PCs, smartphones, tablets, TVs, etc.) 201 of target learners 200 via a network 100 such as the Internet. The target learners 200 can learn in real time from the learning management system 10 using their electronic devices 201.

[0028] Furthermore, as shown in Figure 1(B), the present invention also includes a configuration in which the learning results of the target learner 200, who studied on an electronic terminal 201 at home 202, are stored on an electronic medium and moved to the school 203, and connected to the learning management system 10 via an electronic terminal 204 within the school 203. In addition, while the target learner in the embodiment is, for example, a working adult (company employee, civil servant, self-employed person, etc.), the target learner in the present invention also includes current students and exam candidates in addition to working adults.

[0029] Figure 2 is a block diagram illustrating the functional configuration of the server device 1A shown in Figure 1. Note that while the configuration described later for 1A-2 to 1A-6 involves the processor 10A of the server device 1A executing programs stored in memory, a configuration in which multiple processors process each function in parallel may also be adopted.

[0030] The server device 1A includes a database 1A-1 that stores knowledge information learned through an AI learning process from experts in a specific field, and a first generation unit 1A-2 that individually generates structured deliverables to be presented to the learner by referring to the knowledge information stored in the database 1A-1.

[0031] In this invention, the knowledge information refers to information learned through an AI learning process from educational experts such as web marketers and designers who are active in specific business domains, including their teaching know-how, problem-solving methods, error analysis, learning plan development criteria, model answers, and analysis results of entrance examination questions. Below is an example of a knowledge information table stored in database 1A-1. Furthermore, this knowledge information is just an example, and it is assumed that a person skilled in the art (as defined in Article 36, Paragraph 4, Item 1 of the Patent Act) would foresee including other knowledge.

[0032] Examples include typical error patterns in each subject and their causes, target audience, objectives, budget, and prerequisites.

[0033] The structured data in this invention holds the expertise of specialists in a specific business domain as a data structure that defines multiple variables constituting the deliverables, as well as the dependencies and constraints between these variables in a network-like manner, and consists of a template management table, a variable definition table, a variable dependency table, a constraint rule table, an inference priority table, a network map table, and the like.

[0034] This makes it possible to retain the tacit knowledge of experts as a graph structure of variables and dependencies.

[0035] Specifically, Table 1 shows the Template_Master (structured template management table).

[0036] [Table 1]

[0037] Specifically, Table 2 shows the Template_Variables (variable definition table) which defines the "variables" that make up the deliverable.

[0038] [Table 2]

[0039] Specifically, Table 3 shows Variable_Dependencies (variable dependency table), which stores the dependencies (constraints) between variables as a network.

[0040] [Table 3]

[0041] Specifically, Table 4 shows the Constraint_Rules (constraint rule table), which holds individual constraints for each variable.

[0042] [Table 4]

[0043] Specifically, Table 5 shows the Inference_Priority (inference priority table), which manages priority information for calculating the "next missing parameter to acquire".

[0044] [Table 5]

[0045] Specifically, Table 6 shows a Variable_Network_Map, which stores the network structure between variables as a graph.

[0046] [Table 6]

[0047] Furthermore, the server device 1A includes an acquisition unit 1A-3 that acquires target learning format information for target learners who are scheduled to learn specific content; a selection unit 1A-4 that selects approximate past learning format information that approximates the target learning format information from the past learning format information of each past learner from among the success examples of multiple past learners; a generation unit 1A-5 that generates a target learner curriculum based on the approximate past learning curriculum used by past learners based on the approximate past learning format information; and a first presentation unit 1A-6 that presents the target learner curriculum generated by the generation unit 1A-5 to the target learner.

[0048] Here, the learning unit 1A-3 performs a process to acquire at least one of the following as target learning format information: the days of the week the target learner can study, the times of day they can study, the duration of study, the desired learning level they wish to acquire through learning, the content they have previously acquired through learning, the learning level they have previously acquired through learning, and their current learning environment.

[0049] Furthermore, the server device 1A includes a second presentation unit 1A-7 that generates a daily schedule as the target learner curriculum and presents the target learner with a ToDo list at the time of access.

[0050] Here, the second presentation unit 1A-7 is characterized by creating a next ToDo list that progresses learning from the ToDo list at a predetermined time.

[0051] Furthermore, the system includes an inference unit 1A-8 that has a knowledge database for storing structured data, analyzes input information obtained from learners based on the structured data, infers missing variables in the input information in light of the dependencies between the variables, calculates the priority of missing parameters to be obtained, and dynamically branches and generates question items to be presented to learners based on that priority; and an output generation unit 1A-9 that combines the learner's individual information obtained by the inference unit 1A-8 with the structured data, works backward to determine the components of the output based on the dependencies, and automatically generates a practical-level output tailored to the learner.

[0052] The learning management system shown in this embodiment stores data generation models for generating learning models in database 1A-1. The learning management system functions as so-called generative AI (Artificial Intelligence). An example of a data generation model is ChatGPT (Internet Search<URL:https: / / openai.com / blog / chatgpt> Examples of generative AI include those described above. The data generation model is obtained by performing deep learning on a neural network. The data generation model is input with a prompt containing instructions, and at least one inference data such as audio data representing speech, text data representing text, and image data representing an image.

[0053] The data generation model is configured to infer expert knowledge from input inference data according to prompts and to output the inference results in data formats such as audio data and text data. Here, inference refers to, for example, analysis, classification, prediction, concretization, and / or summarization.

[0054] Figure 3 shows a schematic block diagram of the learning management system 10. The learning management system 10 consists of an input unit 11, a target learning format information storage unit 12, a target learner curriculum selection unit 13, a daily schedule generation unit 14, a to-do list generation unit 15, a reminder generation unit 16, a timing unit 17, a display unit 18, an approximate learning template generation unit 20, a workflow generation unit 30, and an inference model generation unit 40.

[0055] [Information on target learning methods] The input section 11 is where the target learner enters various personal information. Immediately after enrolling in the school, the target learner enters information about their target learning style through an input screen (input section) such as a browser on an information terminal and uploads it to the learning management system 10. The target learning style information is stored in the target learning style information storage section 12.

[0056] As shown in Figure 3, after the target learner enters their name, date of birth, address, gender, highest level of education, subject, work history, etc., they also enter information about the target learning format, such as the days of the week on which they can study (daily, specific days of the week, etc.) 12A, the times when they can start studying 12B, the total time they can study (continuous or total time they can study from the start time) 12C, the desired level of proficiency they wish to acquire through learning (selected from multiple levels such as high, medium, and low) 12D, the content they have previously acquired through learning 12E, the level of proficiency they have previously acquired through learning (selected from multiple levels such as high, medium, and low) 12F, their current learning environment (availability of a PC, etc., availability of a private study room, etc.) 12G, and any special notes (for example, whether they have national qualifications including professional qualifications, whether they have private qualifications, knowledge gained from past work, proficiency in foreign languages, etc.) 12H.

[0057] [Target Learner Curriculum] As shown in Figure 4, in the learning management system 10, the target learner curriculum selection unit 13 matches the target learner's target learning style information (12A to 12H, see Figure 3) with a large amount of past learning style information 13A. Past learning style information 13A is, for example, information of past learners who have studied at the school in the past, and each is stored with a past learning curriculum 13B linked to it.

[0058] From these past learning style information 13A, one that closely matches the target learning style information (12A~12H) entered by the target learner is selected as the approximate past learning style information 13C. This process is performed by AI (Artificial Intelligence).

[0059] The selection criteria for selecting past learning style information 13A as approximate past learning style information 13C, or in other words, the criteria for determining whether a particular past learning style information 13A is similar to the target learning style information (12A~12H), are determined by assigning priority to each item entered by the target learner as the target learning style information (12A~12H), and by the degree of agreement or similarity of each item in the past learning style information 13A. The priority order for each item in the target learning style information (12A~12H) can be changed automatically by AI or manually as appropriate, for example, depending on differences in learning courses.

[0060] Furthermore, the numerous past learning patterns 13A include not only successful cases where past learners reached their desired proficiency level, but also unsuccessful cases where past learners did not reach their desired proficiency level, or even cases where past learners abandoned their studies midway. Even such unsuccessful cases are often partially useful as historical information.

[0061] The approximate past learning curriculum 13D, linked to the approximate past learning format information 13C selected in this manner, is presented to the target learner as the target learner curriculum 13E via the display unit 18. The display unit 18 displays various information, images, etc., for example, on the browser of the electronic device used by the target learner.

[0062] As shown in Figure 2, When a learner agrees to the presented target learner curriculum 13E, the daily schedule generation unit 14 generates a daily schedule, the ToDo list generation unit 15 generates a ToDo list, and the timing unit 17 starts timing based on the daily schedule. Figure 5 is a block diagram illustrating the functions of the target learner curriculum selection unit 13 shown in Figure 3. In Figure 5, 13A is past learning format information. 13B is past learning curriculum. 13C is approximate past learning format information. 13D is approximate past learning curriculum.

[0063] [Presentation of target learner curriculum] Figure 6 shows, Learning Management System A flowchart is provided illustrating the information processing method from the moment the target learner inputs information about their learning style until the target learner is presented with their curriculum.

[0064] As shown in Figure 6, in step 10 (ST10), when the target learner inputs target learning format information, in step 11 (ST11), the target learning format information is compared with a large number of past learning format information entries. In step 12 (ST12), one of the past learning format information entries that is similar to the target learning format information input by the target learner is selected as the approximate past learning format information. In step 13 (ST13), the past learning curriculum linked to the selected past learning format information is presented to the target learner as the target learner curriculum.

[0065] Then, if the target learner agrees to the target learner curriculum presented in step 14 (ST14) (YES), the process proceeds to step 15 (ST15). In step 15 (ST15), the learning management system 10 generates a daily schedule using the daily schedule generation unit 14 (see Figure 2), and generates a ToDo list using the ToDo list generation unit 15 (see Figure 2). In step 16 (ST16), the timing unit 17 (see Figure 2) starts timing based on the daily schedule.

[0066] Furthermore, if the target learner does not agree to the target learner curriculum presented in Step 14 (ST14) (NO), proceed to Step 17 (ST17).

[0067] In Step 17 (ST17), the target learner curriculum presented in Step 14 (ST14) is excluded from the selection, and the process returns to Step 12 (ST12). This allows for the objective and rational creation of individual target learner curricula based on the target learner's learning style information (12A-12H).

[0068] [Reminder function] Figure 7 shows a flowchart illustrating the information processing method corresponding to the reminder function based on the daily schedule. As shown in Figure 7, if the target learner logs into the learning management system 10 in step 20 (ST20) (YES), the process proceeds to step 21 (ST21).

[0069] If the daily schedule is not yet completed in Step 21 (ST21) (YES), proceed to Step 22 (ST22). If the student submits the assignments for the student curriculum in Step 22, proceed to Step 23 (ST23). If the student has achieved 100% completion of the assignments for the student curriculum submitted in Step 23 (ST23), proceed to Step 24 (ST24).

[0070] In step 24 (ST24), the learning management system 10's daily schedule generation unit 14 (see Figure 2) generates the next daily schedule, and the ToDo list generation unit 15 (see Figure 2) generates the next ToDo list. In step 25 (ST25), the timing unit 17 starts timing based on the next daily schedule.

[0071] On the other hand, if the target learner is not logged in at step 20 (ST20) (NO), proceed to step 26 (ST26). If the daily schedule is due before the deadline at step 26 (ST26) (YES), return to step 20 (ST20). If the daily schedule is due after the deadline at step 26 (ST26) (NO), proceed to step 27 (ST27), send a reminder to the target learner prompting them to log in, and return to step 20 (ST20).

[0072] Login reminders are sent via methods such as email, SMS (Short Message Service), SNS (Social Networking Service), and voice guidance via telephone, and are repeated at pre-set times or at specified intervals.

[0073] If, in step 21 (ST21), the daily schedule is overdue (NO), meaning there is a schedule delay, in step 28 (ST28), the learner is notified of the schedule delay via the display unit 18 (see Figure 2), and the process proceeds to step 22 (ST22). If, in step 22 (ST22), the learner does not submit the curriculum assignment (NO), in step 29 (ST29), a reminder is displayed to the learner via the display unit 18 (see Figure 2) to encourage them to submit the curriculum assignment, and the process returns to step 22 (ST22).

[0074] If the completion rate of the curriculum assignments submitted by the target learner in Step 23 (ST23) is less than 100% (NO), in Step 30 (ST30), the daily schedule generation unit 14 (see Figure 3) generates a revised daily schedule taking into account the incomplete parts of the curriculum assignments, and the ToDo list generation unit 15 (see Figure 2) generates a revised ToDo list. The revised daily schedule and revised ToDo list for the next session are then presented to the target learner via the display unit 18 (see Figure 2).

[0075] Subsequently, in step 25 (ST25), the timing unit 17 starts timing based on the next revised daily schedule.

[0076] [Approximate Learning Template] Returning to Figure 3, if the target learner requests advice from the learning management system 10 for the target learner curriculum 13E or higher, as shown in Figure 4, the target learner inputs the desired outcome (the level of achievement desired by the target learner) 12X separately from the target learning format information (12A to 12H). The learning management system 10 extracts the specific elements necessary for the desired outcome from the desired outcome 12X via the specific element extraction unit 21.

[0077] The extraction of specific elements in the specific element extraction unit 21 is performed by the AI ​​comprehensively determining, for example, the wording and sentence structure of the input desired outcome 12X and the target learner's target learning format information (12A to 12H).

[0078] As shown in Figure 8, the learning management system 10 stores, for example, learning templates used by past learners who have studied at the school in the past, as a large number of general-purpose learning templates 22A in the general-purpose learning template storage unit 22.

[0079] General-purpose learning template 22A links past learners' desired outcomes with their desired outcomes and past workflows leading up to those outcomes.

[0080] The general-purpose learning template storage unit 22 includes not only successful cases where past learners achieved their desired outcomes, but also unsuccessful cases where past learners did not achieve their desired outcomes, or unsuccessful cases where past learners abandoned their learning midway. Even such unsuccessful cases are often partially useful as historical information.

[0081] The learning management system 10 selects one of the many general-purpose learning templates 22A that the AI ​​determines to be optimal based on the specific elements extracted by the specific element extraction unit 21 (see Figures 3 and 4), generates an approximate learning template in the approximate learning template generation unit 20, and presents it to the target learner via the display unit 18.

[0082] [Workflow] Returning to Figure 3, the workflow generation unit 30 is activated in conjunction with the approximation learning template generation unit 20, which generates an approximation learning template. The workflow generation unit 30 refers to the approximation learning template in the approximation learning template generation unit 20, as well as to past workflows stored in the workflow database (DB) 31, and selects the workflow that the AI ​​determines to be optimal.

[0083] If the AI ​​determines that there is insufficient information to select a workflow, the learning management system 10 proactively engages with the target learner via the specific element extraction unit 21 to acquire further desired outcomes.

[0084] Specifically, if the desired outcome is abstract, this could involve prompting the learner to provide specific input, presenting the learner with detailed questions about the desired outcome, or presenting the learner with multiple options for selection.

[0085] [Inference Model] Furthermore, the learning management system 10 starts the inference model generation unit 40 in the backend in response to the startup of the workflow generation unit 30. The inference model generation unit 40 refers to the approximate learning template in the approximate learning template generation unit 20 and the workflow in the workflow generation unit 30, as well as the numerous past inference models stored in the inference model database (DB) 41, and selects the inference model that the AI ​​has determined to be optimal.

[0086] The inference model database (DB) 41 is structured by refining (improving quality) and tuning (enhancing functionality) past inference models based on the knowledge of the school's instructors and experts both inside and outside the school. It stores a large amount of data that has high market value and is highly confidential.

[0087] The inference model generation unit 40 complements the workflow generated by the workflow generation unit 30 using the inference model that the AI ​​has determined to be optimal. The workflow generated in this way is generated by working backward from the desired outcome of the target learner, and is designed to enable the target learner to efficiently learn, acquire, and master methods for obtaining high-quality deliverables with high market value in the shortest possible time.

[0088] Figure 9 shows a flowchart illustrating the information processing method until the approximate learning template and workflow are presented to the target learner. As shown in Figure 6, if the target learner inputs the desired outcome in step 40 (ST40) and the specific elements necessary for the desired outcome are extracted in step 41 (ST41) (YES), the process proceeds to step 42 (ST42).

[0089] In step 42 (ST42), the AI ​​selects one of several general-purpose learning templates that it deems optimal to generate an approximate learning template, and in step 43 (ST43), a workflow is generated based on the approximate learning template. Then, in step 44 (ST44), the workflow is supplemented by an inference model, and in step 45 (ST45), the approximate learning template and workflow are presented to the target learner.

[0090] If the target learner agrees to the presented approximate learning template workflow in step 46 (ST46) (YES), the timing unit 17 starts timing based on the daily schedule in step 47 (ST47).

[0091] If the specific elements necessary for the desired outcome could not be extracted in Step 41 (ST41) (NO), proceed to Step 48 (ST48). In Step 48 (ST48), actively engage with the target learners to further acquire the desired outcome. If the specific elements necessary for the desired outcome could be extracted in Step 49 (ST49) (YES), proceed to Step 42 (ST42).

[0092] If the specific elements necessary for the desired outcome could not be extracted in Step 49 (ST49) (NO), return to Step 48 (ST48). If the target learner did not agree to the presented approximate learning template / workflow in Step 46 (ST46) (NO), proceed to Step 50 (ST50).

[0093] In step 50 (ST50), the approximate learning template workflow presented in step 45 (ST45) is excluded from the selection, and then the process proceeds to step 48 (ST48).

[0094] In conventional learning management systems, a "learning-first" workflow is common, where after completing an approximate learning template, or while progressing through learning using the approximate learning template, the learner creates the desired outcome (e.g., text, outline, policy plan, etc.) through trial and error.

[0095] In contrast, the learning management system 10 extracts specific elements necessary for the desired outcome entered by the target learner, and presents an "output-first" learning workflow that starts with the structure and generation process of "high-quality deliverables demanded in the market" based on those specific elements, and then works backward from there, enabling learners to efficiently learn, acquire, and master practical skills in the shortest possible time.

[0096] In particular, the learning management system 10 uses a proprietary AI that has accumulated specific business know-how (such as prompt configuration and domain knowledge), making it easy for even beginner learners to produce results at a practical level.

[0097] The disclosure relating to the present invention described above can be summarized to at least the following: Information on target learning methods 12A to 12H is collected from target learners 200 who are scheduled to learn specific content; approximate past learning method information 13C is selected from the past learning method information 13A of each past learner from among the success examples of multiple past learners, and the approximate past learning curriculum 13D used by past learners based on the approximate past learning method information 13C is presented to the target learner 200 as the target learner curriculum 13E.

[0098] The present invention disclosed in paragraph 0031 above may include at least the following embodiments. These embodiments may be adopted separately or in combination with each other.

[0099] Furthermore, the server device operating the learning management system may be configured to include a knowledge database (see Tables 1 to 6) that stores the expertise of experts in a specific business domain as structured data in which multiple variables constituting the deliverables, and the dependencies and constraints between the multiple variables are defined in a network, and before step 40 (ST40) shown in Figure 9, the processor may perform the following steps: analyze the input information obtained from the learner based on the structured data stored in the knowledge database, and infer the missing variables in the input information in light of the dependencies between the variables; calculate the priority of the missing parameters to be obtained, and dynamically branch and generate question items to be presented to the learner based on that priority; and further combine the learner's individual information obtained by the inference unit with the structured data, and inversely determine the components of the deliverables based on the dependencies, and automatically generate a practical-level deliverable specifically for the learner.

[0100] [Effects of the Embodiment] According to this embodiment, a learning management system can be freely constructed that operates with proprietary software that generates practical-level deliverables in reverse using a unique data processing structure and inference algorithm.

[0101] [Remarkable effects based on the configuration described in Claim 1 of the present invention] Furthermore, there are clear obstacles to a person skilled in the art applying general-purpose generative AI to a conventional learning management system and adopting a configuration that realizes a deliverable generation function in which "the system first generates and presents a practical-level final deliverable (answer) in an end-to-end manner," as in the present invention.

[0102] This is because conventional learning management systems are based on a "learning-first" approach, where students acquire knowledge through trial and error, and aim to cultivate their ability to solve problems independently. Therefore, if this system were to generate and present a "perfect final product" tailored to the individual circumstances of each learner at an early stage, it would deprive students of the opportunity for trial and error, directly contradicting the original purpose of conventional learning systems (promoting self-directed learning).

[0103] Furthermore, by deliberately overcoming the aforementioned strong inhibiting factors and adopting an "output-first" configuration, we achieve a unique and remarkable effect that those skilled in the art could not have predicted.

[0104] Specifically, the deliverable initially generated and presented in this invention is not a general example solution. Instead, it is a "practical deliverable that perfectly meets market demand," generated by storing in the system the "expert thought process (secret recipe) for performing specific tasks" that does not exist in open source, not as mere text but as structured data (workflow) defining dependencies between variables, and comparing it with expert evaluation criteria data.

[0105] In the process of generating such deliverables, this learning management system does not indiscriminately request information from the target learner. Instead, it employs an information processing mechanism in which an inference model analyzes the target learner's input in real time, calculates only the information (parameters) that are truly lacking in the workflow for creating expert-level output, dynamically branches the questions, and selectively extracts only the necessary information.

[0106] This allows students not only to receive perfect deliverables generated by the system, but also to clearly understand and relive the "expert thought process" itself within the system—how professional instructors follow this workflow, input this level of information at each step, and consequently produce this level of output.

[0107] Furthermore, it dramatically accelerates the understanding of "professional thought processes for meeting market demands," which was fundamentally impossible with conventional systems. This results in a significant improvement in the speed and quality of acquiring practical skills compared to traditional trial-and-error learning, producing a uniquely superior educational effect that cannot be achieved with a learning-first flow and can only be obtained through an "output-first" flow.

[0108] Furthermore, the "daily to-do function" and "curriculum generation function" of the system according to the present invention are not merely a collection of existing technologies, but employ an extremely robust, organically coupled configuration that always dynamically links together, with output based on the deliverable generation function as an essential input.

[0109] In particular, the "output-first flow" is a powerful learning method, but it alone does not guarantee that all students will achieve their goals.

[0110] Therefore, the curriculum generation function described herein is expected to have a synergistic effect by taking detailed interviews about the target level, current level, available time, etc., of the target learners and transforming them into an "output-first curriculum optimized for each student," while still being based on the flow of the deliverable generation function.

[0111] Furthermore, even with an individually optimized curriculum, the likelihood of achieving the goals is low if it is not implemented perfectly.

[0112] Therefore, this system employs a configuration in which the daily ToDo function presents the curriculum as subdivided daily ToDos, provides reminders for uncompleted tasks, and executes a "dynamic calculation logic that redefines the next day's ToDos" according to the progress of the day (uncompleted, 100%, over, etc.).

[0113] Specifically, the system recalculates the number of days from the relevant date to the graduation date, the remaining necessary To-Dos and the time required to achieve the goal, and reallocates To-Dos based on the student's available time. Notably, in cases where the allocated To-Dos exceed the student's available time due to progress delays and are deemed "physically difficult to complete," the daily To-Do function's judgment triggers the curriculum generation function to redraw the entire long-term curriculum from that day onward, performing a restructuring process that simplifies the process to achieve the goal. This combination of functions is expected to produce synergistic effects.

[0114] Thus, the present invention does not have independent functions, but rather has a deliverable generation function as its main component. By combining this main function with a curriculum generation function and a daily to-do function, the curriculum generation function generates a curriculum from the learner's absolute goal definition, and the daily to-do function redraws the curriculum in progress based on progress and limit assessments, providing real-time feedback. This configuration is closely and highly integrated, and the powerful synergistic effect that cannot be achieved by design changes of the art makes it possible to bring the student's goal achievement rate as close to 100% as possible.

[0115] The present invention can also be realized by supplying a program that implements one or more of the functions of the above-described embodiments to a system or device via a network or storage medium, and by a process in which one or more processors in the computer of that system or device read and execute the program. It can also be realized by a circuit (e.g., ASIC) that implements one or more functions.

[0116] (1) A learning management system in which a server device operating the learning management system and a data terminal operated by the learner communicate to provide learning support services to the learner, wherein the server device provides knowledge information including teaching know-how, problem-solving methods, optimal configuration of deliverables tailored to the target group, and market trend analysis results possessed by experts active in a specific business domain, to multiple final deliverables at the expert's practical level. Includes target audience, objectives, budget, and assumptions.Variable names defined by variable names, dependency information defining the type of dependency that identifies the dependency between the multiple variables, and constraint conditions identified by rule type, Subtracting the variable names entered by the learner from the variable names including the multiple target groups, objectives, budget, and assumptions mentioned above. Missing variables given name The system includes a knowledge database that stores structured data structured to associate with inference priorities that define the priority of each variable, and further analyzes input information, including the specifications, objectives, conditions, and learner attributes of the desired deliverables, which the learner inputs via the data terminal, with the final deliverables defined in the multiple variable name tables, infers missing parameters for variables not included in the input information based on the dependencies, and further determines the acquisition priority of the missing parameters by referring to the inference priority table. identification The system includes an inference unit that dynamically generates question items to be presented to the learner based on their priority, and further includes a unit that combines the learner's individual information obtained through the presentation of the question items by the inference unit with the structured data, and determines the values ​​of the remaining variables based on the dependencies so as to satisfy the constraints defined in the structured data, and the determined rest The value of the variable In order to produce deliverables of a practical standard, including For learners only sentence It is characterized by having a deliverable generation unit that automatically generates document data.

[0117] (2) The server device includes an acquisition unit that acquires target learning method information for target learners who are scheduled to learn specific content, a selection unit that selects approximate past learning method information that approximates the target learning method information from the past learning method information of each past learner from among the success examples of multiple past learners, and the approximate past learning method information Extract past learning curricula that are linked to and stored in memory as the target learner's curriculum. The system is characterized by comprising a generation unit and a first presentation unit that presents the target learner curriculum generated by the generation unit to the target learner.

[0118] (3) The learning unit is characterized in that it acquires at least one of the following as target learning format information: the days of the week when the target learner can study, the times when the target learner can study, the duration of study, the desired level of learning that the learner wishes to acquire through learning, the content of learning that the learner has acquired in the past, the level of learning that the learner has acquired in the past, and the current learning environment.

[0119] (4) The system is characterized by having a second presentation unit that generates a daily schedule as the target learner curriculum and presents the target learner with a ToDo list at the time of access.

[0120] (5) The second presentation unit is characterized by creating a next ToDo list from the ToDo list at a predetermined time to advance the learning.

[0121] (6) A learning management system information processing method for providing learning support services to learners by communicating between a server device operating the learning management system and a data terminal operated by a learner, wherein the server device includes knowledge information including teaching know-how, problem-solving methods, optimal deliverable configurations tailored to target groups, and market trend analysis results possessed by experts active in a specific business domain, variable names defined by multiple variable names constituting the expert's final deliverables at the practical level, dependency information defining the types of dependencies that identify dependencies between the multiple variables, and constraints identified by rule types, Subtracting the variable names entered by the learner from the variable names including the multiple target groups, objectives, budget, and assumptions mentioned above. Missing variables given name The system includes a knowledge database that stores structured data structured to associate with inference priorities that define the priority of each variable, and further analyzes input information, including the specifications, objectives, conditions, and learner attributes of the desired deliverables, which the learner inputs via the data terminal, with the final deliverables defined in the multiple variable name tables, infers missing parameters for variables not included in the input information based on the dependencies, and further determines the acquisition priority of the missing parameters by referring to the inference priority table. identification The system includes an inference step that dynamically generates question items to be presented to the learner based on their priority, and further includes a method that combines the learner's individual information obtained through the presentation of the question items by the inference step with the structured data, and determines the values ​​of the remaining variables based on the dependencies so as to satisfy the constraints defined in the structured data. rest The value of the variable In order to produce deliverables of a practical standard, including For learners only sentenceIt is characterized by having a deliverable generation step that automatically generates document data.

[0122] (7 ) studies The data terminal operated by the trainee and A computer that functions as a communication server. This includes insights from experts active in a specific business domain, such as guidance know-how, problem-solving methods, optimal deliverable structure tailored to the target audience, and market trend analysis results, defined by variable names that constitute the expert's final deliverable at the practical level, dependency information that defines the type of dependency that identifies the dependencies between the multiple variables, and constraints identified by rule types. Subtracting the variable names entered by the learner from the variable names including the multiple target groups, objectives, budget, and assumptions mentioned above. Missing variables given name A server computer of a learning management system, which has a knowledge database that stores structured data structured to associate with inference priorities that define the priority of each variable, analyzes input information, including the specifications, objectives, conditions, and learner attributes of the desired deliverables, which learners input via the data terminal, by comparing it with the final deliverables defined in the multiple variable name tables, infers missing parameters for variables not included in the input information based on the dependencies, and further determines the acquisition priority of the missing parameters by referring to the inference priority table. identification The system includes an inference unit that dynamically generates question items to be presented to the learner based on their priority, and further combines the learner's individual information obtained through the presentation of the question items by the inference unit with the structured data, so as to satisfy the constraints defined in the structured data. rest The values ​​of the variables are determined based on the aforementioned dependencies, and the values ​​of the remaining variables are determined. In order to produce deliverables of a practical standard, including For learners only sentence It features a program that functions as a deliverable generation unit that automatically generates written data. [Explanation of Symbols]

[0123] 1A Server device 1B Server device 10 Learning Management Systems 12A Available days for learning 12B Learning time 12C Learning Time 12D Desired proficiency level 12E Past learning contents 12F Past Acquisition Level 12G Current learning environment 12X desired outcome 13A Past Learning Format Information 13C Approximate Past Learning Pattern Information 13D Approximate Past Learning Curriculum 13E Target Learner Curriculum 22A General-purpose learning template 200 target learners

Claims

1. A learning management system in which a server device operating the learning management system and a data terminal operated by the learner communicate to provide learning support services to the learner, The server device includes a knowledge database that stores structured data that associates knowledge information, including the teaching know-how, problem-solving methods, optimal deliverable configurations tailored to target groups, and market trend analysis results possessed by experts active in a specific business domain, with variable names defined by variable names including multiple target groups, objectives, budgets, and assumptions that constitute the expert's practical final deliverables, dependency information that defines the types of dependencies that identify dependencies between the multiple variables, constraints identified by rule types, and inference priorities that define the priority of missing variable names obtained by subtracting the variable names entered by the learner from the variable names including the multiple target groups, objectives, budgets, and assumptions. Furthermore, the system includes an inference unit that compares and analyzes input information, including the specifications, objectives, conditions, and learner attributes of the desired deliverables, entered by the learner via the data terminal, with the final deliverables defined in the multiple variable name tables, infers missing parameters based on the dependencies of the input information, identifies the acquisition priority of the missing parameters by referring to the inference priority table, and dynamically generates question items to be presented to the learner based on that priority. Furthermore, the learning management system is characterized by comprising a deliverable generation unit that combines the learner's individual information obtained through the presentation of the question items by the inference unit with the structured data, determines the values ​​of the remaining variables based on the dependencies so as to satisfy the constraints defined in the structured data, and automatically generates learner-specific document data to produce a work-level deliverable, including the determined values ​​of the remaining variables.

2. The server device is An acquisition unit that acquires information on the target learning format for target learners who are scheduled to learn specific content, A selection unit that selects approximate past learning style information that approximates the target learning style information from the past learning style information of each past learner from among the success stories of multiple past learners, A generation unit that extracts past learning curricula stored in association with the aforementioned approximate past learning format information as target learner curricula, The system comprises a first presentation unit that presents the target learner curriculum generated by the generation unit to the target learner, The learning management system according to claim 1, characterized in that the generation unit generates a target learner curriculum for learners in accordance with learning plan formulation standards.

3. The learning management system according to claim 2, characterized in that the acquisition unit acquires at least one of the following as target learning format information: the days of the week when the target learner can study, the times when the target learner can study, the duration of study, the desired level of acquisition that the learner wishes to acquire through learning, the content of past acquisitions acquired through learning in the past, the level of acquisition acquired through past learning in the past, and the current learning environment.

4. The learning management system according to claim 2, further comprising a second presentation unit that generates a daily schedule as the target learner curriculum and presents a To-Do list to the target learner at the time of access.

5. The learning management system according to claim 4, characterized in that the second display unit creates a next To-Do list to advance learning from the To-Do list at a predetermined time.

6. A learning management system information processing method for providing learning support services to learners through communication between a server device operating the learning management system and a data terminal operated by learners, The server device includes a knowledge database that stores structured data that associates knowledge information, including the teaching know-how, problem-solving methods, optimal deliverable configurations tailored to target groups, and market trend analysis results possessed by experts active in a specific business domain, with variable names defined by multiple variable names that constitute the expert's practical final deliverables, dependency information that defines the types of dependencies that identify dependencies between the multiple variables, constraints identified by rule types, and inference priorities that define the priority of missing variable names obtained by subtracting the variable names entered by the learner from the variable names including the multiple target groups, objectives, budgets, and assumptions. Furthermore, the system includes an inference step that compares and analyzes input information, including the specifications, objectives, conditions, and learner attributes of the desired deliverables, entered by the learner via the data terminal, with the final deliverables defined in the multiple variable name tables, infers missing parameters based on the dependencies of the input information, identifies the acquisition priority of the missing parameters by referring to the inference priority table, and dynamically generates question items to be presented to the learner based on that priority. Furthermore, the learning management system's information processing method is characterized by comprising a deliverable generation step which combines the learner's individual information obtained through the presentation of the question items by the inference step with the structured data, determines the values ​​of the remaining variables based on the dependencies so as to satisfy the constraints defined in the structured data, and automatically generates learner-specific document data to produce a work-level deliverable, including the determined values ​​of the remaining variables.

7. A server computer for a learning management system that stores structured data, which is structured such that knowledge information including teaching know-how, problem-solving methods, optimal deliverable configurations tailored to target groups, and market trend analysis results possessed by experts active in a specific business domain is associated with variable names defined by multiple variable names constituting the expert's practical-level final deliverables, dependency information defining the types of dependencies that identify dependencies between the multiple variables, constraints identified by rule types, and inference priorities that define the priority of missing variable names obtained by subtracting the variable names entered by the learner from the variable names including the multiple target groups, objectives, budgets, and assumptions. An inference unit analyzes input information, including the specifications, objectives, conditions, and learner attributes of the desired deliverables, which the learner inputs via the data terminal, by comparing it with the final deliverables defined in the multiple variable name tables, infers missing parameters based on the dependencies of the input information, further identifies the acquisition priority of the missing parameters by referring to the inference priority table, and dynamically generates question items to be presented to the learner based on that priority. Furthermore, the program is characterized by combining the learner's individual information obtained through the presentation of the question items by the inference unit with the structured data, determining the values ​​of the remaining variables based on the dependencies so as to satisfy the constraints defined in the structured data, and functioning as a deliverable generation unit that automatically generates learner-specific document data to produce a workable deliverable, including the determined values ​​of the remaining variables.

Images